AI News Digest 2026-10-16
特集
開発者コーナー
AIツール紹介コーナー
速報コーナー
参考記事一覧
参考記事一覧を表示
- Meet Grok Voice 3.0: SpaceXAI Voice Agent Model Ranks #1 on Artificial Analysis (importance: 55, dev_relevance: 60)
- DeepSeek suffers another major outage as company offers user refunds - South China Morning Post (importance: 30, dev_relevance: 20)
- DeepSeek's Second Funding Round Doubles to 100 Billion Yuan; Pre-Money Valuation Stays at 500 Billion Yuan - BigGo Finance (importance: 30, dev_relevance: 5)
- Anthropic tells investors quarterly revenue up nearly 50% to $16.8bn ahead of IPO (importance: 60, dev_relevance: 5)
- Elon Musk is promising to “make you whole” if Grok messes up a purchase or leaks your credit card. - The Verge (importance: 25, dev_relevance: 5)
- Elon Musk hints at launching Amazon rival after e-commerce giant blocks his Grok bots - India Today (importance: 15, dev_relevance: 0)
- The X Subscription Bundles Grok, Cursor and X Premium From $8 - Tech My Money (importance: 15, dev_relevance: 10)
- OpenAI again shakes up data center leadership, with xAI exec Uday Ruddarraju to lead and Sachin Katti to leave - Data Center Dynamics (importance: 25, dev_relevance: 5)
- Ofcom questions X’s handling of illegal content as Grok faces new restrictions - Firstpost (importance: 25, dev_relevance: 0)
- Figma announces new features to define how agents work on its platform (importance: 45, dev_relevance: 70)
- Google launches its ‘universal’ online shopping cart ahead of the holidays (importance: 20, dev_relevance: 0)
- YouTube Music launches new labels as AI-generated music floods streaming platforms (importance: 15, dev_relevance: 0)
- Muse Spark Open is Meta's answer to Qwen, GLM, and Kimi in the open-weight race (importance: 65, dev_relevance: 65)
- Black Forest Labs' open-weights image model FLUX 3 Image Klein runs on consumer GPUs with just 8 GB of VRAM (importance: 35, dev_relevance: 25)
- Google wants Gemini 4 Argon to be the only AI model robots need (importance: 30, dev_relevance: 35)
- NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning (importance: 40, dev_relevance: 60)
- Bun 1.4.3、TypeScript 7.0.2と同じエラーを返す型検査を同梱 (importance: 0, dev_relevance: 0)
- GitLab、CI設定の正規表現経由で未認証RCEとなる欠陥を修正 (importance: 72, dev_relevance: 80)
- LangGraphとQlib Alpha158、Google TimesFMで作る「自律型AIヘッジファンド」をOSS公開した話 (importance: 0, dev_relevance: 60) (いいね相当スコア: 0)
- 機械学習を実務で回す技術書4選 (importance: 0, dev_relevance: 30) (いいね相当スコア: 1)
- AI Price Prediction: We Asked Grok AI Which Recovers First From the October Selloff: Bitcoin, Ethereum, XRP or Solana - 24/7 Wall St. (importance: 2, dev_relevance: 0)
- An update on independent evaluators (importance: 35, dev_relevance: 15)
- Preparing for the 2026 US midterms (importance: 25, dev_relevance: 0)
- Anthropic opens Singapore office and signs MOU with the Government of Singapore (importance: 40, dev_relevance: 10)
- v2.1.300 (importance: 45, dev_relevance: 90)
- v2.1.299 (importance: 40, dev_relevance: 90)
- Teaching students to think and learn through questions (importance: 15, dev_relevance: 5)
- Amp puts AI agents to work beyond code with OpenAI models (importance: 40, dev_relevance: 60)
- SentinelOne cuts malware analysis from weeks to hours (importance: 25, dev_relevance: 15)
- How Third Moment Research runs 10,000 agents for 36+ hours (importance: 35, dev_relevance: 60)
- How Block is building AI-powered security with OpenAI technology (importance: 25, dev_relevance: 30)
- The EU’s next chapter in AI is agentic (importance: 20, dev_relevance: 10)
- Putting Earth AI to work: Geospatial agents for our planet's biggest challenges (importance: 25, dev_relevance: 25)
- Analyzing technical documents with Deep Review (importance: 25, dev_relevance: 25)
- Research@ 2026: 10 ways we're amplifying human ingenuity (importance: 15, dev_relevance: 5)
- Accelerating the search for antimicrobial peptides with Google Antigravity (importance: 35, dev_relevance: 20)
- Google DeepMind and Singapore Government collaborate to build fine-tuned AI model for scam detection (importance: 35, dev_relevance: 15)
- Proof-based general-purpose computing with Magnetar (importance: 40, dev_relevance: 55)
- Apertus 1.5 Release Notes (importance: 45, dev_relevance: 40)
- HiRO-ACEv2: Simulating extreme weather at 3 km resolution for a fraction of the cost (importance: 25, dev_relevance: 10)
- GitHub availability report: September 2026 (importance: 20, dev_relevance: 45)
- From 2,500 requests to 24: Rebuilding star history without the stargazers (importance: 40, dev_relevance: 85)
- A fully programmable secure web gateway (importance: 25, dev_relevance: 50)
- Deploy with Void (importance: 45, dev_relevance: 85)
- Cloudflare and OpenAI: Making AI discovery better for everyone (importance: 40, dev_relevance: 40)
- Yandex outage: Data shows that instead of going dark, malicious bots went on the move (importance: 20, dev_relevance: 20)
- Inside Hyperdrive: How we solved a rare Hyperdrive MySQL caching bug (importance: 40, dev_relevance: 85)
- How Meta Built Forum With AI-Native Engineering (importance: 35, dev_relevance: 75)
- 2026 World Cup: Breaking Records, Spiky Charts and Reliability at Scale (importance: 20, dev_relevance: 45)
- Sapienz Receives ISSTA 2026 Impact Paper Award (importance: 15, dev_relevance: 40)
- Google Summer of Code 2026 With the Kotlin Foundation: Recap (importance: 10, dev_relevance: 35)
- IntelliJ IDEA’s Java and Kotlin for Cursor and VS Code Hits 1.0 (importance: 45, dev_relevance: 90)
- How Netflix Made AI Coding Agents More Efficient with JetBrains IDE Intelligence (importance: 60, dev_relevance: 90)
- MPS 2026.2 Release Candidate Available (importance: 20, dev_relevance: 80)
- What Makes a Great KotlinConf Proposal? (importance: 5, dev_relevance: 10)
- Boost Your Pipeline: Top 6 CI/CD Integrations for Qodana (importance: 15, dev_relevance: 60)
- How Cloud-Based Agent Sandboxing Works in Air (importance: 0, dev_relevance: 0)
- KotlinConf Recap: Lessons from Verifone’s Architect on Kotlin in Mission-Critical Payments (importance: 10, dev_relevance: 40)
- Lauterbach and JetBrains Announce TRACE32 Integration in CLion IDE (importance: 25, dev_relevance: 80)
- AI Agent Security Risks: What Can Go Wrong When Agents Act on Their Own (importance: 15, dev_relevance: 40)
- Harbor + Docker Sandboxes: Run Safe, Reproducible Agent Evals (importance: 35, dev_relevance: 80)
- Scale Agentic Coding Workflows with Meta’s Muse Spark Open on NVIDIA GB300 NVL72 (importance: 40, dev_relevance: 60)
- Optimizing Agent Reliability and Tool Calling Efficiency for Chip Design Using NVIDIA NeMo Gym (importance: 30, dev_relevance: 60)
- How to Rewrite Apache Spark UDFs for GPU Acceleration with AI Agents (importance: 25, dev_relevance: 65)
- NVIDIA Run:ai Is Now Open Source as Part of NVIDIA DSX AI Factory Platform (importance: 0, dev_relevance: 0)
- ReplaySSM: Rethinking State Storage for Faster Hybrid-Model Decoding (importance: 30, dev_relevance: 50)
- 5 Steps to Create SimReady Assets for Robotics with Frontier AI Models (importance: 20, dev_relevance: 40)
- Building Reliable Data Analytics Agents: Lessons from the KDD Cup (importance: 30, dev_relevance: 60)
- The Machines that Make the Machines (importance: 20, dev_relevance: 40)
- Validate AI Factory Changes with Digital Twins and AI Agents (importance: 20, dev_relevance: 30)
- Scaling Decision Optimization to 100 Million Variables and Beyond with mPDLP in NVIDIA cuOpt (importance: 25, dev_relevance: 60)
- Faster Scientific Image Analysis with NVIDIA cuPhoton (importance: 20, dev_relevance: 40)
- Scale Bitwise-Deterministic Pretraining with NVIDIA Megatron Core (importance: 30, dev_relevance: 60)
- How DOCA GPUNetIO Unifies GPU-Initiated Networking Across the NVIDIA Software Stack (importance: 20, dev_relevance: 60)
- AICR v1.0: Open, stable, and verifiable GPU cluster configuration (importance: 25, dev_relevance: 60)
- Control How Your GPU Shares Work with Green Contexts (importance: 20, dev_relevance: 60)
- Build Applications on NVIDIA BlueField Faster with NVIDIA DOCA Agent Skills (importance: 25, dev_relevance: 65)
- Build Local AI Apps with C++ and NVIDIA TensorRT RTX Samples (importance: 25, dev_relevance: 65)
- Fine-Tuning NVIDIA Nemotron for Saudi Arabic Dialects, with a Path to Other Languages (importance: 20, dev_relevance: 50)
- Deploying an HSTU Generative Recommender with NVIDIA Dynamo-Triton (importance: 25, dev_relevance: 60)
- Expanding AI Storage Access with NVIDIA cuObject and the NVIDIA SCADA Server SDK (importance: 20, dev_relevance: 55)
- Tracing Agent Harness Behavior with NVIDIA NeMo Relay (importance: 35, dev_relevance: 80)
- AI Native by Design: Lessons Learned from Building NVIDIA TensorRT Model Connect (importance: 30, dev_relevance: 70)
- Lower the Cost of Building and Running Visual AI Agents with NVIDIA VSS Blueprint 3.3 (importance: 25, dev_relevance: 50)
- NVIDIA Open Agent Safety Platform: A Reference for Continuous In-Silicon Agent Monitoring (importance: 30, dev_relevance: 60)
- Add Runtime Controls to AI Agents with NVIDIA OpenShell (importance: 35, dev_relevance: 70)
- How NVIDIA DSX MaxLPS Maximizes AI Factory Throughput and Efficiency (importance: 15, dev_relevance: 25)
- Efficient MoE Training for Biological Foundation Models (importance: 20, dev_relevance: 35)
- Introducing NV-Reason-CT Open 3D CT VLM for Radiologist Chain-of-Thought Reasoning (importance: 25, dev_relevance: 15)
- Validate GPU Cluster Readiness Before AI Workloads Land (importance: 20, dev_relevance: 50)
- Manage Kubernetes Node Fleets with NodeWright (importance: 20, dev_relevance: 60)
- Report: FTC probes Anthropic, OpenAI after Trump dismissed AI fears as “hoax” (importance: 60, dev_relevance: 10)
- Epic says its new vibe-coded Epic Games Launcher is faster than Steam (importance: 25, dev_relevance: 35)
- Researchers use AI to design proteins that cut up other proteins (importance: 35, dev_relevance: 5)
- Recording device hidden in toilet paper found in top-secret OpenAI meeting room (importance: 30, dev_relevance: 5)
- Anthropic warns small AI models could self-replicate through IoT devices (importance: 40, dev_relevance: 10)
- OpenAI, Google pushing new AI models easily tricked into making nudes, researchers say (importance: 30, dev_relevance: 5)
- Bluesky expands its AI-powered Attie app to let anyone build websites, games, and more (importance: 25, dev_relevance: 20)
- Last chance to save up to $200 on your TechCrunch Founder Summit pass: Early booking ends October 16 (importance: 1, dev_relevance: 0)
- AI agents are coming to Rivian’s cars (importance: 20, dev_relevance: 5)
- TechCrunch Disrupt 2026: Last day (importance: 3, dev_relevance: 10)
- Amazon’s Bee launches two new personal AI devices, starting at $75 (importance: 35, dev_relevance: 5)
- Ricursive Intelligence is betting on AI-designed chips to help solve the compute crisis (importance: 35, dev_relevance: 20)
- Gudea leverages AI to help organizations prevent social media crises before they erupt (importance: 15, dev_relevance: 5)
- Flock CEO says its microphones detect ‘just gunshots’ — its website says different (importance: 25, dev_relevance: 5)
- Apate sends AI agents undercover to get scammed so they can gather fraud intelligence (importance: 20, dev_relevance: 15)
- Deezer will soon let you filter out AI-generated music (importance: 20, dev_relevance: 3)
- Save 50% on your TechCrunch Disrupt 2026 ticket and join the last 2 days in San Francisco’s Moscone West (importance: 2, dev_relevance: 5)
- Zenzap raises $44M for its AI chat app that aims to replace WhatsApp in the workplace (importance: 25, dev_relevance: 5)
- Floqer’s AI agents research sales prospects so teams don’t have to (importance: 20, dev_relevance: 10)
- TechCrunch Disrupt 2026: Day 2 (importance: 2, dev_relevance: 5)
- Scimplify raises $25M to challenge China’s grip on critical materials with AI (importance: 25, dev_relevance: 5)
- Meta and Apple team up to make Muse AI models run better on Macs (importance: 75, dev_relevance: 45)
- Friends don’t let friends feed their data to AI agents (importance: 15, dev_relevance: 5)
- How Windows is being optimized for Nvidia’s RTX Spark laptops (importance: 30, dev_relevance: 20)
- Nobody is safe from celebrity AI deepfakes (importance: 15, dev_relevance: 3)
- Benn Jordan on Flock, privacy, and the new surveillance backlash (importance: 10, dev_relevance: 2)
- Amazon’s always-listening Bee wearable can now help you buy stuff (importance: 30, dev_relevance: 5)
- Your secrets aren’t safe with this cute robot toy — they’re on Chinese servers ready to be hacked (importance: 15, dev_relevance: 5)
- McKinsey is making job applicants spar with a fake AI client (importance: 20, dev_relevance: 3)
- Eight states sue OpenAI, saying ChatGPT isn’t safe for kids (importance: 55, dev_relevance: 5)
- Presentation: What 25 Trillion Tokens Reveal About How Developers Actually Adopt AI Agents (importance: 35, dev_relevance: 75)
- Meta Extends Confidential Computing to AI Glasses with Private Processing (importance: 25, dev_relevance: 30)
- Google Antigravity SDK Adds Local Model Support for Greater Token Efficiency and Code Privacy (importance: 60, dev_relevance: 85)
- Bank of America puts 4,000 people on fixing vulnerabilities uncovered by Anthropic's Mythos and other AI models (importance: 40, dev_relevance: 35)
- GPT-6-Astra solves 49 of 50 physics research problems at $2.40 apiece (importance: 45, dev_relevance: 15)
- Google says self-improving AI isn't anywhere close (importance: 40, dev_relevance: 40)
- Researchers tracked down a rogue fleet of Chinese AI agents in the wild and actually talked to it (importance: 45, dev_relevance: 20)
- OpenAI will reportedly ship at least one model this year but its flagship still needs alignment work (importance: 45, dev_relevance: 25)
- AnitaB.org、今年のGrace Hopper Celebrationを資金不足で中止 (importance: 20, dev_relevance: 15)
- 米ソフトウェア開発の求人は増加、伸びはシニアとAI職に偏る (importance: 25, dev_relevance: 25)
- Intel、推論向けGPU Dシリーズのメモリと性能をOCPで提示 (importance: 40, dev_relevance: 30)
- Cloudflare、エージェントのブラウザセッションをターン間で維持 (importance: 55, dev_relevance: 90)
- Linux vhostの欠陥Vghost、ローカルrootとVM脱出の詳細が公開 (importance: 50, dev_relevance: 45)
- Shopify、Hydrogenをフレームワーク非依存のライブラリとして一般提供 (importance: 45, dev_relevance: 90)
- Mozilla、自社インフラで管理するFirefox Enterpriseを公開 (importance: 25, dev_relevance: 20)
- Next.js、未認証RCE2件とServer ActionsのDoSを修正 (importance: 68, dev_relevance: 90)
- Bun、次期バージョン向けに組み込みフォーマッタ bun fmt を公開 vscode実測でPrettier比約320倍高速と主張 (importance: 55, dev_relevance: 95)
- Nous Research、Hermes Agent v0.22.0を公開 無料トライアルでの初回セットアップとローカルllama.cppを追加 (importance: 55, dev_relevance: 80)
- XBOW、GPT-5.6 SolでChromeフルチェーンを6日構築 Googleの25万ドル報奨を獲得、修正済み (importance: 60, dev_relevance: 50)
- Anthropic、Claude Managed AgentsとMessages APIのOpenTelemetryトレース出力をパブリックベータ公開 (importance: 65, dev_relevance: 90)
- Framework Laptop 12、Wildcat LakeのLinux実測は前世代の1.26倍 (importance: 20, dev_relevance: 25)
- Linuxカーネル、LLM利用のAssisted-by必須を外すパッチが投稿 (importance: 0, dev_relevance: 70)
- npmのTrusted Publishing、1,500超がリポジトリ名の再利用で露出 (importance: 0, dev_relevance: 85)
- Argo CD、削除フックがAppProjectの制限を迂回する欠陥を修正 (importance: 0, dev_relevance: 85)
- AWS、通常のHTTPサーバーを載せるLambda Web Functionsを公開 (importance: 0, dev_relevance: 75)
- 2Bのコード探索モデルColGrep-Agent、Claude Codeのトークンを34%削減 (importance: 0, dev_relevance: 75)
- CopilotKit Releases CLiF: A Continual Learning Platform That Turns Agent Traces and User Actions into Skills, Evals, and Fine-Tuning Data (importance: 45, dev_relevance: 75)
- openKylin Outlines Its Vision for Reimagining Linux in the AI Era at Open Source Summit Europe 2026 (importance: 20, dev_relevance: 15)
- TypeSafe AI Releases Jev 1.14 With Images, Box Type and 262k Context (importance: 30, dev_relevance: 50)
- Datalab Releases OmniParseBench: An Open OCR Benchmark With 16,288 Pass/Fail Tests to Audit OCR Parsers (importance: 30, dev_relevance: 40)
- Fish Audio Releases Drama 3, a 35B Open-Weight MoE Text-to-Speech Model (importance: 35, dev_relevance: 20)
- 9 Best AI Gateways in 2026: Bifrost, Portkey, LiteLLM, Kong and 5 More Ranked (importance: 20, dev_relevance: 60)
- Liquid AI Releases LFM2.5-24B-A2B: A 2B-Active MoE for On-Device Agents (importance: 40, dev_relevance: 55)
- Artificiality's First Exam: An LLM-Referenced C-test for Supraintelligence (importance: 30, dev_relevance: 15)
- The Silent Failures in Agentic Data Science (importance: 40, dev_relevance: 60)
- ClawLoop: Self-Improving Agents via Harness-Native Training Loop (importance: 45, dev_relevance: 65)
- Correction Training for Diffusion Models (importance: 35, dev_relevance: 10)
- Long-Horizon-Terminal-Bench 2.0: Automating Diverse Task Construction and Measuring Sustained Agent Progress (importance: 40, dev_relevance: 65)
- VETNET: Assembling Benchmark Datasets for AI-Driven Veterinary Innovation (importance: 10, dev_relevance: 5)
- Unconstrained CTI-REALM with Open-Weights (importance: 35, dev_relevance: 40)
- Visualizing the World Model in an Agent's Value Function (importance: 25, dev_relevance: 10)
- Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models (importance: 25, dev_relevance: 10)
- Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs (importance: 35, dev_relevance: 55)
- Harness^2: Recursive Agent Harnessing for an Open World (importance: 40, dev_relevance: 60)
- Echo: Error-Compensated 4-Bit Tensor-Parallel Communication for LLM Training and Inference (importance: 30, dev_relevance: 40)
- Controllable Diverse Sampling through Pullback Directions (importance: 15, dev_relevance: 5)
- ConicFlow: constrained Inference-time optimal control for flow matching models (importance: 15, dev_relevance: 5)
- Which Metadata Matters for LLM Tool Selection in MCP? Evidence from Controlled Factor-Level Interventions (importance: 40, dev_relevance: 75)
- Actionable Meta-Requirements for AI Robustness Auditing (importance: 20, dev_relevance: 25)
- When Pleasing the User Compromises Care: AI Sycophancy in Caregiver-Mediated Healthcare (importance: 25, dev_relevance: 5)
- ACED-Bench: Evaluating How LLM Agents Acquire and Act on Causal Evidence (importance: 30, dev_relevance: 35)
- ANASSA: From High-Level Framework to Low-Level Architecture for Spatial Agentic AI (importance: 25, dev_relevance: 40)
- Measuring the Inference Energy Efficiency of Open-Weight LLMs across Benchmarks, Reasoning Modes, and Hardware (importance: 35, dev_relevance: 45)
- SAT-RL: Training Language Models for Logical Reasoning with Symbolic Solvers (importance: 35, dev_relevance: 40)
- Frustratingly Easy Agentic RAG via Portable Skills: A Case Study in Customer Support (importance: 35, dev_relevance: 60)
- AlignPoison: Conditional Targeted Poisoning of Generative Recommenders via Decoding-Aligned Prefixes (importance: 25, dev_relevance: 20)
- HiPPT: Physiology-Grounded Wearable PPG Representation Learning via Multi-Scope Text Alignment (importance: 15, dev_relevance: 5)
- Reels-RSI: Program-Level Self-Improvement for Code-Rendered Explainer Videos (importance: 35, dev_relevance: 55)
- MontageFlow: Physically Valid High Fidelity EEG Generation (importance: 15, dev_relevance: 5)
- ScienceBuddy-PV: A Verifier-Governed Recursive Improvement Protocol for Panvascular Intervention Research (importance: 15, dev_relevance: 40)
- Energy Ring: Stabilizing Feature Energy for Reliable Test-Time Discovery (importance: 15, dev_relevance: 5)
- FDE-Bench: Evaluating End-to-End Delivery from Underspecified Real-World Business Requests (importance: 30, dev_relevance: 40)
- Are Multimodal Benchmarks Really Useful? Item-Level Multimodal Benchmark Diagnosis via Structure-Response Co-Calibration (importance: 25, dev_relevance: 15)
- PsyMemBehav: Psychology-Informed and Memory-Grounded Two-Level Identity-to-Action Reasoning for Social-Media Behavior Simulation (importance: 20, dev_relevance: 15)
- Learning after Execution with Action-Specific Credit (importance: 25, dev_relevance: 40)
- CB-Agent: Concept-Based In-Context Learning for Sequential Action Selection (importance: 25, dev_relevance: 40)
- Iterate on an LLM, Serve on an SLM (importance: 35, dev_relevance: 60)
- Decoupling Knowledge Forgetting and Refusal Learning in MLLM Unlearning (importance: 15, dev_relevance: 5)
- From Existing Methods to New Research Ideas: Two Perspectives on Recursive Self-Improvement (importance: 25, dev_relevance: 40)
- PEARL: Process-Experience Aware Reinforcement Learning for Multi-Turn Agents (importance: 25, dev_relevance: 40)
- AegisVox: Factorized Privacy and Clinical Preservation for Long-Form Speech (importance: 15, dev_relevance: 10)
- Shaping Latent Mixture-of-Experts: From Expert Geometry to Layerwise Capacity Allocation (importance: 25, dev_relevance: 30)
- RASSG: Repair-Aware Semantic Survival Gating for Streaming Tool Execution (importance: 25, dev_relevance: 45)
- A Hybrid Kinematic GCN and Elastic Transformer for Efficient Indian Sign Language Recognition (importance: 10, dev_relevance: 5)
- AcrossVAM2.0: Conflict-Aware Point Supervision for Parameter-Efficient Video Diffusion Adaptation (importance: 20, dev_relevance: 10)
- Joint Discrete and SE(3) Diffusion for Protein Interface Design (importance: 15, dev_relevance: 3)
- DSABench: Benchmarking Source Discovery, Coordination, and Execution for Multi-Database Agents (importance: 25, dev_relevance: 50)
- How to Build a Benchmark That Actually Measures Taste (importance: 15, dev_relevance: 5)
- Relational Semantic Grounding for Test-Time Discovery (importance: 15, dev_relevance: 5)
- MHWA: Multi-timescale Hierarchical World-Action Model (importance: 30, dev_relevance: 35)
- Think in Strategies, Act with Tools: Traceable Agentic Optimization for Combinatorial Problems (importance: 30, dev_relevance: 40)
- SAGE: Pushing 2-Bit Vector-Quantized LLMs toward Real-World Usability (importance: 35, dev_relevance: 55)
- Equal-information evaluation of credit assignment in modular financial reinforcement learning (importance: 20, dev_relevance: 25)
- GateSight: Recurrent Memory Anticipates Attention Memory for Query-Agnostic KV Eviction in Hybrid LLMs (importance: 35, dev_relevance: 55)
- CUSE: Fine-Grained Multimodal Unlearning with Counterfactual Guidance and On-Policy Self-Distillation (importance: 30, dev_relevance: 30)
- How Should a Camera-First AI Agent Work? (importance: 30, dev_relevance: 35)
- ProgAgent: Learning Tool Orchestration through Programmatic Training (importance: 50, dev_relevance: 70)
- LYNX: Analytical Program Synthesis via Grounding and Closed-Loop Refinement (importance: 30, dev_relevance: 50)
- Counterfactual Regret for Reflective Agents with Lean-Verified Guarantees (importance: 30, dev_relevance: 40)
- GREAT-ESDD: Graph Edge Attention Network for Effectively Steered Delivery Decisions (importance: 15, dev_relevance: 25)
- CESNET-TimeSeries26: A Multi-Year Network Traffic Time Series Dataset for Annual Seasonality, Concept Drift, and Long-Term Anomaly Detection (importance: 20, dev_relevance: 25)
- ASIWorld: Benchmarking Autonomous Agents in a Persistent Survival Sandbox (importance: 25, dev_relevance: 30)
- A Reference Architecture for Integrating Agentic AI into Enterprise Software: Responsibilities, Interactions, and Variability (importance: 40, dev_relevance: 65)
- EvoCowork: A Benchmark for Real-World Office Tasks (importance: 30, dev_relevance: 50)
- Pre-Registered Genesis Accounting for Recursive Self-Improvement: First Audited Operator-Genesis Events in a Physics-Discovery Runtime (importance: 25, dev_relevance: 40)
- PEAM: On-device Communication Agent for Personalized, Efficient and Autonomous Mobile Network Configuration (importance: 25, dev_relevance: 35)
- WeatherBridge: Forecast-Aware Multi-Source Input Adaptation for Pretrained Weather Forecasting Models (importance: 20, dev_relevance: 15)
- Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based Learning (importance: 30, dev_relevance: 40)
- MARL-BattleGrounds: A Novel JAX-native Benchmark for Heterogeneous and Competitive Multi-agent RL (importance: 20, dev_relevance: 35)
- Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents (importance: 20, dev_relevance: 15)
- Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling (importance: 25, dev_relevance: 35)
- WorkBorne: Discovering Agentic Tasks from Real-World Artifact Transitions (importance: 35, dev_relevance: 65)
- SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications (importance: 40, dev_relevance: 60)
- Predicting how model propensities generalize out of distribution (importance: 40, dev_relevance: 25)
- VC-OPD: Visual Counterfactual On-Policy Distillation for Grounded Vision-Language Reasoning (importance: 25, dev_relevance: 25)
- MCPDisCo: Seesaw-Mitigated Tuning for MCP-Native Agents (importance: 40, dev_relevance: 65)
- What Survives When Activation Steering Is Used as Memory (importance: 30, dev_relevance: 35)
- AlphaRLM: Aligning Language Models for Literature-Grounded Symbolic Factor Discovery (importance: 25, dev_relevance: 25)
- SkillHorizon: Constructing Long-Horizon Agent Tasks through Skill Composition (importance: 35, dev_relevance: 55)
- FleetPilot: Fine-Grained Steering Empowers Real-World LLM Personalization (importance: 25, dev_relevance: 25)
- Single-Sample Decoding of a Two-Layer Sound-Word Canvas Model for Thai Klon Pad (importance: 10, dev_relevance: 5)
- Certify the Decision, Not Every Score: Exact Top-k Selection from Few Evaluations (importance: 25, dev_relevance: 40)
- Can You Trust Your Preference Models? Stackelberg Games for Alignment (importance: 30, dev_relevance: 25)
- On the Complexity Gap Between POMDPs and Multi-Environment POMDPs (importance: 15, dev_relevance: 20)
- Belief Embedding Tree Search (importance: 15, dev_relevance: 20)
- Trace Evolution Bench: Measuring Whether Agents Improve from Their Own Environment Traces (importance: 35, dev_relevance: 50)
- Exposing AI Inference as a Telco Network API over the 5G User Plane (importance: 25, dev_relevance: 35)
- Knowledge transfer between simple discrete worlds (importance: 20, dev_relevance: 15)
- Self-Evolving Visual Agent via Margin Reward (importance: 25, dev_relevance: 30)
- Structural Chain-of-Thought: Reasoning with Executable Computational Graphs (importance: 35, dev_relevance: 50)
- GesBench: When and Where Do Multimodal Language Models Perceive Social Meaning? (importance: 15, dev_relevance: 10)
- Trust Hinders Bidirectional Human-AI Alignment (importance: 20, dev_relevance: 15)
- Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback (importance: 20, dev_relevance: 10)
- Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses (importance: 35, dev_relevance: 60)
- Can agents whose sole duty is to keep humans alive be hijacked to kill? Delayed multi-turn derived-action trap attacks in simulated high-stakes environments (importance: 40, dev_relevance: 40)
- Continuous Risk-Based Scheduling for Food Safety Inspection: Integrating Historical Decay with Event-Driven Consumer Review Signals (importance: 15, dev_relevance: 20)
- SocraticRepair: Planning-Guided Socratic Tutoring for Program Debugging and Misconception Repair (importance: 25, dev_relevance: 35)
- Running Fewer Loops Removes Refusal Behaviour in Recurrent-Depth Language Models (importance: 30, dev_relevance: 25)
- HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systems (importance: 35, dev_relevance: 55)
- Understanding Unified VLMs: From Attention Probing to Guidance Tuning for Multimodal Generation (importance: 25, dev_relevance: 15)
- Words to World: Benchmarking LLMs in Interactive Physical Environments (importance: 25, dev_relevance: 25)
- Using Evidence-Centered Design to Guide Prompt Engineering in Measurement: An Illustration with Automated Essay Scoring (importance: 20, dev_relevance: 30)
- MoMaTune: Model, Match, Then Tune Language Models for Numerical Records (importance: 25, dev_relevance: 30)
- ClinEdge: An Expert-Graded Benchmark of Real Patient Edge Cases Where Frontier Multimodal Models Fail (importance: 20, dev_relevance: 5)
- RECAP: Calibrating Low-Bit Reasoning Models on the History They Read (importance: 35, dev_relevance: 50)
- OntoAR-HAR: Ontology-Guided Agentic RAG with Personalized Memory for Human Activity Recognition (importance: 15, dev_relevance: 15)
- CHEAR: Confusion-Hierarchy Evidence-Aware Reasoning for Cost-Efficient Human Activity Recognition (importance: 15, dev_relevance: 15)
- SuperAI: Merging Capabilities Across Heterogeneous Foundation Models (importance: 40, dev_relevance: 35)
- MultiFlow: Vision-Driven Multimodal fMRI Encoding via Stimulus-Informed Source Flow Matching (importance: 15, dev_relevance: 5)
- TRACE: Temporal Risk Assessment and Concept Editing for Safer Language Generation (importance: 25, dev_relevance: 30)
- Fewer Bits, More Tries: Recovering Capability Under Memory Constraints (importance: 30, dev_relevance: 45)
- Analyzing the Dynamics of WSBM-based Echo State Networks using InputDSA (importance: 15, dev_relevance: 10)
- A Decade of PhotoMappers: A Longitudinal Study of Crowdsourced Disaster Photos Geolocalization with Vision-Language Models and Geospatial Reasoning (importance: 15, dev_relevance: 10)
- State Representation Shapes Long-Horizon Agent Learning and Transfer: An Empirical Study (importance: 35, dev_relevance: 50)
- CoverSynth: Coverage-Guided Generation of Retrieval Evaluation Sets (importance: 35, dev_relevance: 60)
- SpecFit: Not All Draft Tokens Cost the Same in Offloaded MoE Inference (importance: 30, dev_relevance: 55)
- Hidden in Plain Sight: Advancing Harmful Content Understanding via Perception-Aware Adversarial Self-Play (importance: 25, dev_relevance: 25)
- Scalable Robust MCTS with Anytime Deterministic Optimality Guarantees (importance: 20, dev_relevance: 25)
- MemeLogos: Knowledge-Driven Retrieval-Augmented Reasoning for Harmful Meme Detection (importance: 15, dev_relevance: 10)
- SkillFusion: Towards Reliable Multi-Skill Composition for LLM Agents (importance: 40, dev_relevance: 65)
- When Skills Remain But Access Fails: Factorized Skill Memory for Continual Skill Internalization (importance: 35, dev_relevance: 45)
- Large Decision Models: Repurposing Large Language Models as Single-Pass Decision Makers (importance: 30, dev_relevance: 35)
- Data Portability Revisited: Toward the Human-Centric, AI-Driven Data Ecosystems of Tomorrow (importance: 15, dev_relevance: 10)
- Agent-Canary-Mas: End-to-End Evaluation of Protocol Risk in Long-Horizon Multi-Agent Systems (importance: 35, dev_relevance: 50)
- Self-Representation Alignment in Diffusion Models for Video Background Music Generation (importance: 15, dev_relevance: 5)
- Inverse Garment Design with Visual Agents (importance: 15, dev_relevance: 10)
- CondMSP: Condition-Informed and Resource-Aware Inorganic Materials Synthesis Planning (importance: 15, dev_relevance: 5)
- From Stages to States: Unified Iterative Reasoning for MLLMs (importance: 25, dev_relevance: 20)
- Big-Big-SycoBench: A Benchmark to Detect Sycophancy Through Model Responses to Valid vs. Invalid Arguments (importance: 25, dev_relevance: 25)
- Recall What Is Missing: Composing Evidence from Long-Term Agent Memory (importance: 35, dev_relevance: 45)
- Knowledge Interaction in Hypernetwork-based Continual Learning: Conflict and Composition (importance: 20, dev_relevance: 20)
- Radioactive Feedback: Tracing Models Trained by Label-Only LLM Supervision (importance: 30, dev_relevance: 25)
- FARL: A Fully Automated Pipeline of RL Interface and Solver Design (importance: 25, dev_relevance: 30)
- SpecForge: Recursive Self-Improvement for Coding Agents via Behavioral Specifications (importance: 45, dev_relevance: 75)
- Suggested Is Not Superseded: A Pre-Registered Evaluation of a System-One Model as a Memory-Retirement Gate (importance: 30, dev_relevance: 40)
- MedHarnessBench: A Unified Benchmark for Healthcare Agent Evaluation (importance: 25, dev_relevance: 30)
- History Replays Itself: Unleashing Long-Context Capability in Linear Attention (importance: 30, dev_relevance: 25)
- Visible but Not Avoided: State-Dependent Re-Proposal of Rejected Candidates in Catalog-Based LLM Optimization Loops (importance: 30, dev_relevance: 50)
- Google Earth AI Geospatial Reasoning Agent: Capabilities and a Systematic Evaluation (importance: 35, dev_relevance: 35)
- Agent Sociodrama: Social-Dynamics-Guided Persona Recasting for LLM Agent Teams (importance: 25, dev_relevance: 30)
- Passing Tests, Missing Semantics: Evidence of Interpretive Lock-In in Plan-Based DataFrame Code Generation (importance: 35, dev_relevance: 75)
- Attention Sinks in VLMs Are Borrowed, Not Seen (importance: 20, dev_relevance: 20)
- Engram Injection: A Study of Temporal Validity in Parametric Language-Model Memory (importance: 20, dev_relevance: 15)
- RetryBench: Separating Retry Permission from Recovery Outcomes in GPT Tool Use (importance: 40, dev_relevance: 65)
- Comparative Study of Canvas- and Coordinate-Based Geometry-Conditioned Neural Networks for Real-Time Prediction of Indoor Environments with Varying Room Dimensions (importance: 10, dev_relevance: 5)
- LURE: a Phase-Labelled Synthetic Corpus of Online-Grooming Exchanges (importance: 15, dev_relevance: 5)
- Multi-Horizon Explainable Queue Forecasting and Resource-Aware Decision Support for Intelligent Border Control Using LightGBM (importance: 10, dev_relevance: 15)
- The Anatomy of a Blind Spot: What an LLM Judge Must Be Shown, and How Much (importance: 35, dev_relevance: 50)
- Vision: "What Don't I Know About User's Preferences?" How Uncertainty Estimation Improves Proactive Assistant Alignment (importance: 20, dev_relevance: 15)
- ADAPT: Audio-Depth Action Planning with Transition-Aware Rewards for Lights-Out Industrial Navigation (importance: 20, dev_relevance: 15)
- Small but Mighty: Test Suites as Improvement Signals for Self-Evolving Agents (importance: 40, dev_relevance: 75)
- PopGen: Generating Semantically Diverse Populations for Zero-Shot Coordination (importance: 20, dev_relevance: 20)
- Revocation Is Not Complete at Receipt: An Executable Outcome Contract for Distributed Agent-Tool Gateways (importance: 35, dev_relevance: 60)
- Handoff Curves: Initial Selection, Later Handoff, and Supervision Targets in Multi-turn LLM Routing (importance: 35, dev_relevance: 50)
- Large language model-generated patterns for epidemiological modeling: a multi-benchmark evaluation framework (importance: 15, dev_relevance: 5)
- Computational Reflective Equilibrium for Agentic AI: A Rawlsian Alignment Scaffold for Governance (importance: 25, dev_relevance: 25)
- Cross-Protocol Personality Assessment with LLMs: Aligning Behavioral Evidence with Human Ratings (importance: 10, dev_relevance: 5)
- Understanding and Reducing Misalignment in Persuasion Optimization (importance: 45, dev_relevance: 15)
- Beyond Single-Pathogen Forecasts: Predicting Shared Respiratory Hospital Burden from Influenza, COVID-19, and RSV (importance: 15, dev_relevance: 5)
- Heterogeneous Agentic Coevolution (importance: 30, dev_relevance: 35)
- The Sharp Directions Are Against You: Curvature Analysis of Activation Steering (importance: 30, dev_relevance: 15)
- GUIDE: Computationally Augmented Consensus for Trustworthy Clinical Practice Guidelines (importance: 15, dev_relevance: 5)
- Constructed regional photovoltaic generation fields improve ultra-short-term distributed solar forecasting (importance: 12, dev_relevance: 5)
- Beyond Perfect Teammates: Multi-Agent Collaboration with Unreliable Colleagues (importance: 30, dev_relevance: 45)
- Imitation Holds a Key to Evolution: Localized Repair and Verified Skill Acquisition for Adaptive Agents (importance: 35, dev_relevance: 55)
- Large Language Model Assisted Transportation Planning through a Differentiable User Equilibrium Layer (importance: 15, dev_relevance: 15)
- Phantom Merge: When Your Large Language Model Agents Pick One but Tell You About Another (importance: 30, dev_relevance: 40)
- Pollard: Governed Semantic Execution Trees for Framework-Independent AI Agent Control and Replay (importance: 40, dev_relevance: 75)
- Uncovering Robustness Weaknesses in LLM Agents via Persona Evolution with Task-Integrity Constraints (importance: 25, dev_relevance: 30)
- Interactive 4D Volumetric Liquid Forecasting under Moving-Solid Interaction (importance: 12, dev_relevance: 5)
- ProCARE: Real-World Study Automation via Profile-Grounded Evidence Contracts (importance: 20, dev_relevance: 25)
- Label-Centric Knowledge Retrieval for Zero-Shot Text Classification (importance: 25, dev_relevance: 30)
- Correct Choices, Failed Sequences: Choice2Seq Benchmarks Correctness Retention Across Output Interfaces (importance: 25, dev_relevance: 30)
- OneShot: Label-Efficient Detector Weighting for Time Series Anomaly Detection (importance: 20, dev_relevance: 25)
- Learning to GUIDE the Group: Self-Learning LLM Facilitation Across Multi-Party Tasks (importance: 20, dev_relevance: 10)
- ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context (importance: 20, dev_relevance: 15)
- Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution (importance: 25, dev_relevance: 40)
- agent-risk: Risk Assessment for Changes to AI Agents (importance: 35, dev_relevance: 55)
- REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces (importance: 30, dev_relevance: 30)
- Memory-Efficient Integer-Only Training of Deep Neural Networks (importance: 30, dev_relevance: 35)
- FON: A Draft-Revise Dual-Organ Language Model without Sequence-Level Quadratic Attention (importance: 25, dev_relevance: 20)
- Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment (importance: 30, dev_relevance: 15)
- ReTAC: Residual Test-Time Agentic Computation for Video Spatial Reasoning (importance: 20, dev_relevance: 15)
- HyWe: Unifying Search and Operation for Real-World Hybrid Web Tasks (importance: 30, dev_relevance: 40)
- When a Conceptor Complement Becomes Gain: Operator Geometry and Behavioral Attribution in Language Models (importance: 25, dev_relevance: 15)
- TRAP: Tool-result Replay Attack for Probing Misbinding in LLM Agents (importance: 35, dev_relevance: 60)
- Efficient Multi-Agent Reasoning via Confidence-Guided Adaptive Debate (importance: 30, dev_relevance: 25)
- ChronoScribe: Semantic-Numerical Specialization for Instruction-Driven Time Series Forecasting (importance: 15, dev_relevance: 10)
- World Models Need a Physical Registry: Learning Internal Predictive States for Physically Plausible World Models (importance: 20, dev_relevance: 10)
- RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM-Based Retrosynthesis Planning (importance: 20, dev_relevance: 10)
- Recursive Environment Scaling for General Long-Horizon Agents (importance: 30, dev_relevance: 40)
- EvoWeb: A Benchmark for Coding Agents on Evolving Web Applications (importance: 40, dev_relevance: 75)
- SLS-Bench: A Benchmark for Incident Log Summarization with Synthetic Observability Data (importance: 25, dev_relevance: 50)
- Chain Dimension Scaling: Communication-Efficient Loop-Style Scaling for Industrial Short-Video Ranking (importance: 20, dev_relevance: 25)
- AuthGateBench: Benchmarking LLMs as Authorization Gate (importance: 45, dev_relevance: 70)
- RAP-WAM: Learning World-Action Models through Predictive Rehearsal and Action Perturbations (importance: 15, dev_relevance: 5)
- XAI-Guided Fourier Domain Generalisation for Pure-Simulation Sim-to-Real Deep Reinforcement Learning (importance: 15, dev_relevance: 10)
- PXRD-Bench: How Close Are We to Determining Crystal Structures from Powder Diffraction? (importance: 15, dev_relevance: 5)
- Boosting the Learning for Ranking Patterns (importance: 12, dev_relevance: 20)
- EAU: Environment-Agent-User Co-evolution for Training User-Centric Agents (importance: 30, dev_relevance: 30)
- Does the Explanation Drift Before the Guarantee? Label-Free Early Warning on Evolving Graphs (importance: 15, dev_relevance: 15)
- Overtaught, Underlearned: Learner-Centered Distillation from Multi-Agent Discussion (importance: 25, dev_relevance: 25)
- Making Language the Low-Rate Code of Vision: Successive Refinement from Words to Pixels (importance: 20, dev_relevance: 10)
- AI Discovery in Combinatorics: From Problems to Solutions and Back to Bigger Problems (importance: 35, dev_relevance: 10)
- G$^2$WAM: World Action Models for Efficient and Geometry Grounded Image-Goal Navigation (importance: 15, dev_relevance: 5)
- Double-Rollout Reinforcement Learning with Sampling-based Reward Estimation (importance: 30, dev_relevance: 35)
- When 1+1 Backfires: Rethinking Relevance-Diversity Balancing for Sample-Aware Visual Token Reduction (importance: 20, dev_relevance: 25)
- Weak Segmentation-Guided GAN for~Realistic Color Edition (importance: 10, dev_relevance: 5)
- LATCH: LLM-Assisted Timing Closure Harness for Post-Route FPGA Optimisation (importance: 25, dev_relevance: 40)
- Does a Hosted Decision Model Respect Irrelevant Alternatives? A Pre-Registered Test of Option-Set Dependence in Jev (importance: 25, dev_relevance: 40)
- Structural Baselines Matter for Agent Failure Detection and Attribution (importance: 30, dev_relevance: 55)
- Trace-to-Architecture: Capability Bottleneck Diagnosis for Harness Revision (importance: 35, dev_relevance: 65)
- Queue-Aware Token Communications: Bridging AI Representation and Wireless Service (importance: 12, dev_relevance: 10)
- Separating Prompt Harmfulness from Interaction Safety via Partial Generation (importance: 25, dev_relevance: 30)
- Geospatial Vector Geometry as a Native Spatial Modality for Vision-Language Models (importance: 15, dev_relevance: 10)
- Game-Creation-Agents: Autonomous Game Production (importance: 30, dev_relevance: 50)
- LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints (importance: 15, dev_relevance: 5)
- Same Operation, Different Permission: Evaluating Whether Guardrails Honor Authorization (importance: 35, dev_relevance: 50)
- CLIVO: Differentiable Cluster-Based Instruction Mixing with Learned Validation Objectives (importance: 15, dev_relevance: 25)
- Acceptable Misalignment (importance: 12, dev_relevance: 5)
- Transparency as Answerability: Role-Dependent Information Flow Modeling (importance: 15, dev_relevance: 10)
- Observe, Share, or Predict? Disentangling Partner Information in Embodied Multi-Agent Collaboration (importance: 20, dev_relevance: 15)
- Neural Bayesian Networks for Modular and Scalable Decision-Making Under Uncertainty (importance: 20, dev_relevance: 25)
- Audit tools of the trade: Automated Auditor Elicitation Can Be Sensitive to the Context the Auditor Is Given (importance: 35, dev_relevance: 50)
- Transparent Information Flow Modeling for Role-Aware AI Act Assessment (importance: 25, dev_relevance: 15)
- WeFair: Fidelity-Aware Alignment from Human Feedback for Personalized Query Rewriting in Short-Video Search (importance: 15, dev_relevance: 10)
- MADDIE: Multi-Agent Drug Discovery Engine (importance: 15, dev_relevance: 5)
- PRAISE: A Reproducible Workflow for Non-Programmer Scientists to Implement Published AI Models (importance: 25, dev_relevance: 20)
- Glasseek: A Transparent Recipe for Training and Trustworthy Evaluation of DeepSearch Agents (importance: 30, dev_relevance: 40)
- Understanding Personalization Gains in Whisper (importance: 20, dev_relevance: 25)
- Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints (importance: 10, dev_relevance: 10)
- Neos: Learning to Interleave Search and Visual Reasoning (importance: 25, dev_relevance: 20)
- VL-DocIR: A Benchmark for Vision-Based Long Document Information Retrieval (importance: 15, dev_relevance: 15)
- PatientWM: Towards Generalizable Clinical World Models via Unified Patient States (importance: 15, dev_relevance: 5)
- SFT Leaves a Reproducible Signature in Activation Geometry: Calibrating Post-Training Interpretability Against Run-to-Run Noise (importance: 25, dev_relevance: 20)
- When LLM Agents See a Conflict but Still Proceed: Cross-Domain Dependencies in Network Slicing (importance: 25, dev_relevance: 30)
- EvoQuantLab: Training-Free Dual-Loop Evolution for Auditable Quantitative Research (importance: 20, dev_relevance: 25)
- The Format is the Filter: Scoring Format Fixes What an LLM Judge Can Express (importance: 25, dev_relevance: 35)
- Learning Perceptual Geometry of Natural Odor Mixtures from LLM-Derived Descriptors (importance: 10, dev_relevance: 3)
- ChemExplain: A Multimodal Benchmark for Explanatory Chemical Reasoning (importance: 12, dev_relevance: 5)
- Size Generalization for Diffusion Routing Solvers (importance: 12, dev_relevance: 10)
- Releasing the Inner Jev: A Simple Recipe for Open Decision Models (importance: 25, dev_relevance: 35)
- AutoMARS: Automatic and Scalable Multi-Agent Large Language Model Framework for Supply Chain Risk Assessment (importance: 15, dev_relevance: 15)
- IsoVRBench: Video Reward Modeling Without Human Feedback (importance: 15, dev_relevance: 15)
- Does a Governed Confidence Score Know When a Text-to-SQL Answer Is Right? An Exploratory Study on BIRD and a Pre-registered Confirmation on Spider (importance: 30, dev_relevance: 50)
- RL Post-Training Diffusion Language Models Along Their Own Generation Paths (importance: 25, dev_relevance: 25)
- Reasoning for Propaganda Detection via Label-Guided CoT Optimisation (importance: 15, dev_relevance: 10)
- PhysCodeGym: Benchmarking Coding Agents for Physically Verifiable 4D Reconstruction (importance: 25, dev_relevance: 45)
- Towards Resilience: Slice-Driven Orchestration of Mobile IoMT in 6G-Edge-Continuum (importance: 12, dev_relevance: 10)
- The Agent Knew: Internal States Forecast Unauthorized Actions and Stop Them, and the Lies That Follow (importance: 45, dev_relevance: 45)
- Reliable Text-to-SQL in Resource-Constrained Settings with Small LLMs via Iterative Retry Workflows (importance: 30, dev_relevance: 50)
- Beyond Successful Play: Diagnosing Opponent Modeling in Large Language Models (importance: 25, dev_relevance: 15)
- Do Theorem Provers Need Theorem-Proving Agents? mini-Lean-agent: A Coding-Agent Baseline for Lean (importance: 40, dev_relevance: 60)
- Budgeted KV-Cache Repair for Evolving Contexts (importance: 35, dev_relevance: 55)
- When Goals Leak Across Relationships: How Cached Goal Representations Shape Cross-Client Decisions (importance: 25, dev_relevance: 35)
- What Makes a Latent Space Shared? Direct and Recoverable Compatibility in Motion VAEs (importance: 10, dev_relevance: 10)
- Get In, Decide, Prove: A Merchant-Side Readiness Benchmark for Agentic Commerce (importance: 30, dev_relevance: 40)
- EPOCH: Coverage-Aware Evolution with Experiential Preference Optimization for LLM-based Automated Heuristic Design (importance: 20, dev_relevance: 30)
- Learning to Play: Reinforcement Learning and Games (importance: 10, dev_relevance: 10)
- Generalizable field of view extension for diffusion MRI via conditional flow matching (importance: 15, dev_relevance: 5)
- Agency as a New Organizing Abstraction for Computer Science (importance: 35, dev_relevance: 40)
- SchemaWeaver: Accelerating On-Device Mobile Agentic AI via Schema-Guided Parallel LLM Decoding (importance: 30, dev_relevance: 45)
- Large Language Models Extend Known Research But Do Not Create New Research Directions (importance: 25, dev_relevance: 10)
- An FPGA-Based Spiking Neural Network for Spatiotemporal Pattern Learning and Classification (importance: 15, dev_relevance: 20)
- REA: From Correct LLM Judgments to Safe Decision Delivery (importance: 30, dev_relevance: 50)
- TacitVLA: Reasoning Vision-Language-Action Models Without Reasoning Latency (importance: 25, dev_relevance: 15)
- Reference Wording Shapes Retrieval of Robot Task Instructions (importance: 10, dev_relevance: 15)
- RAN-JEPA: An Action-Conditioned World Model for Energy-efficient Mobile Network Control (importance: 15, dev_relevance: 10)
- Verifier-Harness Co-Evolution for Training-Free Self-Improvement on Open-Ended Tasks (importance: 40, dev_relevance: 60)
- OpenDepth: Repurposing Image Editing Priors for Robust Open-World Monocular Depth Estimation (importance: 10, dev_relevance: 5)
- MAS-Gate: Unlocking Multi-Agent System Lifecycle via Node-Centric Quantification (importance: 30, dev_relevance: 50)
- Executable Prefix Values for Locating Agent Failures (importance: 40, dev_relevance: 60)
- Beyond Bias Scores: Answerability Confounds in Activation Steering with Gradient-Constrained Direction Design (GCD) (importance: 15, dev_relevance: 10)
- When Should Enterprise Policy Decisions Be Learned, and When Should They Be Reasoned? (importance: 30, dev_relevance: 40)
- Can Agents Truly Forget? Explore Machine Unlearning for LLM Agents (importance: 25, dev_relevance: 35)
- RoTA-Bench: Evaluating 3D Spatial Reasoning through Interactive Projection Alignment (importance: 10, dev_relevance: 5)
- Cowork-Gym: Synthesizing Interactive Work Environments at Scale (importance: 30, dev_relevance: 45)
- Encounter-Driven City-Scale Human Mobility Generation with Large Language Models (importance: 10, dev_relevance: 10)
- Stable Token-level Scheduling for Serverless LLM Inference (importance: 30, dev_relevance: 50)
- MedEvo: Medical Reasoning Evolution through Knowledge Navigation (importance: 15, dev_relevance: 10)
- Understanding Gene Expression-to-Image Learning: Spatial Structure and Information Fidelity (importance: 10, dev_relevance: 5)
- A Node Embedding Framework for Model-Centric Conformance Checking in Process Mining (importance: 10, dev_relevance: 15)
- Adaptive-Group Process Rewards for GUI Agents (importance: 25, dev_relevance: 40)
- ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution (importance: 15, dev_relevance: 25)
- Keep Your Receipts: Trustworthy Cleanup of Agent-Built Knowledge Graphs (importance: 30, dev_relevance: 45)
- Upstream Influence, Downstream Re-Anchoring in LLM Agent Chains (importance: 25, dev_relevance: 35)
- Evidence-First Agentic Verification (importance: 35, dev_relevance: 55)
- Safety for Whom? Benchmarking Understanding Beyond Refusal in Egocentric Multimodal Assistants (importance: 15, dev_relevance: 10)
- DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models? (importance: 15, dev_relevance: 15)
- Coupled Harness--Model Evolution for Scientific Equation Discovery (importance: 25, dev_relevance: 40)
- Instructions Are Evidence, Not Objectives: Measuring Outcome Alignment of Agents (importance: 35, dev_relevance: 60)
- Autonomous 3D Visual-Thermal Inspection for Utility-Scale Solar Farms (importance: 15, dev_relevance: 10)
- Memory Echo in Single-Agent Long-Term Memory: Lifecycle Evaluation and Provenance-Guided Mitigation (importance: 35, dev_relevance: 55)
- Agentic4PDM: Host-Governed Evidence Admissibility for Industrial Fault Diagnosis (importance: 20, dev_relevance: 35)
- Empowering Edge Agents: A Systematic Analysis of Edge-Cloud Collaboration in LLM Agent Execution (importance: 35, dev_relevance: 55)
- Benchmarking and Training Reward Models for Cross-Source Evidence Grounding in Search Agents (importance: 30, dev_relevance: 40)
- Learning Adaptive Deliberation Control for Reliable Recognition with Multi-Agent Language Models (importance: 25, dev_relevance: 35)
- Socratic Skill Learning: Interleaving Teaching and Practice for Transferable Agent Skills (importance: 40, dev_relevance: 60)
- ECMR: Evidence-Aware Deliberation Control for Multi-Agent LLM Reasoning (importance: 25, dev_relevance: 35)
- Training LLMs for Multi-Agent Collaboration with Verifiable Information Rewards (importance: 25, dev_relevance: 30)
- PaceReasoner: Label-Free Learning to Answer Briefly from Long Reasoning (importance: 25, dev_relevance: 40)
- Learning How to Remember: Memory Modality Selection for GUI Agents (importance: 25, dev_relevance: 35)
- Seeing Depth but Choosing Cues in Vision-Language Models (importance: 10, dev_relevance: 5)
- SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces (importance: 20, dev_relevance: 25)
- Beyond Black Boxes: Visual Grounding for Text Embeddings (importance: 15, dev_relevance: 15)
- Can Agents Forget When Asked? A Benchmark for Memory Unlearning in LLM Agents (importance: 25, dev_relevance: 40)
- WorkRover: Scaling Long-Horizon Cowork Agent Training Via Recursive Environment Evolution (importance: 25, dev_relevance: 40)
- AReaL-RSI: Adaptive Scaling and Coordination of Recursive Self-Improvement (importance: 30, dev_relevance: 40)
- ResearchFlow-Eval: Tracing the Execution Gap in End-to-End Multi-Claim Autoresearch (importance: 25, dev_relevance: 35)
- One-Sided Verification Under Limited Expert Review (importance: 20, dev_relevance: 20)
- AM-Bench: Understanding Agents under Task Underspecification across Models and Harnesses (importance: 35, dev_relevance: 55)
- LifeSearch: Benchmarking Deep Research Agents for Real-World Life and Productivity (importance: 20, dev_relevance: 20)
- CompanionSafe: From Multi-turn Safety Evaluation to Guardrails for AI Companions (importance: 20, dev_relevance: 10)
- Chain of Thought Monitoring Can Fail Even When CoT Is Necessary (importance: 40, dev_relevance: 25)
- Fresh at Decision, Stale at Action: Routing New Information to LLM Decisions Already Made (importance: 25, dev_relevance: 35)
- Monitoring latent reasoning in looped language models (importance: 20, dev_relevance: 15)
- Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs (importance: 15, dev_relevance: 10)
- RadixFlow: Filling Execution Bubbles in Prefix-Cached Hybrid LLM Serving (importance: 25, dev_relevance: 50)
- When and How Order-Flow Interventions Work After Dealers Adapt (importance: 10, dev_relevance: 10)
- Back to the Future: Reverse-Consequence Modeling for Long-Horizon Particle Dynamics (importance: 10, dev_relevance: 5)
- Neuro-Symbolic Graph Transformers for Hierarchical Intention Recognition from Multimodal Human Activity Sequences (importance: 10, dev_relevance: 10)
- QORH: Learning Query-Specific Ordered Rule Hypergraphs for Inductive Knowledge Graph Link Prediction (importance: 10, dev_relevance: 15)
- Consider, Then Choose: Order-Invariant Decisions in System One Models (importance: 25, dev_relevance: 35)
- Beyond Answer Sharing: TrueMAD for Targeted Peer Engagement in Multi-Agent Debate (importance: 20, dev_relevance: 30)
- Output Discipline Sets the Floor: A Falsification-Guided, Model-Agnostic Agent Protocol for Auditable Hypothesis-Driven Decision Making (importance: 15, dev_relevance: 30)
- A dataset of formalised theorem statements from the Annals of Mathematics (importance: 25, dev_relevance: 30)
- Efficient Safety Guardrails through Latent Dual-Hypothesis Reasoning with On-Demand Trace Reconstruction (importance: 20, dev_relevance: 35)
- Complex materials engineering via Provenance Action Graphs: potential, structure and demonstration (importance: 10, dev_relevance: 15)
- AssetForge: Asset-Augmented Task Synthesis for Business Automation Agents (importance: 25, dev_relevance: 40)
- Choosing LLMs for Strategic Agent Interactions (importance: 25, dev_relevance: 30)
- ABot-ER: Unified Embodied Reasoning through Agentic Data-Model Co-Evolution (importance: 25, dev_relevance: 15)
- Delegating Earth science model calibration to knowledge-guided LLM agents (importance: 20, dev_relevance: 30)
- The Singularity Has a Distribution: Exact Timing Laws and Detection Limits for a Stochastic Intelligence Explosion (importance: 25, dev_relevance: 15)
- Thinker-WM: A Unified World Model with Shared Multimodal Representations for Embodied Intelligence (importance: 15, dev_relevance: 10)
- Autonomous Science Bench: Evaluating AI Agents for Autonomous Scientific Discovery (importance: 25, dev_relevance: 25)
- ReTerm: Recovery Training for Self-Improving Terminal Agents (importance: 35, dev_relevance: 60)
- Judges Follow the Reference: The Derivation Gap in Verifying Tool-Call Arguments (importance: 35, dev_relevance: 60)
- The Smart Buildings Challenge: Learning Foundation Models for Real-World Optimization at Scale (importance: 15, dev_relevance: 10)
- Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale (importance: 45, dev_relevance: 65)
- Input-Conditioned Search for Efficient Test-Time Scaling of Recursive Reasoners (importance: 20, dev_relevance: 25)
- Rethinking Quality in Large Language Model Routing: Security-Conditioned Model-and-Budget Selection (importance: 30, dev_relevance: 45)
- Configuring RAG Systems: What Breaks, Where Originates, and How to Fix (importance: 35, dev_relevance: 60)
- Order-Consistent Latent Geometry for Reliable Short-Horizon Planning in World Models (importance: 15, dev_relevance: 15)
- LLM Agents for Automated Post-Training via Structured Recipe Optimization (importance: 25, dev_relevance: 30)
- Reading Right, Acting Wrong: Diagnosing Decision Failures of System One Models in Agent Loops (importance: 25, dev_relevance: 40)
- Compare to Correct: Pairwise-Guided Logit Adjustment for LLM Test-Time Scaling (importance: 20, dev_relevance: 30)
- Auditing Future-Memory Use in Point-in-Time Language Agents (importance: 20, dev_relevance: 25)
- AgentStripes: An Evidence-Graded Maturity Model for AI Agents (importance: 35, dev_relevance: 50)
- Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces (importance: 20, dev_relevance: 25)
- Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents (importance: 40, dev_relevance: 55)
- Two Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference (importance: 12, dev_relevance: 20)
- To Align or Not To Align: Check Your COMPASS Before You Train (importance: 35, dev_relevance: 45)
- Knowing the User Is Not Knowing What We Established: Grounding Awareness in Personal Agents (importance: 35, dev_relevance: 50)
- Unlocking Token Rewards via Training-Free Reward Attribution (importance: 30, dev_relevance: 35)
- Beyond the Verdict: Agentic Code Judging with Verifiable Feedback (importance: 45, dev_relevance: 70)
- Jeva: System One Models from First Principles (importance: 20, dev_relevance: 30)
- BLADE: Auditing Information Use for Failure Attribution in LLM Multi-Agent Systems (importance: 40, dev_relevance: 60)
- Pack Light, CarryOn: Building Object-Oriented Context for Long-Horizon Agents (importance: 42, dev_relevance: 70)
- Self-Evolving Agent Harness (importance: 42, dev_relevance: 65)
- Read the Room Before Your Model Generates (importance: 25, dev_relevance: 40)
- Who Improves the Robot? Toward Physical Recursive Self-Improvement (importance: 15, dev_relevance: 15)
- MFG: Multi-turn Feedback Guided Molecular Optimization (importance: 12, dev_relevance: 10)
- AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewards (importance: 30, dev_relevance: 45)
- Subgraph-Aware Training of Language Models for Knowledge Graph Completion Using Structure-Aware Contrastive Learning (importance: 15, dev_relevance: 15)
- EmoXpt: Analyzing Emotional Variances in Human Comments and LLM-Generated Responses (importance: 10, dev_relevance: 10)
- Latent Algorithmic Structure Precedes Grokking: A Mechanistic Study of ReLU MLPs on Modular Arithmetic (importance: 20, dev_relevance: 15)
- Selective Knowledge Control for Continual GUI Agent Learning over Application Streams (importance: 22, dev_relevance: 35)
- Dependency, Compression, and Synergy: A Unified Information-Theoretic View of Multimodal Learning (importance: 18, dev_relevance: 15)
- MRFlow: Mechanism-Structured Residual Flow Matching for Virtual Cell Perturbation Prediction (importance: 12, dev_relevance: 10)
- Predict Before You Modify: Learning Improvement Dynamics for Self-Evolving Agents (importance: 38, dev_relevance: 60)
- Posterior Flow Matching (importance: 20, dev_relevance: 25)
- GameSmith: Scaling Zero-Sum Self-Play with Synthesised Games (importance: 28, dev_relevance: 35)
- Splicing Hints into Reasoning Traces: Learning from Demonstrator Failures (importance: 25, dev_relevance: 30)
- SharedSAE: a unified sparse dictionary across u-net blocks and denoising time for diffusion models (importance: 15, dev_relevance: 15)
- Enhanced Trading Systems with Intelligent Market Analysis (importance: 10, dev_relevance: 10)
- SemTree: Judge-Aware Semantic Decision Trees for Inspectable and Repairable Rule Decisions (importance: 25, dev_relevance: 40)
- VH-GPO: Verified Handoff for LLM Agent Reinforcement Learning (importance: 28, dev_relevance: 40)
- PreSO: Privileged Representation Isolation for On-Policy Self-Distillation (importance: 25, dev_relevance: 30)
- Showing versus Doing: Teaching by Demonstration (importance: 10, dev_relevance: 10)
- Coding LLMs Can Be Harnessed to Learn Interpretable Models from Data (importance: 35, dev_relevance: 50)
- What Can Language Models Inherit from Solomonoff Induction? Limits, Attainable Properties, and What Served Conditionals Reveal (importance: 15, dev_relevance: 10)
- When Log Parsing Removes What Matters: Detecting Failures in Distributed GPU Workloads (importance: 30, dev_relevance: 55)
- Physics-Informed State Space Models for Day-Ahead Photovoltaic Power Forecasting (importance: 8, dev_relevance: 10)
- Bias Robustness of Genetic Multi-Criteria Sorting versus Learned Rankers under Proxy Skew and Label Contamination (importance: 12, dev_relevance: 15)
- Barycentric Anchored Weight Search for Robust Model Merging (importance: 18, dev_relevance: 25)
- RoboHarness: A Simple Harness Outperforms VLA and World Actions Models (importance: 30, dev_relevance: 40)
- LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures (importance: 12, dev_relevance: 5)
- CoEvoke: Evoking Multi-Agent Collaboration for Trajectory Collection (importance: 28, dev_relevance: 40)
- OPMD: On-Policy Mechanism Distillation Beyond Output Supervision (importance: 22, dev_relevance: 25)
- Contrastive Covariance Subspaces for Activation Steering (importance: 22, dev_relevance: 25)
- Pick the Mixture, Not the Samples: Combinatorial Data Selection for LLMs (importance: 25, dev_relevance: 30)
- Can We Trust This Evaluation? Instance-Level Reliability of LLM-as-a-Judge (importance: 30, dev_relevance: 40)
- Rotate the Loop: Blockwise Conformality for Looped Transformers (importance: 18, dev_relevance: 15)
- PLDiv: Measuring Dataset Diversity and Semantic Collapse via Persistence Landscapes (importance: 22, dev_relevance: 35)
- A Geometric Perspective on Reward Function Updates in Inverse Reinforcement Learning (importance: 10, dev_relevance: 10)
- Reinforcement Learning from Real-Robot Interaction without High-Quality State Estimates (importance: 12, dev_relevance: 10)
- The Future of AI-assisted Physics: Discovering, Verifying and Organizing Results via Formalization (importance: 18, dev_relevance: 25)
- Auditing LLM Program Search for Bermudan Stopping with Primal--Dual Certificates (importance: 15, dev_relevance: 30)
- Loose Lips: Understanding Privacy Leakage in Agentic Dialogue (importance: 30, dev_relevance: 35)
- Is the Future Predictable? Modeling Future Predictability for Reliable Temporal Reasoning (importance: 20, dev_relevance: 15)
- One Model, Multiple Compute Budgets: Multi-Ratio ConceptMoE for Flexible Compute Across Turns (importance: 25, dev_relevance: 30)
- Two-Stage Multi-modal Learning for Dynamic Audio-Visual Navigation: From Physical Alignment to Scene Aggregation (importance: 10, dev_relevance: 10)
- SAGE: Efficient Protected Vision-Language Inference on Smart Glasses (importance: 18, dev_relevance: 25)
- Embodied Zero-Shot Task Planning via Hardware-Aware LLM-VLM Integration for Grounded Robotic Manipulation (importance: 18, dev_relevance: 25)
- Beyond Discrimination: Evaluating Probability Calibration in Radiomics-Image Fusion for Renal Cell Carcinoma Classification (importance: 6, dev_relevance: 10)
- Beyond Gate Weights: Decision-Level Interpretation of Radiomics-Image Feature Fusion for RCC Classification (importance: 6, dev_relevance: 10)
- Next-Concept Modeling: Fast, Parallel Coarse-to-Fine Text Generation (importance: 28, dev_relevance: 30)
- Do you "see" what I "see"? A Multi-panel Visual Question Answer Dataset for Large Language Model Chart Analysis (importance: 15, dev_relevance: 20)
- Cortex: Measurement-Gated Evolution and Pruning of Coding-Agent Skills and Rules (importance: 50, dev_relevance: 80)
- Learnable Low-Rank Polynomial Sketch for Effective Linear Attention (importance: 22, dev_relevance: 30)
- CAPTOR: a vision-language foundation model for Chest-Abdomen-Pelvis Oncologic CT (importance: 8, dev_relevance: 10)
- Latent World Action Model for GUI Agents (importance: 25, dev_relevance: 35)
- Agentic Data Engineering for LLMs: System Design and a Controlled Empirical Study (importance: 32, dev_relevance: 40)
- Akule: Fast and Scalable Multi-Robot Motion Planning via Sparse Interaction Diffusion (importance: 10, dev_relevance: 15)
- The RNN Homunculus: Learning to Gate Long-Term Memory (importance: 12, dev_relevance: 15)
- GeoFix: Geometry Fixing for 4D Consistent Generation (importance: 12, dev_relevance: 10)
- RUBATO: Decoupled Clocks Allow Full-Duplex Models to Think as Fast as They Can (importance: 18, dev_relevance: 20)
- MvFFN: Multi-view Floor plan Feed-forward Network for Unposed Wide-Baseline Panorama Layout Reconstruction (importance: 8, dev_relevance: 10)
- EV-JEPA: Elastic V-JEPA for Any Device and Cross-Device Inference (importance: 15, dev_relevance: 25)
- TraceSampler: Making Sampling Errors Predictable and Correctable for Accelerated Generative Sampling (importance: 15, dev_relevance: 20)
- Natural Code Transformation Confuses Agents in Vulnerability Analysis (importance: 40, dev_relevance: 60)
- DexFLEX: Contact-Aware Foundation Controller for Command-Guided Dexterity (importance: 10, dev_relevance: 10)
- Deep learning from a control vantage point (importance: 15, dev_relevance: 20)
- RealityBench: Simulated Environments to Evaluate a Vision Language Model's Ability to Document Human Physical Work (importance: 15, dev_relevance: 15)
- Memory Without Curation: Learning Multi-Stream Latent Memory for Robot Policies (importance: 12, dev_relevance: 15)
- Coarse-Graining Recursive Expert Computation: Continuation-Relative Interfaces and Failure Modes (importance: 18, dev_relevance: 30)
- SceneForesee: Learning Scene Evolution for Spatiotemporal Reasoning (importance: 12, dev_relevance: 10)
- Target-ESS Tempered Policy Optimization (TETPO): Smooth Variance Control via Bridge Importance Sampling (importance: 22, dev_relevance: 30)
- Specification-Based Testing of Large Language Models (importance: 30, dev_relevance: 45)
- KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization (importance: 25, dev_relevance: 45)
- Jevolution: A Framework for Studying Communication and Collective Dynamics in AI Agents (importance: 10, dev_relevance: 25)
- Co-adaptive Rule Routing and Constraining for Structured Knowledge Retrieval (importance: 15, dev_relevance: 20)
- Diffusion-Based Auctions for Multi-Winner Banner Ads (importance: 8, dev_relevance: 10)
- RES-MR: Risk-Aware Reasoning for Explainable and Safe Medication Recommendation (importance: 6, dev_relevance: 10)
- Scale, Drivers, and Limits of Representational Alignment Across Large Language Models (importance: 18, dev_relevance: 20)
- What Counts as the Same Task? Executable Contracts for Robustness Evaluation of Tool-Using Agents (importance: 32, dev_relevance: 50)
- AphaVoice: Text-to-Speech Model for Aphasia Patient Simulation (importance: 8, dev_relevance: 10)
- Rethinking Activation Functions: Reducing Bias Can Improve Learning (importance: 15, dev_relevance: 20)
- Compositional Concepts Erasing: Disentangling Object-Style via Domain Manifold Unlearning (importance: 12, dev_relevance: 10)
- Who to Trust in Multi-Robot Systems: Joint Robot and Object Hypothesis Trust Inference via Neural-Symbolic Cross-Validation (importance: 8, dev_relevance: 10)
- GG-Nav: Goal-oriented Grounding Chain-of-Thought Elicits Reasoning in Navigation Foundation Models (importance: 10, dev_relevance: 10)
- A Self-Defining Operator for Distributed Applications (importance: 28, dev_relevance: 45)
- Active Neural Matter: A Finite-Window Response Framework for Physical Systems and Distributed AI Workflows (importance: 10, dev_relevance: 15)
- LLM-based User-Product Relevance Evaluation for Dynamic Product Advertising (importance: 12, dev_relevance: 15)
- LASER: Defeasible Certificate Search for Long-Range Protocol Dependencies (importance: 18, dev_relevance: 40)
- Limitations on Autoregressive Token Generation via Sequential Sampling (importance: 15, dev_relevance: 15)
- EarthAlign: Learning Language-Aligned Representations of Earth System Data (importance: 8, dev_relevance: 10)
- Prospective Forcing: Learning to Look Ahead for Autoregressive Long Video Diffusion (importance: 12, dev_relevance: 10)
- DexCast: Learning Real-World Dexterous Control from Human Videos (importance: 10, dev_relevance: 10)
- AlignAudio: A Unified Dual-Alignment Framework for Noise-Robust Speech-Conditioned Generation (importance: 30, dev_relevance: 15)
- Towards Merging-conditional Backdoor Unalignment against Large Language Models (importance: 45, dev_relevance: 45)
- Lingdan-V2: Large language models for clinically grounded reasoning in traditional Chinese medicine (importance: 25, dev_relevance: 5)
- Vela: User Behavior as Supervision Sparks LLM Intelligence (importance: 35, dev_relevance: 20)
- Integrating AIRL-Based Reward Estimation into a Control Method Using Fuzzy Inference and Policy Gradient (importance: 15, dev_relevance: 10)
- Self-improving data synthesis can scale embedding models across domains and tasks (importance: 35, dev_relevance: 35)
- PTGPO: Prefix Tree Group Policy Optimization (importance: 35, dev_relevance: 40)
- Last Secure Code Benchmark: Is the Code You Vibe Secure? (importance: 50, dev_relevance: 75)
- Prior-Fitted and Frozen Scorers for Multimodal Contrastive Learning (importance: 20, dev_relevance: 10)
- Training a Compact, Strong, and Realistic Text-to-Speech Model (importance: 30, dev_relevance: 10)
- YOUDrive: Driving Style as a Composable Direction for End-to-End Autonomous Driving (importance: 15, dev_relevance: 5)
- Depth-progressive 3D seismic velocity model building via 2D generative diffusion models (importance: 12, dev_relevance: 5)
- Seamless Deployment of Machine Learning Inference at the Far Edge: An Empirical AWS IoT Greengrass Architecture for Industrial Predictive Maintenance (importance: 15, dev_relevance: 40)
- AI-Native Product Management: A Vision (importance: 30, dev_relevance: 20)
- Cross-Lingual TTS for Indian Languages: A Benchmark and Inference-Time Duration Estimation Strategies (importance: 25, dev_relevance: 10)
- Getting Space Right: Agentic Rewards for Image Generation (importance: 25, dev_relevance: 5)
- Probabilistic Multi-Source Data Selection via Learnable Categorical Priors: Variance Reduction and Guaranteed Derandomization (importance: 15, dev_relevance: 10)
- Counterfactual Rollouts for Offline Contrastive Goal-Conditioned Reinforcement Learning (importance: 15, dev_relevance: 5)
- Humans vs. Machines: A Comparison With AI Agents in Software Reverse Engineering (importance: 45, dev_relevance: 55)
- Fitting Every Hint Is Not Enough: Premature Floods in Recurrent Algorithmic Reasoners (importance: 15, dev_relevance: 10)
- Too Late to Prune: Generative AI, the Screen, and Team Creative Decisions (importance: 30, dev_relevance: 10)
- Mixing Regularization for Average-Reward $Q$-Learning and $Q$-Planning (importance: 10, dev_relevance: 5)
- Recurrent In-Context Learning for Tabular Foundation Models (importance: 20, dev_relevance: 10)
- UltraBERT: Beyond the Limit of BERT Pre-Training (importance: 30, dev_relevance: 20)
- Toward System One Clinical Intelligence: Parameter-Efficient Fine-Tuning and Counterfactual Explainability of Decision Foundation Models (importance: 15, dev_relevance: 5)
- InCLAD: A Continual Learning Benchmark for Industrial Visual Anomaly Detection (importance: 12, dev_relevance: 5)
- Select or Rewrite: Oracle-Guided Fine-Tuning for Generative ASR Post-Editing from N-Best Hypotheses (importance: 25, dev_relevance: 10)
- Deep Learning-based Semantic Segmentation of Dental Radiographs: Comparative Evaluation of PSPNet, DeepLabv3, UNet, and UNet++ (importance: 8, dev_relevance: 3)
- GAMW: Generation-Aware Mel-Domain Watermarking for TTS Traceability (importance: 20, dev_relevance: 15)
- Have You Been Convinced? A Neuro-symbolic Argumentation Reasoning Approach to Detecting Persuasiveness in ChangeMyView Discussions (importance: 12, dev_relevance: 5)
- VLCQ: Post-training quantization for deep neural networks using variable length coding (importance: 20, dev_relevance: 30)
- TurnCost: Turns, Not Observations: Cache-Aware Loop Engineering for Coding Agents (importance: 55, dev_relevance: 80)
- scHFBiMamba: a bidirectional Mamba autoencoder for cardiac disease-state classification from single-nucleus RNA-seq data (importance: 10, dev_relevance: 5)
- ShellElastic: Scalable Elastic Pretraining with Simultaneous Maximal Updates and Hyperparameter Transfer (importance: 25, dev_relevance: 15)
- RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition (importance: 10, dev_relevance: 5)
- To Defer or Not to Defer? Online Learning with LLMs and Congested Experts (importance: 20, dev_relevance: 15)
- UDHRBench: Benchmarking Human Rights in Large Language Models (importance: 20, dev_relevance: 10)
- Exploring Jailbreak Vulnerability Across Depth-Based Dynamic Inference in Large Language Models (importance: 25, dev_relevance: 25)
- PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces (importance: 35, dev_relevance: 40)
- Framework for an AI-Ready ARPES Community Database (importance: 12, dev_relevance: 10)
- EvolvAttack: Backdooring self Evolving VLM Agents (importance: 30, dev_relevance: 25)
- Sequent-Prover: Training Agents for Formal, Checkable SMT-based Reasoning (importance: 30, dev_relevance: 30)
- LeHami: Symplectic Latent Dynamics for World Models (importance: 15, dev_relevance: 5)
- GeoBridge: Evolving Transferable Decoder Topologies for Geometric Vision (importance: 12, dev_relevance: 5)
- North or South? Diagnosing LLMs' Understanding of North Korean Language, Society, and Knowledge (importance: 25, dev_relevance: 5)
- Protein Crystallization Segmentation and Classification Using Subordinate Color Channel in Fluorescence Microscopy Images (importance: 8, dev_relevance: 3)
- Co-DSL: Human-LLM Co-design of Reusable Spatial Concepts for Layout Generation (importance: 15, dev_relevance: 10)
- HGA-Quant: Heterogeneity-Aware Quantization for AutoRegressive Visual Generation Models (importance: 12, dev_relevance: 15)
- WisdomNet: trustable machine learning toward error-free classification (importance: 10, dev_relevance: 5)
- Accuracy and Uncertainty in Machine Deception Detection (Research White Paper) (importance: 25, dev_relevance: 10)
- Anchor-and-Harvest Curriculum: A Fully Unmasked Non-LLM Curriculum for Craftax (importance: 12, dev_relevance: 5)
- ZOIF: Zeroth-Order Influence Function Estimation for LLM Data Selection Using Only Forward Passes (importance: 35, dev_relevance: 25)
- StimFlow: Synthesizing Brain Activity from Images via Stimulus-conditioned Flow Matching (importance: 10, dev_relevance: 3)
- Across Viewpoints and Time: Trajectory-Generalizable Physical Attacks on World Models and Video Understanding Models (importance: 15, dev_relevance: 5)
- CodeWalk: Generating Coding Benchmarks by Walking a Problem Graph (importance: 35, dev_relevance: 60)
- Gugak-VocalSet: A Multi-Genre Studio-Quality Dataset of Korean Traditional Singing with Sigimsae Annotations (importance: 10, dev_relevance: 3)
- Siam-ALS: An Explainable Siamese Deep Learning Framework for Early ALS Detection Using EEG and Eye-Tracking Signals (importance: 8, dev_relevance: 3)
- Adversarial Detection for Pre-trained Classifiers via Lipschitz-Margin Analysis with Certification (importance: 15, dev_relevance: 10)
- A Sovereign, Indic-First, Reasoning Model (importance: 35, dev_relevance: 15)
- In-Place Indic Tokenizer Expansion and Continual Pretraining of Nemotron-3-Nano-30B-A3B (importance: 30, dev_relevance: 25)
- E8 Lattice Key Quantization for Low-Bit KV-Cache Compression (importance: 30, dev_relevance: 40)
- Sensitivity or Criterion? Measuring and Moving When Language Models Act (importance: 25, dev_relevance: 30)
- Pessimistic Latent Task-aware Optimization for Robust Offline Meta-Reinforcement Learning (importance: 10, dev_relevance: 5)
- World Modeling through Spectral Alignment (importance: 12, dev_relevance: 5)
- BridgeCodec: Mamba Enhanced Neural Audio Codec with Schr\"odinger Bridge at Low Bitrate (importance: 15, dev_relevance: 5)
- When Messages Go Wrong: Understanding Communication Faults in LLM-Based Multi-Agent Systems (importance: 30, dev_relevance: 40)
- Security in LLM-Assisted Coding: A Systematic Literature Review (importance: 40, dev_relevance: 70)
- The Imperishable Record, Forgiveness, and the Future of Criminal Justice (importance: 10, dev_relevance: 3)
- CacheLink: Functional KV Transport for Zero-Prefill Cross-Model Context Handoff (importance: 25, dev_relevance: 35)
- REPRO: Reflective Evolutionary Prompt Optimizer for Automated LLM Red-Teaming (importance: 25, dev_relevance: 20)
- SARATHI: Decoupled Data-Free Probing and Adaptive Reconstruction for Structured LLM Pruning (importance: 25, dev_relevance: 30)
- OceanFlow: Coherent Flow Matching for Probabilistic Ocean Modeling (importance: 10, dev_relevance: 3)
- Improving the Bounds on Approximating Weighted Maximin Share for Chores (importance: 3, dev_relevance: 2)
- ELECT: Settling Identity Before Rendering for Scalable Story Visualization (importance: 15, dev_relevance: 5)
- Dynamic Conflict Clustering with Lightweight Learning and Strategic Repair for Distributed Constraint Satisfaction (importance: 8, dev_relevance: 10)
- Rethinking Human Intent-to-CAD: Parametric CAD Model Generation via Cooperative Multi-Task Alignment and Spatial-Aware Reinforcement Learning (importance: 15, dev_relevance: 20)
- TabMosaic: Unified In-Context Diffusion for Tabular Generative Tasks (importance: 15, dev_relevance: 5)
- RespFocus: Learning Temporal Evidence for Weakly Supervised Respiratory Sound Classification (importance: 8, dev_relevance: 3)
- bharatSetu: a 1M x 22 bridge for indic translation (importance: 20, dev_relevance: 10)
- What DNA Foundation Models Learn Beyond Sequence Composition (importance: 10, dev_relevance: 5)
- Uncovering the Generative Landscape of Diffusion Models: A Spectral-Guided Quality-Diversity Diffusion Framework (importance: 12, dev_relevance: 5)
- SkillMarker: A KGW-Based Watermarking Method for Agent Skill Ownership Verification (importance: 30, dev_relevance: 50)
- Hyperbolic Time Series Forecasting with Learnable Decomposition and Manifold Dynamics (importance: 10, dev_relevance: 5)
- A Decade in a Day: Synthesizing Agent-Native Interfaces from Human-Oriented Software Evolution (importance: 30, dev_relevance: 50)
- SurgiCLAW: An Interpretable Atomic-Skill Agent for Precision Robotic Micromanipulation (importance: 12, dev_relevance: 5)
- An Explainable System for Offensive Content Detection in Nastaliq Urdu: Corpus Curation, Transformer Benchmarking, and Word-Level Attribution (importance: 12, dev_relevance: 5)
- Adaptive Look, Flexible Thought: Learning How and How Long to Think in Multimodal Reasoning (importance: 20, dev_relevance: 5)
- LLM Value Functions for Credit Assignment in Agentic RL (importance: 30, dev_relevance: 25)
- Personalized LLM Recommendation with Semantic IDs: Autoregressive Beam Search vs. Multi-Token Prediction (importance: 20, dev_relevance: 10)
- What the Model Cannot See: Covert and Provable Text Invisibility in LLMs (importance: 35, dev_relevance: 35)
- Variance-Optimal State Resampling for Reinforcement Learning with Verifiable Rewards (importance: 20, dev_relevance: 15)
- Learning Multimodal Sensor Fusion with JEPA Representations for Breathing Activity and Pose Recognition (importance: 10, dev_relevance: 5)
- Make Every Expansion Count: Learning to Search Efficiently in LLM Code Repair (importance: 30, dev_relevance: 65)
- Review Rarely Beats the Stronger Model: Separating Verification from Capability Transfer in LLM Pipelines (importance: 30, dev_relevance: 25)
- Gen Through the Lens: Camera-Controllable Video Generation via Cross-Coordinate Conditioning Fusion (importance: 15, dev_relevance: 5)
- RINO: Renormalization Group Distillation with No Labels (importance: 8, dev_relevance: 3)
- Scalable Weak-to-Strong Verifier Hardening for Agent Environments (importance: 30, dev_relevance: 35)
- DiffNash: Finding Nash Equilibrium in High-Dimensional Continuous Action Spaces with Diffusion Policies (importance: 10, dev_relevance: 5)
- Hardware-Assisted Multi-Stage Ransomware Detection with Adaptive Machine Learning (importance: 12, dev_relevance: 15)
- Unfair and Unsafe: Everyday Bias Triggers Emergent Misalignment (importance: 35, dev_relevance: 15)
- Embedding Information in Policy Randomness: Distribution-Preserving Implicit Communication in MDPs (importance: 22, dev_relevance: 25)
- AutoSec-RAN: Specification-Grounded LLM Fine-Tuning for Automated Security Analysis of O-RAN (importance: 22, dev_relevance: 30)
- When Does Compensating Inherited Quantization Error Help? Mean-Square Guarantees for Layer-Wise Quantization and Their Limits (importance: 30, dev_relevance: 45)
- Of Errors and Echoes: Responsibility-Aware Credit Assignment for Tool-Use Agents (importance: 35, dev_relevance: 55)
- OT-FlowRec: Optimal Transport-Guided Multi-Bridge Flow for Multimodal Sequential Recommendation (importance: 15, dev_relevance: 25)
- When Agents Talk, Secrets Travel: Topology-Driven Leakage and Recipient-Conditioned Sanitization in Multi-Agent LLM Systems (importance: 40, dev_relevance: 55)
- Autoregressive Decoding as a Martingale: Characterizing the Stochastic Dynamics of LLM Generation (importance: 25, dev_relevance: 30)
- AI Representations of Mental Health Stigma: Quantifying the Gap Between AI Estimations and Human Perceptions Across Gender Groups Using GPT-Based LLMs (importance: 20, dev_relevance: 10)
- Adaptive Compression and Targeted Perturbation: A Unified Framework for Generalized Audio Deepfake Detection (importance: 20, dev_relevance: 15)
- When Agent Evaluation Becomes Unreliable: An Empirical Study of Oracle Failures and Configuration-Dependent Measurement in an LLM Agent Benchmark (importance: 40, dev_relevance: 70)
- Dream2Discern: Learning from Hard Latent Variants for Runtime Failure Prediction in Multi-Agent Systems (importance: 30, dev_relevance: 55)
- Where Polanyi Meets Galois: Tacit Knowing as an Algebro-Deterministic Operation, Toward Re-rooting AI-Generated Science (importance: 10, dev_relevance: 15)
- Towards a Benchmark for Fire Safety Code Interpretation in the Hong Kong AEC Industry (importance: 20, dev_relevance: 10)
- PID-Gated Continual Prompt Adaptation Under Noisy Feedback (importance: 30, dev_relevance: 60)
- Query Your Way Out: Benchmarking Multi-Turn Knowledge Reasoning through Generalized Puzzle Solving (importance: 20, dev_relevance: 25)
- Omni-Modal LLMs Know More Than They Say (importance: 30, dev_relevance: 20)
- Unravel: Recovering an Interactive 3D Scene from a Single Image (importance: 15, dev_relevance: 15)
- WildTrace: Evaluating Training Data Detection in Realistic Large-Scale LLM Post-Training (importance: 25, dev_relevance: 25)
- Fisher information and the geometry of memorization in neural networks (importance: 15, dev_relevance: 10)
- EventLens: Event-Structure Reinforcement Learning for Video Understanding (importance: 20, dev_relevance: 10)
- MP-MNCA: Morphology-Prior Masked Neighbor Cross-Attention for Efficient Spatial Domain Identification (importance: 10, dev_relevance: 10)
- SafeDSE: Joint Exploration of Safety Architecture Patterns, Mapping, and Scheduling for Dataflow Applications With Safety Arguments (Extended Version) (importance: 15, dev_relevance: 40)
- Look Back to Move Forward: Learning to Revisit for Agentic Compositional Image Generation (importance: 20, dev_relevance: 15)
- AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms (importance: 25, dev_relevance: 30)
- Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling (importance: 10, dev_relevance: 10)
- Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation (importance: 25, dev_relevance: 40)
- A Survey of Flow Matching in Reinforcement Learning (importance: 25, dev_relevance: 15)
- SLDBench v2: Active Scaling Law Discovery (importance: 20, dev_relevance: 20)
- BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction (importance: 10, dev_relevance: 10)
- Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories (importance: 15, dev_relevance: 5)
- BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates (importance: 15, dev_relevance: 10)
- WorldRoPE: Probabilistic World-Centric Rotary Embedding for Multi-View Attention (importance: 15, dev_relevance: 15)
- Rescuing Points Trapped Inside Incorrect Decision Regions: A target for Evaluating and Improving Forward-Forward Learning (importance: 15, dev_relevance: 10)
- DWI Connectome-based Sex Classification in Young using Supervised Variational Autoencoder (importance: 10, dev_relevance: 5)
- LoopQuant: Accurate and Efficient Post-Training Quantization for Looped Transformers (importance: 20, dev_relevance: 35)
- Estimating the Causal Impact of Social Navigation Signals in Corridor Scenarios (importance: 15, dev_relevance: 10)
- ARMED: Adaptive Retrieval-Augmented Medication Recommendation (importance: 15, dev_relevance: 10)
- VisExMEM: Visual Evidence-Based Explainable Multimodal Evaluation Metric (importance: 15, dev_relevance: 15)
- CryptoFlowBench: Benchmarking Multi-Hop Reasoning on Blockchains for Cryptocurrency Fund Tracing (importance: 15, dev_relevance: 10)
- StructZO: Structured Zeroth-Order Optimization for Memory-Efficient LLM Fine-Tuning (importance: 25, dev_relevance: 45)
- PATMT2: Two-stage domain adaptation for multilingual patent translation (importance: 12, dev_relevance: 20)
- SLEM: Spectrum-Aware Latent Expert Merging (importance: 20, dev_relevance: 25)
- Growing and Transferring Pathways for LLM Reinforcement Learning (importance: 20, dev_relevance: 15)
- Drafting Agents, Not Just Tokens: Speculative Recursive Self-Improvement for Agentic Systems (importance: 35, dev_relevance: 55)
- From Caricature to Collective Artifact: How Persona Construction Drives Polarization in Generative Social Simulations (importance: 15, dev_relevance: 5)
- pi-Net: Empirical Latent Representations for Differentiable Physics Heads in Data-Scarce Systems (importance: 10, dev_relevance: 15)
- PatTransBench: an ipc-stratified, multi-dimensional benchmark for patent machine translation evaluation (importance: 12, dev_relevance: 15)
- Heuristics are Not Enough: Calibrated and Corrected Caching for Video Diffusion Acceleration (importance: 20, dev_relevance: 30)
- VLA-GAUGE: Systematically Evaluating Generalization and Robustness in Vision-Language-Action Models (importance: 15, dev_relevance: 10)
- Towards Modality-Agnostic Seizure Classification: A Unified Deep Learning Framework for Heterogeneous EEG and Intracranial EEG Signals (importance: 10, dev_relevance: 5)
- MixLID: Token-Level Language Identification in Mixed-Language Text (importance: 15, dev_relevance: 25)
- Answer It or Not? Towards Prompt-Adapted Adversarial Images on Vision Language Models (importance: 15, dev_relevance: 10)
- Integrating Feature Understandability into Feature Selection for Human-Centred Tabular Explanations (importance: 10, dev_relevance: 10)
- AeroChem: Closed-loop Physics-Informed State Space Modeling for Long-term Chemically-Reactive Air Quality Forecasting--Extended Version (importance: 10, dev_relevance: 5)
- REEL: Learning to Retrieve, Evaluate, and Evolve Visual Memory for Any Long-Horizon Multi-Shot Video Generation (importance: 20, dev_relevance: 15)
- SuPER-Mamba: Surprisal Prioritization for Extended-Context Routing in Mamba (importance: 15, dev_relevance: 25)
- Learning Incompressible 3D Flows from Sparse Samples: A Set Transformer Approach to Divergence-Free Kernel Fitting (importance: 10, dev_relevance: 10)
- Feature-Context Consistency: Unsupervised Adversarial Detection with Drift Stability on Attributed Graphs (importance: 12, dev_relevance: 10)
- Audit Before You Distill: A Controlled Workflow for Reasoning Transfer into Compact Medical Language Models (importance: 20, dev_relevance: 35)
- VAttnPO: Visual Attention-Guided Policy Optimization (importance: 15, dev_relevance: 30)
- Before Attention Sinks Form: Hidden-State Geometry Reveals and Modulates Their Formation (importance: 25, dev_relevance: 35)
- Relational Learning as World Modeling (importance: 15, dev_relevance: 25)
- PrivacyGuard: Source-Linked Sanitization and Locally Resolvable PII References for Cloud GUI Agents (importance: 30, dev_relevance: 50)
- InfoVis: Unified Slide Image Generation and Editing (importance: 15, dev_relevance: 15)
- Logit Leakage through Floating-Point Bias Addition (importance: 35, dev_relevance: 50)
- StableJudge: A Lightweight Adversarial Defense Framework for LLM-as-a-Judge (importance: 20, dev_relevance: 40)
- Growing to See: Efficient Pretraining of Self-Supervised Vision Transformers (importance: 15, dev_relevance: 20)
- DTA-GT: Direction- and Topology-Aware Graph Transformer for Neural Network Representation Learning (importance: 10, dev_relevance: 10)
- Unsupervised Data Drift Monitoring for Deployed Deepfake Speech Detectors (importance: 15, dev_relevance: 25)
- A Scaled-Up Empirical Study of Syntactic Language Models (importance: 15, dev_relevance: 10)
- Few-for-Many Best-Arm Identification in Multi-Objective Bandits (importance: 10, dev_relevance: 10)
- Towards Cost-Efficient Video Dataset Construction with Large Language Models (importance: 15, dev_relevance: 25)
- Davinci-Tool: Scaling Agentic Tool-Use Supervision with Verifiable Executable Environments (importance: 35, dev_relevance: 50)
- Comparing GenAI to Machine Learning and Manual Review for eDiscovery (importance: 25, dev_relevance: 15)
- KNN Implementation Details Can Dramatically Change Performance: An Example from Cover Trees (importance: 25, dev_relevance: 75)
- UA-Bench: Benchmarking Large Audio-Language Models on Underwater Acoustic Recognition (importance: 12, dev_relevance: 10)
- Social Priors: Descriptions of AI Users Shape How Language Models Trust and Cooperate (importance: 25, dev_relevance: 10)
- SarcasTraps: Effect of Sarcasm on Turkish Hate Speech Detection (importance: 8, dev_relevance: 5)
- Proactive Physical AI through Continuous Observation and Timely Intervention (importance: 20, dev_relevance: 15)
- The Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors (importance: 15, dev_relevance: 15)
- WeGR: Unified Generative Retrieval for Industrial Search Systems in WeChat (importance: 15, dev_relevance: 20)
- From Suboptimal to Optimal: One-Step Latent Flow Policies for Offline Reinforcement Learning (importance: 15, dev_relevance: 15)
- AUGUstus: Autonomous Driving VLA Model with Unified Generation and Understanding (importance: 15, dev_relevance: 10)
- TabGift: A Context-Efficient Foundation Model for Tabular Data Generation (importance: 15, dev_relevance: 15)
- Governing Copyright Responsibly in the Era of Generative AI (importance: 25, dev_relevance: 15)
- MAGNET: Manifold-Aware Graph Diffusion Network for Connectome Generation (importance: 10, dev_relevance: 5)
- ProbabilityBench: Model Evaluation and Calibration with Exact Ground-Truth Uncertainty (importance: 15, dev_relevance: 25)
- CWE-bench: Refocusing Agentic Security from Exploitation to Defense (importance: 45, dev_relevance: 70)
- SPLICE: Sparse Physics Library Identification and Compositional Expansion (importance: 12, dev_relevance: 15)
- Gradient Routing Localizes and Removes Unintended Behaviors in RL (importance: 30, dev_relevance: 35)
- Superintelligence and the Triple Risk of an AI Race (importance: 25, dev_relevance: 5)
- Interactive Navigation Beyond Reachability: Benchmark and Agentic Framework (importance: 15, dev_relevance: 15)
- What the Machine Invented: Mining New Mathematical Instruments from OpenAI's October 2026 Release of AI-Generated Mathematics (importance: 45, dev_relevance: 10)
- AIPOCH Open-Science: An Open-Source AI Research Workbench with Reproducible .science Packages (importance: 25, dev_relevance: 55)
- Calibrated Typed Decisions for Entity, Relation, and Event Extraction (importance: 15, dev_relevance: 25)
- D-World: An Explicit World-Language-Action Model with Recursive Self-Improvement under Dual Control (importance: 15, dev_relevance: 15)
- D2: High-Capability Full-Duplex Spoken Language Models from Half-Duplex Backbones (importance: 20, dev_relevance: 10)
- Who is the 'I' in AI? Establishing Decentralized Identity and Accountability for Autonomous Agents (importance: 25, dev_relevance: 35)
- SonicWorld: Source-Centric Latent Dynamics for Spatial Audio Reasoning (importance: 10, dev_relevance: 5)
- Force-Modulated Dexterity: Unified Compliance for Contact-Rich Manipulation (importance: 12, dev_relevance: 10)
- Generative Flow on Symbolic Abstractions for Embodiment Capability Discovery (importance: 25, dev_relevance: 40)
- FreeSolo: Linear-time Sparse Attention with Zero-Shot Length Extrapolation (importance: 40, dev_relevance: 55)
- Deleting Code Without Losing Knowledge: Regenerative Software for Python Libraries (importance: 35, dev_relevance: 70)
- Compute-optimal data scaling for neural surrogates via multi-fidelity training (importance: 25, dev_relevance: 40)
- From Conceptual EneMonIE to an Executable Backend: A Modular Architecture for Energy Information Extractionand Grounded Question Answering (importance: 15, dev_relevance: 65)
- Measuring Learned Equivariance: An Information Theoretic Approach (importance: 25, dev_relevance: 45)
- Phase-Conditioned Affordance Fields for SE(3)-Equivariant Visuomotor Policies (importance: 22, dev_relevance: 35)
- StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions (importance: 35, dev_relevance: 50)
- Joint Embedding with Dynamic Integration network (JEDI-net) (importance: 15, dev_relevance: 40)
- Thinking Outside the Bots: Human Ideas in AI's Semantic Blind Spots Are Not Inferior (importance: 25, dev_relevance: 20)
- On the Statistical Mechanics of Attention (importance: 30, dev_relevance: 55)
- MID: Mask-Image Distributional Divergence for Evaluating Medical Image Segmentation (importance: 20, dev_relevance: 35)
- claimcheck: Measuring and Improving Specification Faithfulness (importance: 35, dev_relevance: 75)
- HASTE: Hierarchical and Attribute-aware Supervision for Generative Recommendation (importance: 25, dev_relevance: 40)
- Mitigating Black-Box Score-Based Query Attacks via Gradient Confusion (importance: 25, dev_relevance: 45)
- Cross-Market Meta-Learning for Electricity Price Forecasting (importance: 15, dev_relevance: 25)
- Learning to Detect Shifts in Unstructured Data Streams via Monitoring-by-Betting (importance: 30, dev_relevance: 50)
- Event-Centric Earthquake Monitoring with a Multi-Station Transformer (importance: 20, dev_relevance: 25)
- Cross-View Time-Frequency Diffusion for Extreme-Aware Time-Series Generation (importance: 20, dev_relevance: 40)
- Bottlenecked semantic compression by representation reconstruction (importance: 25, dev_relevance: 50)
- Why Agentic Workflows Fail: A Taxonomy Derived from Literature and Empirical Data (importance: 45, dev_relevance: 70)
- ClearNav: Perception-Aligned Quadrotor Flight Using Differential-Physics (importance: 20, dev_relevance: 30)
- On Making $\mathrm{SE}(2)$-Invariant Networks Optimal (importance: 20, dev_relevance: 40)
- Reinforcement Learning In Partially Verifiable Environments for Colon Cancer Treatment Planning (importance: 15, dev_relevance: 20)
- World-T3: Test-Time Training as Persistent Memory for Real-Time Video World Models (importance: 25, dev_relevance: 35)
- Know-Thy-Not: Training-Free Targeted Negation in Dual-Encoder Vision-Language Models (importance: 20, dev_relevance: 35)
- The belief-action gap in language model agents (importance: 35, dev_relevance: 55)
- Compositional RL: Evaluating and Training LLMs for Collaborative Reasoning (importance: 35, dev_relevance: 55)
- Assessing the Performance of AI Models in Misinformation Detection Using Predefined Social Noise Constructs (importance: 15, dev_relevance: 25)
- VeriPIE: Checkable Query Plans and Evidence-Guided Selection for Text-to-SQL (importance: 35, dev_relevance: 60)
- Foundation Models for Event Prediction Across Heterogeneous Event Logs (importance: 20, dev_relevance: 40)
- YUBI-Real2Sim: 3DGS Scene Verification for Reliable Policy Evaluation (importance: 20, dev_relevance: 35)
- Enhanced Detective: A Dual-Stage Proactive Defense Against Internal Poisoning Attacks in Financial Fraud Detection Using Federated Learning (importance: 20, dev_relevance: 40)
- Box-WAM: Box Trajectories as Compact State Representations for World Action Models (importance: 20, dev_relevance: 35)
- Towards Green AI: Evaluating Energy Efficiency for Quantized LLM Inference (importance: 35, dev_relevance: 65)
- Convex-Analytic View of Query Selection for Safety Verification of Markov Decision Processes (importance: 15, dev_relevance: 40)
- Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents (importance: 40, dev_relevance: 65)
- FlashVSR+: Efficient Video Super-Resolution via Hybrid Attention and Reward-Yield Distribution Matching (importance: 25, dev_relevance: 35)
- Benchmarking Self-Supervised 3D Medical Vision Foundation Models and Transferability Estimation (importance: 20, dev_relevance: 25)
- Dissociating Answerability from Extraction in LLM Question Answering over Radiology Reports (importance: 20, dev_relevance: 30)
- Chasing Label Shifters: A Change-Aware Framework for Dynamic Graph Node Classification (importance: 15, dev_relevance: 35)
- Taming Softmax at the Limits of Muon: A Unified Approach through Hilbert Geometry (importance: 30, dev_relevance: 55)
- Revisiting the 16x Pretraining Compute Gap between Diffusion and Autoregressive Language Models (importance: 40, dev_relevance: 55)
- Tanawpamana: Benchmarking Vision-Language Models on Philippine Cultural Heritage Sites (importance: 15, dev_relevance: 25)
- Morphology-aware deep learning predicts colorectal cancer survival from whole-slide images (importance: 15, dev_relevance: 15)
- RELAY: Selective Self-Distillation with Adaptive Hand-off for Rubric-Based Reinforcement Learning (importance: 30, dev_relevance: 50)
- 4LC: 4D Latent Cache for Efficient Conditioning in Dynamic World Modeling (importance: 20, dev_relevance: 35)
- The Tokens Strike Back: Safety-Aware Denoising with Directional Embedding-based Revision (importance: 20, dev_relevance: 40)
- DeltaTTG: Training Multi-Turn Agents with a Token-to-Go Progress Meter (importance: 30, dev_relevance: 55)
- PrT-Flow: Probability-Tractable Flow Matching for Reinforcement Learning (importance: 20, dev_relevance: 40)
- Dynamic K-Head: Difficulty-Supervised Budget Selection for Continuous Latent Reasoning (importance: 25, dev_relevance: 45)
- LatentRiskBench: Measuring State-Dependent Risks in Enterprise Agents (importance: 30, dev_relevance: 50)
- HelpMeOut: A Neuropsychology-Grounded Benchmark for Error-Aware Monitoring of Household Task Execution (importance: 15, dev_relevance: 25)
- zPocket: Better Pocket Finding with Frozen Co-Folding Features (importance: 15, dev_relevance: 15)
- LeanPRM: Lean-Grounded Process Supervision for Natural Language Proof Verification (importance: 25, dev_relevance: 40)
- From Token Correction to Claim Admission: Training-Free Hallucination Mitigation for Large Vision-Language Models (importance: 20, dev_relevance: 35)
- Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning (importance: 20, dev_relevance: 40)
- TSRL: Trigger-based Supervised Reinforcement Learning for Continual Agent Training (importance: 25, dev_relevance: 45)
- SkillScreen: High-Recall Screening of Malicious Agent Skills with Decision Models (importance: 40, dev_relevance: 70)
- CoCA: Criterion-Aware Credit Assignment for Rubric-Based Reinforcement Learning (importance: 25, dev_relevance: 45)
- Explore with Memory, Learn without It: In-Context Exploration and Self-Distillation for LLM Reasoning (importance: 25, dev_relevance: 45)
- Train for Many, Update with One: Improving Multi-Sample Reasoning via Fenchel-Bregman Posterior Learning (importance: 20, dev_relevance: 40)
- HiSTOR: Structured Treatment-outcome Representation Learning for High-Dimensional Longitudinal Data (importance: 15, dev_relevance: 30)
- Auditing Failure Attribution: Label Predictability and Judge Behavior (importance: 30, dev_relevance: 60)
- I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation (importance: 20, dev_relevance: 35)
- Degrees of Freedom in Multi-Agent Coordination: Between Centralized and Decentralized Regimes (importance: 35, dev_relevance: 55)
- Gaussian Process Bandits for AI System Optimization with Selectively Verified Feedback (importance: 25, dev_relevance: 50)
- Efficiently Learning To Reason or Not to Reason: Root-token Policy Optimization for Adaptive Thinking (importance: 30, dev_relevance: 45)
- Posterior-Averaged Bellman Flow Matching for Distributional Critics (importance: 20, dev_relevance: 35)
- GIFT-Eval-CTX: A Benchmark For Contextual Forecasting Model Evaluation (importance: 20, dev_relevance: 40)
- Learning from fracture: Teaching language models to design through physical experience (importance: 20, dev_relevance: 30)
- Same Attack, Different Judge: A First Controlled Adversarial Comparison of Jev, LLM Judges, and Purpose-Built Safety Classifiers (importance: 30, dev_relevance: 55)
- Checking the Checkers: A Synthesis of Verifier Noise and Robustness in RLVR (importance: 30, dev_relevance: 50)
- QK-Edit: Revisiting Attention-based Injection in MM-DiT for Image and Video Editing (importance: 20, dev_relevance: 30)
- Diff Diffusion: Predicting Edit Paths with Diffusion Models for Efficient Code Editing (importance: 40, dev_relevance: 75)
- Agentic-MatriBO: Capability-Compiled Agentic Bayesian Optimization for Materials and Chemical Design (importance: 20, dev_relevance: 35)
- A Tutorial on Quantum Amplitude Embedding with Data Re-Uploading (importance: 15, dev_relevance: 30)
- ProtoArena: Can AI Agents Discover and Exploit Real-World Cross-Component Protocol Vulnerabilities? (importance: 35, dev_relevance: 55)
- FASHION-RQ: Mining Hierarchical Semantic Structure for Automated Evaluation of Fashion Retrieval (importance: 15, dev_relevance: 25)
- MMTA: Benchmarking Multimodal Temporal Analysis with Time Series, Text, and Vision (importance: 20, dev_relevance: 30)
- TStruct: Learning Shared Temporal Structures for Long-Term Time-Series Forecasting (importance: 15, dev_relevance: 30)
- RT-J: Large-Scale Pretraining of Relational Transformers for Context-Efficient Predictions (importance: 20, dev_relevance: 40)
- Cross-Lingual Transfer of Evil Vectors Is Directionally Asymmetric (importance: 25, dev_relevance: 35)
- Interaction-as-a-Judge (importance: 40, dev_relevance: 70)
- Evaluating the Causal Faithfulness of SHAP as a Candidate Attribution Method for SAE Features (importance: 20, dev_relevance: 35)
- Visual Dependence Does Not Imply Answerability: A Systematic Analysis of Answerability Signals in Frozen InstructBLIP (importance: 15, dev_relevance: 30)
- DriveWLA: Unifying Future Dynamics and Scene Semantics through World-Language-Action Modeling for Autonomous Driving (importance: 20, dev_relevance: 30)
- SpecCloak: Probing Harmful Code Generation through Engineering Specifications (importance: 25, dev_relevance: 40)
- LinMorph: Learning to Morph Weights for Efficient Video DiT Linearization (importance: 15, dev_relevance: 30)
- CineMem: Attention-Aligned Fine-grained Memory Retrieval for Multi-Shot Long Video Generation (importance: 15, dev_relevance: 25)
- GLST: Global-Local Spatio-Temporal Memory Decomposition Framework for Online Chained-Goal Navigation (importance: 15, dev_relevance: 25)
- MERIT: Multimodal Evidence Reasoning over Image Tables (importance: 20, dev_relevance: 35)
- AI-Enhanced Hyperspectral Imaging for Intraoperative Detection of Low-Grade Glioma: A Clinical Study (importance: 15, dev_relevance: 15)
- HaloBlock: Faster and Better Decoding for Diffusion Large Language Models (importance: 30, dev_relevance: 50)
- FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation (importance: 15, dev_relevance: 15)
- Virtual Head Attention (importance: 35, dev_relevance: 55)
- Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling (importance: 15, dev_relevance: 25)
- BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation (importance: 35, dev_relevance: 60)
- Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models (importance: 30, dev_relevance: 40)
- Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks (importance: 15, dev_relevance: 25)
- Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models (importance: 30, dev_relevance: 35)
- SWEPatchTest: Evaluating Software Tests Beyond Reproduction (importance: 40, dev_relevance: 65)
- Cost-Efficient Privacy Enforcement for Email-Generating LLM Agents under Strategic Information Requests (importance: 30, dev_relevance: 40)
- Correcting or Continuing? What On-Policy Distillation Learns from Student Errors (importance: 30, dev_relevance: 25)
- LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems (importance: 35, dev_relevance: 50)
- ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression (importance: 25, dev_relevance: 40)
- Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On? (importance: 15, dev_relevance: 10)
- Scaling Vision Transformers Without Training: Prune-and-Merge Token Compression (importance: 20, dev_relevance: 20)
- High-fidelity Three-Dimensional Flow Field Reconstruction Driven by Turbulent Fluid Dynamics (importance: 10, dev_relevance: 5)
- inLoRA-MINT: Internal Representations for Auditing Training Data in Domain-adapted LLMs (importance: 25, dev_relevance: 35)
- From Policy Evaluation to Off-Policy Control: Provable Emergence of In-Context Q-Learning (importance: 20, dev_relevance: 10)
- Uncertainty Evaluation: Bridging Machine Learning and Measurement Science (importance: 15, dev_relevance: 15)
- APT: Feedback-Driven Agentic Post-Training for Text-to-Image Generation (importance: 25, dev_relevance: 15)
- A CaLLM Judge is a Better Judge: Mitigating Bias in LLM-as-a-judge (importance: 35, dev_relevance: 35)
- When Missingness Follows the Graph: Beyond Feature Propagation for Missing Features (importance: 15, dev_relevance: 15)
- Certifying Almost-Sure Reachability in Unknown MDPs (importance: 15, dev_relevance: 15)
- H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression (importance: 20, dev_relevance: 20)
- When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions (importance: 15, dev_relevance: 10)
- Fan Out or Relay? Ownership-Aware Routing for LLM Agents Querying Siloed Tables (importance: 35, dev_relevance: 55)
- Claim Framing as a Driver of LLM Sycophancy (importance: 30, dev_relevance: 30)
- LoOPSD: On-Policy Self-Distillation with Looped Teachers (importance: 25, dev_relevance: 30)
- Lifting Finite-Sample Gaussian Approximation of Wide Neural Networks to Functional Convergence (importance: 10, dev_relevance: 10)
- Looped Reasoning Finishes Earlier Than You Think (importance: 20, dev_relevance: 15)
- A Few More Tokens Ahead: Reinforcing Parallelism in Diffusion LLMs (importance: 25, dev_relevance: 25)
- How Reliable Are Large Language Models for Soil Organic Carbon Assessment? An Australian Benchmark (importance: 15, dev_relevance: 10)
- Reset Early, Reset Often, Eliminate Models (importance: 15, dev_relevance: 25)
- TextualRL: Textual Reinforcement Learning for Context Management (importance: 35, dev_relevance: 60)
- A Controlled Evaluation of Curvature Inspired Scalar Feature Scaling for Domain Generalization (importance: 10, dev_relevance: 10)
- Holding Agents Accountable with Verifiable Agent Insurance (importance: 35, dev_relevance: 40)
- Heading Hierarchy Reconstruction for Complex and Long Documents with Level-Wise Multimodal Memory (importance: 25, dev_relevance: 45)
- LiFUSE: Frequency-Spatial Unified Self-Supervised Pretraining for Liver Disease Analysis on Non-contrast CT (importance: 10, dev_relevance: 5)
- FALLNEST: Dual-Domain Dataset Condensation of Imbalanced IMU Data for Wearable Fall Detection (importance: 10, dev_relevance: 5)
- Encrypted Skills Can Become New Security Exploits If AI Agents Know How to Decipher (importance: 45, dev_relevance: 65)
- JevGPT: Thinking Fast and Slow in Autonomous Research (importance: 25, dev_relevance: 35)
- Which Optimizer Leaks Most? Optimizer Implicit Bias as a Training-Data Privacy Controller (importance: 20, dev_relevance: 25)
- AffectWorld: Benchmarking Emotion-Driven Future World Prediction (importance: 10, dev_relevance: 5)
- SimInter: Evidence-Grounded User Simulation with Interview-Based Profile Enrichment for Conversational Recommender Systems (importance: 15, dev_relevance: 10)
- ShoeFit: A New Dataset and Dual-image-stream DiT Framework for Virtual Footwear Try-On (importance: 10, dev_relevance: 5)
- Lightweight but Linguistically Fair? Quantisation Robustness across Bavarian, German, and English (importance: 20, dev_relevance: 30)
- Let the Mind Conceive, Let the Hand Achieve: Following Multimodal Instructions for Long-Horizon Robotic Manipulation (importance: 15, dev_relevance: 25)
- Urbex: Agentic Spatial Grounding for City-Scale 3D Scenes (importance: 15, dev_relevance: 10)
- PRISM: Calibrating Black-Box Judges with Counterfactual Probes (importance: 30, dev_relevance: 40)
- Graph-free Retrieval with Graph-Modified Embeddings (importance: 25, dev_relevance: 40)
- Not All Drift Is a Mechanism: Tracing Prompt Adaptation in LLMs (importance: 20, dev_relevance: 25)
- Do Factuality Methods Make LLMs More Factual? Reproducibility and Fair Comparison (importance: 25, dev_relevance: 25)
- Resource-Efficient 3D Shape and Orientation Estimation of Road Users using Roadside RGB Camera and 3D Radar (importance: 10, dev_relevance: 10)
- Agaram 2.5B: The First Tamil-First Language Model Trained from Scratch with Tool Calling. A Playbook for Under-Represented Languages (importance: 35, dev_relevance: 40)
- FedGraph: Defending Federated Large Language Model Fine-Tuning Against Backdoor Attacks via Graph-Based Aggregation (importance: 25, dev_relevance: 30)
- How Well Do LLMs Perform in Peer Review? Evidence from Finance and Accounting (importance: 20, dev_relevance: 15)
- STAMP as a topology-aware molecular language for fragment-level learning and editing (importance: 15, dev_relevance: 10)
- Learning Perturbation-Specific Representations through Conditional Energy Prior for Cellular Response Prediction (importance: 10, dev_relevance: 5)
- Readout Scaling Shapes Sparse Feature Recovery (importance: 20, dev_relevance: 15)
- MemoryVLN: From Ego-Aware Memory Representation to Cross-Context Memory Reuse in Vision-Language Navigation (importance: 15, dev_relevance: 10)
- Task-evoked eye movements as candidate digital biomarkers of depression severity in adolescents and young adults (importance: 10, dev_relevance: 5)
- Aligning Faithfulness for Theory-Scale Autoformalization (importance: 20, dev_relevance: 20)
- EgoOmniWorld: Unifying Egocentric Understanding via Ego-Action-World Interactions (importance: 10, dev_relevance: 5)
- A Global View of How LLMs Value Human Time (importance: 20, dev_relevance: 10)
- The Geometry of Language Model Logits: Generic Structure and Learned Departures (importance: 25, dev_relevance: 25)
- APTree: Extended Behavior Trees for Adaptive and Scalable Automated Robotic Task Planning in Construction (importance: 10, dev_relevance: 15)
- On Symmetry for Characterizing Agent Heterogeneity in Multi-Agent Systems (importance: 10, dev_relevance: 10)
- The Many Morals of AI: The cautionary bias of North American news narratives in global perspective (importance: 10, dev_relevance: 5)
- When More Objectives Make Identification Easier: The Role of Preferences and Feedback (importance: 10, dev_relevance: 10)
- GeoMER: A Joint Euclidean-Hyperbolic Geometry Framework for Multimodal Emotion Recognition (importance: 10, dev_relevance: 5)
- ACCRUE: Continual Off-Policy Reinforcement Learning for Robot Foundation Models (importance: 20, dev_relevance: 15)
- PhysicsSliders: Learning Physical Controls for Audio Foley Generation (importance: 15, dev_relevance: 10)
- Rethinking Projector Training in Multimodal LLMs (importance: 25, dev_relevance: 20)
- LLM2R: From Large Language Models to 3D Reassemblers (importance: 10, dev_relevance: 5)
- DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization (importance: 25, dev_relevance: 25)
- Jev-ToM: Evaluating Theory of Mind in Typed Decision Models (importance: 15, dev_relevance: 15)
- Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications (importance: 25, dev_relevance: 35)
- Does self-distillation improve reasoning? (importance: 25, dev_relevance: 25)
- Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve (importance: 35, dev_relevance: 60)
- Falsification-proof non-market allocation mechanisms (importance: 10, dev_relevance: 5)
- Geometric Calibration of Supervised Fine-Tuning (importance: 20, dev_relevance: 20)
- DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models (importance: 15, dev_relevance: 25)
- ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction (importance: 10, dev_relevance: 10)
- Tiny Recursive Message Passing (importance: 15, dev_relevance: 10)
- AMemOps: Evaluating Memory Updates in Long Audio Streams (importance: 15, dev_relevance: 15)
- Kwai Summary Attention V2 Technical Report (importance: 25, dev_relevance: 30)
- Known-Voice Cloning and the Limits of Personality Transfer for Character Design (importance: 15, dev_relevance: 10)
- SMI: Ontology-Grounded Semantic IDs for Hierarchy-Aware Medical Concept Representation (importance: 10, dev_relevance: 15)
- From Relation-Specific Heads to Shared Interventions: Contextual Entrainment in Language Models (importance: 20, dev_relevance: 20)
- Model-Based Online Decision Making via Generative Trajectory Planning (importance: 15, dev_relevance: 10)
- ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning (importance: 30, dev_relevance: 35)
- RadarGSOcc: Doppler-Aware 3D Semantic Occupancy Prediction Using 4D Radar Tensor with Geometric-Semantic Feature Extraction (importance: 10, dev_relevance: 5)
- Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning (importance: 20, dev_relevance: 15)
- GeoReader: Learning Latent Geometry for Spatial Reasoning in Vision-Language Models (importance: 15, dev_relevance: 10)
- UniW(A)M: A Unified Mixture-of-Experts Framework for World and World-Action Models (importance: 15, dev_relevance: 10)
- Steertuning Policies from a Few Human Videos (importance: 15, dev_relevance: 10)
- A Rate Transfer Method: Convergence Rates in Stochastic Gradient Descent via Stochastic Calculus (importance: 15, dev_relevance: 15)
- SolBehave: Evaluating the Behavioural Correctness of LLM-Generated Smart Contracts (importance: 30, dev_relevance: 50)
- PreciseCoder: Faithful and Generalizable Debugging Beyond Unit-Test Rewards (importance: 30, dev_relevance: 55)
- Where Do LLM Ensemble Gains Come From? Decomposing Candidate Value, Diversity, and Aggregation (importance: 20, dev_relevance: 25)
- Bytes to Threats: Byte-Sequence Transformer For Malware Detection (importance: 15, dev_relevance: 25)
- SEAR: Self-Evolving Agentic Robotics (importance: 20, dev_relevance: 35)
- Optimizing Team Behavior via Extremum-seeking Control in Multi-agent Reinforcement Learning (importance: 10, dev_relevance: 10)
- Back-Translation Still Matters: Leveraging Monolingual Data for Low-Resource Machine Translation with LLMs (importance: 20, dev_relevance: 20)
- RoboSimStudio: An Interactive Toolkit for Human and Agent Rapid Real2Sim of Robot Manipulation Tasks (importance: 20, dev_relevance: 40)
- Fast Rates for Offline Q-Learning with KL Regularization (importance: 10, dev_relevance: 10)
- LIBERO-RIG: A Hardware-Setup Robustness Benchmark for Vision-Language-Action Models (importance: 15, dev_relevance: 10)
- Two Clocks: Why Larger Rollout Batches Learn Slower but Fail on Schedule in Reinforcement Learning with Verifiable Rewards (importance: 35, dev_relevance: 35)
- Hybrid Reasoning Agents (importance: 30, dev_relevance: 30)
- DoubtLess: Training-Free Contrastive Decoding for Efficient Reasoning (importance: 30, dev_relevance: 30)
- Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards (importance: 35, dev_relevance: 40)
- What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study (importance: 25, dev_relevance: 10)
- ChunkAct: Action Chunking for Agentic Robotics (importance: 25, dev_relevance: 25)
- Decidable Reasoning About Time in Finite-Domain Situation Calculus Theories (importance: 10, dev_relevance: 10)
- Identification of Entailment and Contradiction Relations for Natural Language Argumentation: A Neurosymbolic Approach (importance: 8, dev_relevance: 5)
- Overcoming a Learning Bottleneck in Impartial Games: A Circuit Complexity Approach (importance: 10, dev_relevance: 5)
- Robustness and Cybersecurity in the EU Artificial Intelligence Act (importance: 20, dev_relevance: 15)
- Physics: Benchmarking Foundation Models on University-Level Physics Problem Solving (importance: 20, dev_relevance: 10)
- LLM-BabyBench: Can Language Models Plan in Worlds They Can Simulate? (importance: 15, dev_relevance: 20)
- Agent-Environment Alignment via Automated Interface Generation (importance: 20, dev_relevance: 35)
- Decomposability-Guided Cooperative Coevolution for Large-Scale Itinerary Planning (importance: 10, dev_relevance: 10)
- Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models (importance: 20, dev_relevance: 20)
- AgentFly: Scaling Agentic Reinforcement Learning with Unified Resource System (importance: 25, dev_relevance: 40)
- Toward a Universal Characterization of Sudoku Puzzle Difficulty (importance: 5, dev_relevance: 10)
- OPTIC-ER: A Reinforcement Learning Framework for Real-Time Emergency Response and Equitable Resource Allocation in Underserved African Communities (importance: 8, dev_relevance: 5)
- Are LLM Agents Behaviorally Coherent? A Latent Profiling Approach for Social Simulation (importance: 15, dev_relevance: 15)
- LIFT: Likelihood-Based Intervention Faithfulness Training (importance: 25, dev_relevance: 25)
- Compositional Machine Design as Program Synthesis with LLMs (importance: 15, dev_relevance: 25)
- Toward Agents That Reason About Their Computation (importance: 20, dev_relevance: 25)
- e1: Learning Adaptive Control of Reasoning Effort (importance: 25, dev_relevance: 35)
- Multi-agent Undercover Gaming: Hallucination Removal via Counterfactual Test for Multimodal Reasoning (importance: 15, dev_relevance: 15)
- Alignment Faking - the Train -> Deploy Asymmetry: Through a Game-Theoretic Lens with Bayesian-Stackelberg Equilibria (importance: 25, dev_relevance: 10)
- AlignMerge - Alignment-Preserving Large Language Model Merging via Fisher-Guided Geometric Constraints (importance: 20, dev_relevance: 25)
- ROI-Reasoning: Rational Optimization for Inference via Pre-Computation Meta-Cognition (importance: 20, dev_relevance: 25)
- GDEPO: Group Dual-dynamic and Equal-right Advantage Policy Optimization for Sample-Efficient Reinforcement Learning (importance: 15, dev_relevance: 15)
- Stochastic CHAOS: Why Deterministic Inference Kills, and Distributional Variability Is the Heartbeat of Artifical Cognition (importance: 25, dev_relevance: 40)
- ReCreate: Reasoning and Creating Domain Agents Driven by Experience (importance: 25, dev_relevance: 35)
- Risky-Bench: Evaluating Agentic Safety Risks in Real-World Life-Assist Deployments (importance: 20, dev_relevance: 20)
- Simple Baselines are Competitive with Code Evolution (importance: 20, dev_relevance: 30)
- The Token Games: Evaluating Language Model Reasoning with Puzzle Duels (importance: 15, dev_relevance: 15)
- Strategy Executability in Mathematical Reasoning: Leveraging Human-Model Differences for Effective Guidance (importance: 12, dev_relevance: 10)
- Causal Identification from Counterfactual Data: Completeness and Bounding Results (importance: 8, dev_relevance: 5)
- Don't Ask: Understanding What Numeric Self-Reports Measure in Language Models (importance: 20, dev_relevance: 10)
- Beyond Preset Identities: Selective Stance Accommodation and Interaction Reorganisation in Generative Agent Societies (importance: 12, dev_relevance: 10)
- ScoringBench: A Benchmark for Evaluating Tabular Foundation Models with Proper Scoring Rules (importance: 15, dev_relevance: 20)
- Logarithmic Scores, Power-Law Discoveries: Disentangling Measurement from Coverage in Agent-Based Evaluation (importance: 25, dev_relevance: 35)
- Detecting Multi-Agent Collusion Through Multi-Agent Interpretability (importance: 20, dev_relevance: 15)
- Mitigating Distribution Sharpening in Math RLVR via Distribution-Aligned Hint Synthesis and Backward Hint Annealing (importance: 15, dev_relevance: 15)
- Agentic Architect: An Agentic AI Framework for Architecture Design Exploration and Optimization (importance: 25, dev_relevance: 35)
- OracleProto: A Reproducible Framework for Benchmarking LLM Forecasting via Knowledge Cutoffs and Temporal Masking (importance: 15, dev_relevance: 10)
- VibeServe: Can AI Agents Build Bespoke LLM Serving Systems? (importance: 30, dev_relevance: 60)
- Tools as Continuous Flow for Evolving Agentic Reasoning (importance: 15, dev_relevance: 25)
- MemQ: Integrating Q-Learning into Self-Evolving Memory Agents over Provenance DAGs (importance: 15, dev_relevance: 25)
- RewardHarness: Self-Evolving Agentic Post-Training (importance: 12, dev_relevance: 15)
- Data-driven Circuit Discovery for Interpretability of Language Models (importance: 15, dev_relevance: 10)
- Verifiable Process Rewards for Agentic Reasoning (importance: 15, dev_relevance: 20)
- Argumentation-Based Person-Tailored Storytelling for Older Adults: A Reflective Hybrid AI Architecture (importance: 8, dev_relevance: 10)
- Energy per Successful Goal: Goal-Level Energy Accounting for Agentic AI Systems (importance: 25, dev_relevance: 40)
- TADDLE: A Tool-Augmented Agent for Detecting Deficient LLM-Generated Peer Reviews (importance: 12, dev_relevance: 20)
- ConMoE: MoE Compression as Expert Selection and Functional Assignment (importance: 20, dev_relevance: 20)
- Conformal Certification of Reasoning Trace Prefixes (importance: 15, dev_relevance: 15)
- CoMIC: Collaborative Memory and Insights Circulation for Long-Horizon LLM Agents in Web Cloud-Edge Systems (importance: 15, dev_relevance: 25)
- Revisiting Ripple Effects in Knowledge Editing through Pressure-Aware Joint Neighborhood Optimization (importance: 15, dev_relevance: 15)
- Planning Takes More Than Token Prediction: Causal Plan for Benchmarking and Building Physically Grounded Embodied Reasoners (importance: 12, dev_relevance: 10)
- Trajectory-Refined Distillation (importance: 25, dev_relevance: 25)
- LATTEArena: An Evaluation Framework for LLM-powered Tabular Feature Engineering (Extended Version) (importance: 12, dev_relevance: 20)
- Business World Model (importance: 15, dev_relevance: 15)
- OSGuard: A Benchmark for Safety in Computer-Use Agents (importance: 25, dev_relevance: 30)
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems (importance: 12, dev_relevance: 5)
- From Brewing to Resolution: Tracing the Internal Lifecycle of Code Reasoning in LLMs (importance: 20, dev_relevance: 30)
- PseudoBench: Measuring How Agentic Auto-Research Fuels Pseudoscience (importance: 15, dev_relevance: 15)
- Decoupling Search from Reasoning: A Vendor-Agnostic Grounding Architecture for LLM Agents (importance: 25, dev_relevance: 55)
- Reward as An Agent for Embodied World Models (importance: 12, dev_relevance: 10)
- When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents (importance: 20, dev_relevance: 30)
- Plans Fade Fast: Why Context Management Is Load-Bearing for LLM Agents (importance: 30, dev_relevance: 60)
- Litmus: Zero-Label, Code-Driven Metric Specification for Evaluating AI Systems (importance: 20, dev_relevance: 40)
- Beyond Trajectory Imitation: Strategy-Guided Policy Optimization for LLM Reasoning (importance: 15, dev_relevance: 20)
- Understanding Rollout Error in Graph World Models (importance: 10, dev_relevance: 10)
- MedEvoEval: Evaluating Continual Evolution of Doctor Agents through Simulated Clinical Episodes (importance: 10, dev_relevance: 10)
- What a Pooled Leaderboard Does Not Say: Auditing the Scope and Provenance of Method Comparisons in Microservice RCA Benchmarks (importance: 15, dev_relevance: 50)
- Process Advantage Signal Shaping: Advantage Construction for Process-Supervised Reinforcement Learning in LLM Reasoners (importance: 15, dev_relevance: 20)
- When LLMs Develop Languages: Symbolic Communication for Efficient Multi-Agent Reasoning (importance: 15, dev_relevance: 25)
- Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading (importance: 30, dev_relevance: 60)
- Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Self-Improving Personal Agents (importance: 30, dev_relevance: 45)
- What We Talk About When We Talk About LLM Planning: Evidence for Two Distinct Planning Abilities (importance: 10, dev_relevance: 10)
- Tracing Agentic Failure from the Flow of Success (importance: 25, dev_relevance: 50)
- PRO-LONG: Programmatic Memory Enables Long-Horizon Reasoning (importance: 25, dev_relevance: 45)
- Agent-UCT: Upper Confidence Bounds Applied to Trees for Cost-Aware Agentic Workflow Optimization (importance: 20, dev_relevance: 35)
- Instruction-Tuned Language Models Cannot Sample from Distributions They Can Describe (importance: 20, dev_relevance: 15)
- SKILL-KD: Contrastive Skill Distillation for LLM Agents (importance: 25, dev_relevance: 40)
- MIND: Lightweight and Effective Memory Injection Defense for LLM Agents via Intent-Aware Information Bottleneck (importance: 25, dev_relevance: 40)
- LEMUR: Learning to Align with Multi-Objective Reinforcement Learning from Preference Feedback (importance: 12, dev_relevance: 10)
- Toward Fine-Grained Forgetting:Attribute Unlearning for Multimodal Large Language Models (importance: 15, dev_relevance: 10)
- LaCache: Robust Semantic Caching for LLM Serving (importance: 20, dev_relevance: 45)
- CockpitHAT: Dependency-Graph-Driven Hierarchical Attribution for Embodied Multi-Agent Cockpits (importance: 15, dev_relevance: 20)
- SearchMaster: Grounded and Regulated Self-Play for Search Agents (importance: 20, dev_relevance: 25)
- Surrogate Substitution Preserves PHI Detectability: A Multi-Detector Equivalence Study (importance: 12, dev_relevance: 20)
- Large language models for partial differential equation workflows (importance: 15, dev_relevance: 20)
- $\mathrm{R}{\scriptstyle\mathrm{EAL}}\mathrm{G}{\scriptstyle\mathrm{UARD}}$: Reliable, Efficient, and Auditable LLM Safeguards via Latent Reasoning (importance: 25, dev_relevance: 30)
- ContinualSkillBench: Can LLM Agents Truly Evolve Their Capabilities? (importance: 22, dev_relevance: 35)
- OctoLong: Mid-Training On Cross-Repository Code Contexts Enhances Long-Context Modeling (importance: 25, dev_relevance: 40)
- SkillEval: Decomposing Agent Skill Quality into Interpretable Signals (importance: 25, dev_relevance: 40)
- Illusion of Alignment: Detecting Hidden Disagreement in Collaborative Dialogue (importance: 12, dev_relevance: 10)
- Reading is not Reasoning: Bridging the Agentic Policy Gap in Vision-Text Compression (importance: 22, dev_relevance: 30)
- Emotion2Skill: Model-Internal Emotion Signals for Adaptive Skill Selection and Evolution (importance: 15, dev_relevance: 25)
- Bench2Robust: Scenario-Controlled Recovery for Tool-Using Agents (importance: 25, dev_relevance: 50)
- AI Model Design and Physician Liability (importance: 10, dev_relevance: 5)
- SafeBranch: Branch-Pair Safety Alignment for Embodied Agents (importance: 25, dev_relevance: 20)
- ReguSim: Evaluating LLM Agent Rule Grounding in Financial Compliance (importance: 30, dev_relevance: 30)
- Agentic neural architecture search for clinical prediction from electronic health records: Cross-hospital knowledge transfer and computational efficiency (importance: 20, dev_relevance: 35)
- ParallelWorld: Test-Time Scaling for Agentic Embodied Reasoning (importance: 25, dev_relevance: 25)
- When "Must" Becomes "Maybe": Erosion of Workflow Constraints in LLM Agent Handoffs (importance: 40, dev_relevance: 60)
- Parason: Revealing Subtask and Trial Parallelism in LLM Reasoning (importance: 35, dev_relevance: 45)
- PhysElite: How Far Are LLMs from Solving Olympiad-Level Physics Problems? (importance: 20, dev_relevance: 10)
- LoopArena: Benchmarking Models as Runtime Controllers for Loop Engineering (importance: 45, dev_relevance: 70)
- RetailAgent: Structured Adverse Timing in Self-Conditioned Multimodal LLM Trading Agents (importance: 20, dev_relevance: 10)
- Composition without a Host: Carrying the Spatiotemporal-Composability Guarantee of Plugin-Based Agents across Processes (importance: 30, dev_relevance: 65)
- FractalNet-Based Heterogeneous Federated Learning for Orbital Edge Intelligence in Satellite Mega-Constellations: A Wildfire Case Study (importance: 10, dev_relevance: 20)
- EmoStance: Response-Side Affective-Orientation Control for Empathetic Response Generation via Emoji Weak Supervision (importance: 15, dev_relevance: 15)
- STAIR (STructure Aware Information Retriever): A novel dataset and LLM based retriever for document structure augmentation (importance: 40, dev_relevance: 60)
- LayerRoute: Action-Conditioned Mixture-of-Layers Routing for Vision-Language-Action Policies (importance: 15, dev_relevance: 20)
- Style Over Substance: Content-Invariant Wrappers Flip LLM Safety-Judge Verdicts (importance: 35, dev_relevance: 35)
- Do Agents Know When They Succeed? Calibrating Agent Confidence from Internal Representations (importance: 30, dev_relevance: 45)
- GeoSkill:Experience-Driven Hierarchical Skill Learning with Collaborative Revision forGeospatialAgents (importance: 25, dev_relevance: 45)
- JaxAHT: A JAX-Based Library for Ad Hoc Teamwork (importance: 12, dev_relevance: 45)
- Enabling Creative Exploration for Vibe Design Agents (importance: 30, dev_relevance: 40)
- Not All AI Agents Are Equal: Characterizing Resource and Performance Dynamics (importance: 30, dev_relevance: 55)
- Spatial-Interactor: Learning Spatial Reasoning through Interaction with the Observable Physical World (importance: 15, dev_relevance: 10)
- ScholarStack: Layered Research Asset Orchestration and Cross-Task Reuse for Scientific Agents (importance: 20, dev_relevance: 35)
- Silent Failures in Agent-Tool Interaction: An Audit of ToolUniverse (importance: 35, dev_relevance: 60)
- Analyzing and Mitigating Cost-Inefficient Behaviors in Coding Agents (importance: 50, dev_relevance: 80)
- Prediction Limits and Koopman Closure of Geometry-Induced Soft State Abstractions (importance: 10, dev_relevance: 15)
- Beyond Accuracy: Counterfactual Fragility and Demographic Bias in Clinical Evaluation of LLMs (importance: 25, dev_relevance: 15)
- Right Answer, Wrong Reason: Accuracy, Consistency, and Consensus Are Misleading Indicators of LLM Faithfulness in Clinical Decision Support (importance: 25, dev_relevance: 15)
- Are Benchmarks Reliable? Toward Structural Diagnosis via Item-Level Capability Boundaries (importance: 30, dev_relevance: 35)
- Jev in Medicine: A Benchmark Evaluation (importance: 20, dev_relevance: 15)
- Test-Time Scaling via Budgeted Multi-Attribute Verification (importance: 30, dev_relevance: 40)
- Can Generative AI Automate Data Extraction for Meta-Analysis? A Case Study on Intercropping Research (importance: 15, dev_relevance: 15)
- Visual sensitivity is not claim retractability: persistence-aware credit assignment for multimodal reinforcement learning (importance: 25, dev_relevance: 25)
- Rep2Skill: Representation-Guided Skill Self-Evolution for LLM Agents (importance: 35, dev_relevance: 50)
- Beyond the Remembered World: Predictive 4D Belief for Persistent Navigation in Evolving Worlds (importance: 15, dev_relevance: 15)
- HHR: Hierarchical Hash Retrieval for Efficient LLM Generation (importance: 30, dev_relevance: 60)
- Code Owns the Simulation, Jev Owns the Evaluation (importance: 25, dev_relevance: 35)
- The AI Risk Observatory: What Can We Learn from AI Disclosures in Annual Reports About Societal Resilience? (importance: 15, dev_relevance: 5)
- LUMOS: Tracing Parametric Knowledge from Training Data to Behavioral Outputs in LLMs (importance: 30, dev_relevance: 30)
- RIFAR: Reliability and Forgetting-Aware Replay for Continual Robot Learning (importance: 15, dev_relevance: 15)
- Learning a Fact Is Not Learning How to Retrieve It (importance: 25, dev_relevance: 25)
- Retrieval-Augmented Large Language Model Decision-Making for Autonomous Driving Guided by Chinese Philosophical Wisdom (importance: 12, dev_relevance: 15)
- ROAR: Unifying Runs across Heterogeneous AI-Driven Research Systems (importance: 25, dev_relevance: 45)
- Behavioral History Outperforms Descriptions of the Person for LLM Synthetic Personas (importance: 20, dev_relevance: 10)
- SkillScriptBench: Benchmarking Self-Evolution of Executable Agent Skill Packages Beyond Markdown (importance: 40, dev_relevance: 70)
- Recursive Improvement of a Differentiable Scientific Software Ecosystem (importance: 30, dev_relevance: 60)
- Penumbra: Sample-Efficient Adversarial Search for Regulatory Obligations (importance: 25, dev_relevance: 30)
- PyINE: A Framework for Scalable Elicitation and Oversight via Code Execution (importance: 30, dev_relevance: 50)
- Do More Modalities Always Help? A Geometric Perspective on Missing-Modality Robustness (importance: 20, dev_relevance: 15)
- MemTrace: State-Consistent Memory for Long-Horizon Coding Agents (importance: 40, dev_relevance: 70)
- DREAM: Dynamic Resolution Assignment For Multimodal Multi-agent Debate (importance: 15, dev_relevance: 15)
- How Should a Prompt Optimizer Spend a Tight Budget? BudgetAPO with Noise-Adaptive Evaluation (importance: 30, dev_relevance: 55)
- MS-Exam-Gen: An Automatic Benchmark Construction Toolkit for Evaluating LLMs on Multiple Sclerosis-MRI Imaging Knowledge (importance: 12, dev_relevance: 20)
- Evolving in Thought Space: Training a Small Model at Test Time Unlocks Better Discoveries (importance: 25, dev_relevance: 30)
- GraphDecisionBench: Benchmarking System One Models on Graph Tasks (importance: 15, dev_relevance: 15)
- A Governed Agent Architecture for Natural-Language Dashboard Generation over Enterprise DataBrain (importance: 35, dev_relevance: 60)
- Adaptive Gait Biofeedback With Participant-Held-Out Modeling and Participant-Specific Updating in Chronic Ankle Instability (importance: 8, dev_relevance: 5)
- Where Rules End and Judges Begin: Measuring the Judgment Boundary in Multi-Agent Systems Security (importance: 35, dev_relevance: 50)
- zkLLMPoT: Efficient Zero Knowledge Proof of Training for Large Language Models (importance: 20, dev_relevance: 30)
- Open-MMUnlearning: Unifying Methods and Evaluation for MLLM Unlearning (importance: 20, dev_relevance: 30)
- Reading the Room: Foundations, Design, and Challenges of Normative Competence in LLMs (importance: 15, dev_relevance: 10)
- How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense (importance: 25, dev_relevance: 25)
- Defensive Writing in GPT Models (importance: 25, dev_relevance: 20)
- AtomWorld-Mirror: Macro-Step World Modeling of Critical Evolution Backbones for Materials Dynamics (importance: 12, dev_relevance: 10)
- MemTrial: Learning When to Trust Memory in LLM Portfolio Agents (importance: 20, dev_relevance: 25)
- Safe Actions Alone Do Not Ensure Safe Agents: Identifying Unfulfilled Obligations with Guard Models (importance: 30, dev_relevance: 45)
- PulseBound: Future-Beat State Forecasting Under an Explicit Information Boundary (importance: 8, dev_relevance: 15)
- Reward Hacking as a Transmissible Strategy in Multi-Agent Coding Systems (importance: 40, dev_relevance: 65)
- VETR: Execution-Grounded Self-Improving Agent for Symbolic Graphics Reasoning (importance: 18, dev_relevance: 30)
- SWE-Critix: Post-Training Generalist Verifiers for Assessing Coding-Agent Trajectories (importance: 40, dev_relevance: 70)
- Rollout-Time Intervention: Unlocking Agentic Reinforcement Learning Beyond Policy Limits (importance: 30, dev_relevance: 40)
- NeuSE: Neural SE(3)-Equivariant Embedding for Long-Term Object-Based Simultaneous Localization and Mapping (importance: 8, dev_relevance: 10)
- LLMRec: Benchmarking Large Language Models on Recommendation Task (importance: 10, dev_relevance: 10)
- Fast Diffusion-GAN Model for Generating Symbolic Music with Emotions (importance: 8, dev_relevance: 5)
- Optimal Transport Aggregation for Distributed Mixture-of-Experts (importance: 12, dev_relevance: 25)
- FlowWise: Stateful Fast-Slow Model Serving for Streaming Traffic Intelligence (importance: 15, dev_relevance: 35)
- Teaching MLP More Graph Information: A Three-stage Multitask Knowledge Distillation Framework (importance: 8, dev_relevance: 15)
- Prediction Is All MoE Needs: Expert Load Distribution Goes from Fluctuating to Stabilizing (importance: 25, dev_relevance: 40)
- GREAT: A Graph Edge Attention Network for the Electric Vehicle Routing Problem with Asymmetric Costs (importance: 8, dev_relevance: 15)
- Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation (importance: 20, dev_relevance: 30)
- Scale-specific auxiliary multi-task contrastive learning for deep liver vessel segmentation (importance: 5, dev_relevance: 5)
- Is Complex Query Answering Really Complex? (importance: 15, dev_relevance: 20)
- ReMAP: Lossless Residual Neural Lifting for Scalable MAP Inference in Arbitrary-Order Markov Random Fields (importance: 8, dev_relevance: 15)
- Open Foundation Models in Healthcare: Challenges, Paradoxes, and Opportunities with GenAI Driven Personalized Prescription (importance: 20, dev_relevance: 15)
- Randomly Wrong Signals: Bayesian Auction Design with ML Predictions (importance: 8, dev_relevance: 5)
- Natural Language Generation from Visual Events: State-of-the-Art and Key Open Questions (importance: 10, dev_relevance: 10)
- VQEL: Enabling Self-Play in Emergent Language Games via Agent-Internal Vector Quantization (importance: 8, dev_relevance: 10)
- A Generalist Hanabi Agent (importance: 8, dev_relevance: 10)
- Toxicity Is a Property of Communicative Events: A Contextual Stress Measurement Framework for NLP (importance: 15, dev_relevance: 15)
- Horizon Scans can be accelerated using novel information retrieval and artificial intelligence tools (importance: 15, dev_relevance: 35)
- SAEs Can Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs (importance: 15, dev_relevance: 20)
- FPGA-Based Neural Network Accelerators for Space Applications: A Survey (importance: 8, dev_relevance: 35)
- A Comprehensive Review on RNA Subcellular Localization Prediction (importance: 5, dev_relevance: 5)
- Graph Neural Network-Based Reinforcement Learning for Controlling Biological Networks - the GATTACA Framework (importance: 8, dev_relevance: 10)
- Towards Understanding Deep Learning Model in Image Recognition via Coverage Test (importance: 10, dev_relevance: 20)
- Generative AI for Autonomous Driving: A Review (importance: 15, dev_relevance: 15)
- Localizing Persona Representations in LLMs (importance: 12, dev_relevance: 10)
- Benchmarks Are Not Atomic: Composition-Aware LLM Evaluation using BenchHub (importance: 25, dev_relevance: 35)
- Simulating Student-Tutor Interactions with LLMs for Multilingual Math Feedback (importance: 10, dev_relevance: 10)
- GUARD: Guided Unlearning and Retention via Data Attribution for Large Language Models (importance: 12, dev_relevance: 20)
- From Empirical Evaluation to Context-Aware Enhancement: Repairing Regression Errors with LLMs (importance: 35, dev_relevance: 70)
- Co-Creative Learning via Metropolis-Hastings Interaction between Humans and AI (importance: 15, dev_relevance: 10)
- RCStat: A Statistical Framework for using Relative Contextualization in Transformers (importance: 25, dev_relevance: 35)
- Curriculum-Adapted Robust Reinforcement Learning for UAV Deconfliction in Adversarial Environments (importance: 15, dev_relevance: 10)
- From Individuals to Interactions: Benchmarking Gender Bias in Multimodal Large Language Models from the Lens of Social Relationship (importance: 20, dev_relevance: 15)
- Physics-Informed Graph Neural Networks to Reconstruct Local Fields Considering Finite Strain Hyperelasticity (importance: 12, dev_relevance: 5)
- Unequal Voices: How LLMs Construct Constrained Queer Narratives (importance: 15, dev_relevance: 5)
- Web3 x AI Agents: Landscape, Integrations, and Foundational Challenges (importance: 15, dev_relevance: 20)
- There are Levels to It: Red Teaming Dialogue With Hierarchical Reinforcement Learning (importance: 20, dev_relevance: 25)
- RCR-Router: Efficient Role-Aware Context Routing for Multi-Agent LLM Systems with Structured Memory (importance: 25, dev_relevance: 60)
- One Size Does Not Fit All: Distribution-Guided Tensor-Wise Adaptive Sparsification for Model Merging (importance: 20, dev_relevance: 35)
- SciRerankBench: Benchmarking Rerankers Towards Scientific Retrieval-Augmented Generation with LLMs (importance: 25, dev_relevance: 50)
- PPGFlowECG: Latent Rectified Flow with Cross-Modal Encoding for PPG-Guided ECG Generation and Cardiovascular Disease Detection (importance: 10, dev_relevance: 5)
- TENET: One Step Toward Test-Driven Development for Repository-Level Code Generation (importance: 40, dev_relevance: 75)
- Cognifying Education: Mapping AI's transformative role in emotional, creative, and collaborative learning (importance: 15, dev_relevance: 10)
- Frequency-Aware Model Parameter Explorer: A new attribution method for improving explainability (importance: 15, dev_relevance: 15)
- Rethinking Langevin Thompson Sampling from A Stochastic Approximation Perspective (importance: 10, dev_relevance: 5)
- Auditing a KB Elicitation of Frontier LLM Knowledge: A Multi-dimensional Analysis of GPTKB v1.5 (importance: 20, dev_relevance: 10)
- Batch Bayesian Active Learning with Partial Batch Label Sampling (importance: 15, dev_relevance: 15)
- A Function-Centric Perspective on Flat and Sharp Minima (importance: 15, dev_relevance: 10)
- SoK: Systematizing LLM Prompt Security: Taxonomies, Datasets, and Unified Evaluation of Attacks and Defenses (importance: 35, dev_relevance: 40)
- Iterative Critique-Refine Framework for Enhancing LLM Personalization (importance: 15, dev_relevance: 20)
- Training LLMs Beyond Next Token Prediction - Filling the Mutual Information Gap (importance: 25, dev_relevance: 15)
- Energy-Efficient Deep Learning Without Backpropagation: A Rigorous Hardware-Validated Benchmarking Study of Forward-Only Algorithms (importance: 20, dev_relevance: 15)
- ProDER: A Continual Learning Approach for Fault Classification and Localization in Evolving Smart Grids (importance: 10, dev_relevance: 5)
- Exploring the Underwater World Segmentation without Extra Training (importance: 10, dev_relevance: 5)
- Naga: Vedic Encoding for Deep State Space Models (importance: 10, dev_relevance: 5)
- UnSAMv2: Self-Supervised Learning Enables Segment Anything at Any Granularity (importance: 20, dev_relevance: 15)
- Neural Evidence and Behavioural Welfare Economics: When Can Neuroeconomics Inform Policy? (importance: 10, dev_relevance: 2)
- Explainable Visual Anomaly Detection via Concept Bottleneck Models (importance: 15, dev_relevance: 10)
- Semantics as a Shield: Label Disguise Defense (LDD) against Prompt Injection in LLM Sentiment Classification (importance: 20, dev_relevance: 25)
- Measuring all the noises of LLM Evals (importance: 30, dev_relevance: 45)
- PhyAVBench: A Challenging Audio Physics-Sensitivity Benchmark for Physically Grounded Text-to-Audio-Video Generation (importance: 12, dev_relevance: 5)
- Mechanistic Knobs in LLMs: Retrieving and Steering High-Order Semantic Features via Sparse Autoencoders (importance: 20, dev_relevance: 15)
- HERO: Hierarchical Evidential Reasoning Optimization for Radiology Report Generation via Reason-then-Summarize (importance: 10, dev_relevance: 5)
- ECLIPTICA -- A Framework for Switchable LLM Alignment via CITA - Contrastive Instruction-Tuned Alignment (importance: 15, dev_relevance: 10)
- GeoMotionGPT: Geometry-Aligned Motion Understanding with Large Language Models (importance: 10, dev_relevance: 5)
- Hiking in the Wild: A Scalable Perceptive Parkour Framework for Humanoids (importance: 12, dev_relevance: 3)
- Representational Similarity Alignment of Figurative Language in Humans and Large Language Models (importance: 12, dev_relevance: 5)
- Architecture Before the Formula: Training-Dependent Computational Semantics in Transformers (importance: 15, dev_relevance: 10)
- Less is More: Label-Guided Efficient Summarization of Procedural and Instructional Videos (importance: 10, dev_relevance: 5)
- NeuroVLA: A Brain-inspired Embodied Intelligence for Fluid and Fast Reflexive Robotics Control (importance: 12, dev_relevance: 3)
- NeuroFilter: Activation-Based Guardrails for Privacy-Conscious LLM Agents (importance: 25, dev_relevance: 40)
- Where is the multimodal goal post? On the Ability of Foundation Models to Recognize Highlights in Videos (importance: 10, dev_relevance: 5)
- Retell, Reward, Repeat: Reinforcement Learning for Narrative Theory-Informed Story Retelling (importance: 10, dev_relevance: 5)
- Uni-RS: A Spatially Faithful Unified Understanding and Generation Model for Remote Sensing (importance: 10, dev_relevance: 5)
- PixSearch: Region-Grounded Retrieval for Knowledgeable Large Multimodal Models (importance: 15, dev_relevance: 20)
- The pretraining domain outweighs the training objective in setting the privacy-utility trade-off of differentially private medical image analysis (importance: 12, dev_relevance: 5)
- VERGE: Formal Refinement and Guidance Engine for Verifiable LLM Reasoning (importance: 20, dev_relevance: 30)
- Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening (importance: 25, dev_relevance: 30)
- OpenVTON-Bench: A Large-Scale High-Resolution Benchmark for Controllable Virtual Try-On Evaluation (importance: 10, dev_relevance: 5)
- From Associations to Activations: Comparing Behavioral and Hidden-State Semantic Geometry in LLMs (importance: 10, dev_relevance: 5)
- DCD: Decomposition-based Causal Discovery from Autocorrelated and Non-Stationary Temporal Data (importance: 10, dev_relevance: 5)
- TTT-Parkour: Rapid Test-Time Training for Perceptive Robot Parkour (importance: 12, dev_relevance: 3)
- Flow Policy Gradients for Robot Control (importance: 10, dev_relevance: 5)
- DRMOT: A Dataset and Framework for RGBD Referring Multi-Object Tracking (importance: 10, dev_relevance: 5)
- Coupled Local and Global World Models for Efficient First Order RL (importance: 10, dev_relevance: 5)
- On the Dynamics of Delayed Generalization in Linear Models (importance: 10, dev_relevance: 3)
- Combining scEEG and PPG for reliable sleep staging (importance: 8, dev_relevance: 3)
- Articulated 3D Scene Graphs for Open-World Mobile Manipulation (importance: 10, dev_relevance: 3)
- Formal Policy Enforcement for Agentic Systems (importance: 30, dev_relevance: 60)
- Prompt Interference in LLM Post-Training: Why Pass@k Optimization Can Degrade Pass@1 (importance: 20, dev_relevance: 30)
- An Analysis of Role Semantics in Directed Dynamic Graph Modeling for Edge Classification (importance: 8, dev_relevance: 5)
- Toward Guarantees for Clinical Reasoning in Vision Language Models via Formal Verification (importance: 10, dev_relevance: 5)
- Learning Shortest Paths with Generative Flow Networks (importance: 10, dev_relevance: 5)
- CodecFlow: Efficient Bandwidth Extension via Conditional Flow Matching in Neural Codec Latent Space (importance: 10, dev_relevance: 5)
- A Counterfactual Graph Approach to Oppositional Narrative Analysis (importance: 10, dev_relevance: 3)
- Neural-Surrogate Co-Design of Aerodynamic Shape and Control (importance: 8, dev_relevance: 3)
- Scheduling Parallel Optical Circuit Switches for AI Training (importance: 15, dev_relevance: 20)
- HybridStitch: Token-Adaptive Model Stitching for Efficient Diffusion Generation (importance: 10, dev_relevance: 5)
- Grammar of the Wave: Towards Explainable Multivariate Time Series Event Detection via Neuro-Symbolic VLM Agents (importance: 10, dev_relevance: 5)
- OxyGen: Unified KV Cache Management for VLA Inference under Multi-Task Parallelism (importance: 12, dev_relevance: 20)
- AnolisVLA: Lightweight Vision-Language-Action Model with Deep State Space Models for Mobile Manipulation (importance: 8, dev_relevance: 3)
- DyJR: Preserving Diversity in Reinforcement Learning with Verifiable Rewards via Dynamic Jensen-Shannon Replay (importance: 15, dev_relevance: 15)
- Contrastive Metric Learning for Point Cloud Segmentation in Highly Granular Detectors (importance: 8, dev_relevance: 3)
- TopoPilot: Reliable Conversational Workflow Automation for Topological Data Analysis and Visualization (importance: 12, dev_relevance: 25)
- JAL-Turn: Joint Acoustic-Linguistic Modeling for Real-Time and Robust Turn-Taking Detection in Full-Duplex Spoken Dialogue Systems (importance: 10, dev_relevance: 15)
- IMAGAgent: Orchestrating Multi-Turn Image Editing via Constraint-Aware Planning and Reflection (importance: 10, dev_relevance: 5)
- Spec-Harness: Assessing and Guiding LLM-Synthesized Formal Specifications Beyond Verifiability (importance: 25, dev_relevance: 60)
- AromaGen: Interactive Generation of Rich Olfactory Experiences with Multimodal Language Models (importance: 8, dev_relevance: 5)
- Process Rewards for Outcome-Guided Reasoning Steps in LLM Reinforcement Learning (importance: 15, dev_relevance: 10)
- Pickalo: Leveraging 6D Pose Estimation for Low-Cost Industrial Bin Picking (importance: 8, dev_relevance: 3)
- Generating Synthetic Doctor-Patient Conversations for Long-form Audio Summarization (importance: 10, dev_relevance: 10)
- CMP: Robust Whole-Body Tracking for Loco-Manipulation via Competence Manifold Projection (importance: 8, dev_relevance: 3)
- A Survey of Secure Retrieval-Augmented Generation (importance: 25, dev_relevance: 40)
- PR-MaGIC: Prompt Refinement Via Mask Decoder Gradient Flow For In-Context Segmentation (importance: 8, dev_relevance: 5)
- Feed-Forward 3D Scene Modeling: A Problem-Driven Perspective (importance: 8, dev_relevance: 3)
- Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems (importance: 0, dev_relevance: 85)
- Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learning (importance: 8, dev_relevance: 3)
- Negative Advantages Is a Double-Edged Sword: Calibrating advantages in GRPO for Search Agents (importance: 15, dev_relevance: 20)
- TurboEvolve: Towards Fast and Robust LLM-Driven Program Evolution (importance: 12, dev_relevance: 35)
- UniT: Toward a Unified Physical Language for Human-to-Humanoid Policy Learning and World Modeling (importance: 8, dev_relevance: 3)
- MILP the Gap: An Automated Reasoning Approach to Condorcet's Paradox (importance: 8, dev_relevance: 5)
- Call-Chain-Aware LLM-Based Test Generation for Java Projects (importance: 30, dev_relevance: 75)
- PermaFrost-Attack: Stealth Pretraining Seeding(SPS) for planting Logic Landmines During LLM Training (importance: 15, dev_relevance: 15)
- Cross-Stage Coherence in Hierarchical Driving VQA: A Consistency Protocol and Parameter-Efficient Gated Context Routing (importance: 8, dev_relevance: 5)
- Hiding in Plain Sight: Detectability-Aware Antidistillation of Reasoning Models (importance: 25, dev_relevance: 15)
- Structure Liberates: How Constrained Sensemaking Produces More Novel Research Output (importance: 15, dev_relevance: 20)
- Dependency Parsing Across the Resource Spectrum: Evaluating Architectures on High and Low-Resource Languages (importance: 8, dev_relevance: 5)
- Refining Compositional Diffusion for Reliable Long-Horizon Planning (importance: 8, dev_relevance: 3)
- How Do Novice Describers Edit AI Drafts? The Role of Draft Quality in Audio Description Authoring (importance: 10, dev_relevance: 5)
- XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity (importance: 25, dev_relevance: 15)
- Selective Successor Eligibility Traces for RLVR (importance: 25, dev_relevance: 30)
- AdaGamma: State-Dependent Discounting for Temporal Adaptation in Reinforcement Learning (importance: 15, dev_relevance: 15)
- Multimodal synthesis of MRI and tabular data with diffusion in a joint latent space via cross-attention (importance: 15, dev_relevance: 10)
- APEX: Assumption-free Projection-based Embedding eXamination Metric for Generative Distribution Evaluation (importance: 20, dev_relevance: 25)
- HIMMEL: Hierarchical Interleaved Multi-stream Motion Encoding for Long Video Understanding (importance: 20, dev_relevance: 15)
- When and How to Canonize: A Generalization Perspective (importance: 15, dev_relevance: 10)
- Discrete Flow Matching for Offline-to-Online Reinforcement Learning (importance: 15, dev_relevance: 10)
- Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training (importance: 30, dev_relevance: 40)
- Discrete MeanFlow: One-Step Generation via Conditional Transition Kernels (importance: 12, dev_relevance: 10)
- Revealing the Gap in Human and VLM Scene Description through Counterfactual Semantic Saliency (importance: 12, dev_relevance: 10)
- Towards Long-horizon Embodied Agents with Tool-Aligned Vision-Language-Action Models (importance: 18, dev_relevance: 15)
- A Few Steps Further: Why Defenses Against Malicious Finetuning Erode Under Continued Training (importance: 25, dev_relevance: 15)
- Concurrency without Model Changes: Future-based Asynchronous Function Calling for LLMs (importance: 35, dev_relevance: 70)
- $\alpha$-TCAV: A Unified Framework for Testing with Concept Activation Vectors (importance: 12, dev_relevance: 15)
- Conditioned in Training, Dispensable at Inference: Typological Conditioning for Multilingual Speech Translation (importance: 15, dev_relevance: 10)
- LoopQ: Quantization for Looped Language Models (importance: 18, dev_relevance: 40)
- The Impact of AI Search on the Online Content Ecosystem: Evidence from Google and Reddit (importance: 25, dev_relevance: 15)
- Global Automation Atlas (importance: 15, dev_relevance: 5)
- Learning Higher-Order Interactions from Incomplete Spatiotemporal Data with Mixtures of Hypergraph Laplacians (importance: 12, dev_relevance: 10)
- $\boldsymbol{f}$-OPD: Stabilizing Long-Horizon On-Policy Distillation with Freshness-Aware Control (importance: 22, dev_relevance: 35)
- Training data attribution in diffusion models via mirrored unlearning and noise-consistent skew (importance: 12, dev_relevance: 10)
- Exact Symmetry as Algebra: Group-Structured Tensors that Enforce Physical Selection Rules (importance: 15, dev_relevance: 10)
- Stdlib or Third-Party? Empirical Performance and Correctness of LLM-Assisted Zero-Dependency Python Libraries (importance: 35, dev_relevance: 85)
- CTI2TTP: Evidence-Grounded TTP Extraction from Cyber Threat Intelligence via Diverge-then-Converge Verification (importance: 15, dev_relevance: 15)
- GAC: Noise-Aware Adaptive Mixing for Hybrid SFT-RL Post-Training (importance: 22, dev_relevance: 30)
- Beyond Questions: Evaluating the Stated Knowledge of LLMs (importance: 18, dev_relevance: 10)
- How Do Uncertainty Estimators Align with Hallucination Occurrence in LLMs? (importance: 25, dev_relevance: 25)
- When LLM Reward Design Fails: Diagnostic-Driven Refinement for Sparse Structured RL (importance: 20, dev_relevance: 40)
- GEO-Bench: Benchmarking Ranking Manipulation in Generative Engine Optimization (importance: 15, dev_relevance: 10)
- Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems (importance: 25, dev_relevance: 40)
- idSCD: Identifying Training Datasets through Semantic Correlation Descriptors (importance: 12, dev_relevance: 10)
- Hoeffding Concept Bottleneck Models with Applications to Overhead Images (importance: 10, dev_relevance: 8)
- Real2SAM2Real: Generative 3D Caches as Complementary Context for Video Diffusion (importance: 10, dev_relevance: 5)
- Shape Your Body: Value Gradients for Multi-Embodiment Robot Design (importance: 10, dev_relevance: 5)
- How Much Can Repeated Running Buy? A Measured Variance Ceiling for LLM Code Evaluation (importance: 30, dev_relevance: 55)
- The Value of Spike Timing: A Leakage-Resistant Benchmark of SNN Design Choices for Network Intrusion Detection (importance: 10, dev_relevance: 10)
- Learning When to Translate for Multilingual Reasoning (importance: 18, dev_relevance: 20)
- Economy of Minds: Emerging Multi-Agent Intelligence with Economic Interactions (importance: 20, dev_relevance: 25)
- OpenAgenet / OAN White Paper: Open Infrastructure for Trusted Agent Interconnection (importance: 20, dev_relevance: 45)
- OpenAgenet / OAN Yellow Paper: Technical Architecture for Trust-Governed Resource Identity and Discovery (importance: 22, dev_relevance: 50)
- Adaptive Patching Is Harder Than It Looks For Time-Series Forecasting (importance: 12, dev_relevance: 10)
- Instant-Fold: In-Context Imitation Learning for Deformable Object Manipulation (importance: 10, dev_relevance: 5)
- NoRA: Evaluating Grounded Reasonableness in Visual First-person Normative Action Reasoning (importance: 15, dev_relevance: 10)
- Plan, Watch, Recover: A Benchmark and Architectures for Proactive Procedural Assistance (importance: 12, dev_relevance: 10)
- Improving Heart-Focused Medical Question Answering in LLMs via Variance-Aware Rubric Rewards with GRPO (importance: 15, dev_relevance: 25)
- TAM: Torque Adaptation Module for Robust Motion Transfer in Manipulation (importance: 10, dev_relevance: 5)
- The Geography of Algorithmic Judgment: LLM Intermediaries, Place Identity, and Racial Steering in Housing Search (importance: 25, dev_relevance: 10)
- DataEvolver: Automatic Data Preparation for Large Language Models through Multi-Level Self-Evolving (importance: 25, dev_relevance: 35)
- One Skill, Two Masters: Serving Users and Attackers through Skill Injection (importance: 30, dev_relevance: 60)
- FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning (importance: 10, dev_relevance: 5)
- Beyond Locality: Feature-Boundary Auditing for OS-Integrated AI (importance: 20, dev_relevance: 25)
- RAIL: Rethinking Auditory Intelligence in Large Audio-Language Models with a CHC-Grounded Benchmark (importance: 12, dev_relevance: 8)
- Task-Aligned Stability Analysis of Vision-Language Models for Autonomous Driving Hazard Detection (importance: 10, dev_relevance: 5)
- HiLo-Token: Input-Adaptive High-Low Frequency Token Compression for Efficient Image Editing (importance: 15, dev_relevance: 25)
- ThreadBench: Benchmarking Realism in Agentic Simulation of Real-world Discussions (importance: 12, dev_relevance: 10)
- LLM Judges Have Dark Current: Stratified Null and Utility Tests for LLM Judge Reliability (importance: 22, dev_relevance: 40)
- Retrievable Gradients: Continual Post-Training Without Cumulative Weight Drift (importance: 18, dev_relevance: 15)
- Beyond Reward Engineering: A Data Recipe for Long-Context Reinforcement Learning (importance: 20, dev_relevance: 25)
- Data-Aware Model Merging in the Essential Subspace for Multi-Task Knowledge Aggregation (importance: 18, dev_relevance: 25)
- Human-like autonomy emerges from self-play and a pinch of human data (importance: 10, dev_relevance: 5)
- FinRED: An Expert-Guided Benchmark Generation and Evaluation Framework for Financial LLM Red-Teaming (importance: 15, dev_relevance: 15)
- ScholarQuest: A Taxonomy-Guided Benchmark for Agentic Academic Paper Search in Open Literature Environments (importance: 15, dev_relevance: 20)
- RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulation (importance: 10, dev_relevance: 5)
- On the Sparsity-Storage-Accuracy Tradeoff in Parsimoniously Activated Dictionary Learning (importance: 8, dev_relevance: 8)
- Learning with a Single Rollout via Pass@k Policy Optimization (importance: 25, dev_relevance: 35)
- From Clicks to Intent: Cross-Platform Session Embeddings with LLM-Distilled Taxonomy for Financial Services Recommendations (importance: 10, dev_relevance: 5)
- Error-Conditioned Neural Solvers (importance: 10, dev_relevance: 10)
- SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation (importance: 20, dev_relevance: 25)
- RoboGaze: Evaluating Robot World Models via Structured Vision-Language Analysis (importance: 10, dev_relevance: 8)
- When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs (importance: 35, dev_relevance: 65)
- The Speedup Paradox: Rethinking Inference Speed-Quality Trade-off in Embodied Tasks (importance: 10, dev_relevance: 5)
- Behavior Uncloning: Distilling Mode Redirection into Policy Weights without Inference-Time Steering (importance: 8, dev_relevance: 3)
- Critical Interval MSE: Toward Reliable Offline Validation for Robot Manipulation Policies (importance: 8, dev_relevance: 3)
- Federated Learning with Energy-Based Structured Probabilistic Inference (importance: 10, dev_relevance: 10)
- Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs (importance: 18, dev_relevance: 15)
- Optimal any-angle path planning in static and dynamic environments (importance: 8, dev_relevance: 15)
- Ember: A Kilobyte-State Optimizer for Token Interfaces (importance: 22, dev_relevance: 30)
- Beyond Cultural Persona Bias: Response Geometry as a Measure for Language Models and People (importance: 12, dev_relevance: 8)
- AutoResearch: An Execution-Grounded Multi-Agent Framework for Reliable Research Workflow Automation (importance: 22, dev_relevance: 45)
- TabQueryBench: A Query-Centric Benchmark for Synthetic Tabular Data (importance: 12, dev_relevance: 20)
- Auto-AEG: Scalable Data Construction for Open-Vocabulary Audio Event Grounding (importance: 8, dev_relevance: 5)
- HarnessMDP: Learning to Control LLM Agent Harnesses with Offline Reinforcement Learning (importance: 28, dev_relevance: 65)
- Graph Convolutional Attention: A Spectral Perspective on Graph Denoising and Diffusion (importance: 8, dev_relevance: 5)
- Multi-Agent AI Control: Distributed Attacks Hamper Per-Instance Monitors (importance: 25, dev_relevance: 30)
- EgoWAM: World Action Models Beyond Pixels with In-the-Wild Egocentric Human Data (importance: 10, dev_relevance: 5)
- Phone Segmentation and Recognition through Phonological Activation Mapping (importance: 8, dev_relevance: 5)
- Automatic Thematic Indexing of Large Literary Corpora: A Machine Learning Approach to Voltaire's Complete Works (importance: 8, dev_relevance: 5)
- An LLM-powered Agentic Recommendation System for Connected TV Content Discovery (importance: 12, dev_relevance: 15)
- Comparing Socially-Equitable Renewable Energy Budget Allocation MDP Policies in Mature and Emerging Economies (importance: 8, dev_relevance: 10)
- Scaling Point-in-Time Language Models: Enabling Economic Evaluation of Embeddings (importance: 20, dev_relevance: 20)
- Bulkhead: Automated Semantic Detection and Remediation of Container Escape Vulnerabilities (importance: 30, dev_relevance: 60)
- Memory as a Controlled Process: Learned Adaptive Memory Management for LLM Agents (importance: 25, dev_relevance: 45)
- Never Too Late for Force: Accelerating VLA Post-Training with Reactive Force Injection (importance: 8, dev_relevance: 3)
- HALLMARK: How to Diagnose LLM Citation Verifiers and When to Deploy Them (importance: 25, dev_relevance: 35)
- Audit Before You Update: Recording Artifacts and the Limits of Bounded Model Updates in MIMIC-IV ICU Prediction (importance: 10, dev_relevance: 10)
- Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents (importance: 12, dev_relevance: 25)
- ChannelGuard: From Borrowed Safety to Application-Controlled Defenses in Multi-Agent LLMs (importance: 25, dev_relevance: 40)
- REGEN: Replay-recycling for Expert-to-Generalist distillation with Offline Reinforcement Learning (importance: 20, dev_relevance: 25)
- Active Inference as a Convex Markov Decision Process (importance: 12, dev_relevance: 12)
- Activation Steering with Jungian Cognitive Functions (importance: 25, dev_relevance: 30)
- DBA-Bench: A Production-Fidelity Benchmark for LLM-Based Database Operations Agents (importance: 35, dev_relevance: 60)
- Constraint-Aware Influence Estimation in Deep Constrained Learning (importance: 20, dev_relevance: 25)
- Mission-Level Runtime Assurance for LLM-Assisted ISR Swarms over a Verification-Aware Fabric (importance: 20, dev_relevance: 30)
- This Is a Good Paper, Right? How Tag Questions Affect Sycophancy Across 106 Language Models (importance: 40, dev_relevance: 25)
- ForgetBench: Benchmarking Forgetting Dynamics of Long-Term Parametric Memory in Language Models (importance: 30, dev_relevance: 15)
- Depth, Not Breadth: Best-of-N Jailbreaking Beyond Surface Noise (importance: 35, dev_relevance: 20)
- Same Illness, Different Voice: Restating the Patient's Account Narrows Narrative Anchoring in Clinical Language Models (importance: 25, dev_relevance: 10)
- Specification-Guided Synthesis of Deadlock-Free Communication Protocol Refinements with Large Language Models (importance: 35, dev_relevance: 55)
- Flux-OPD: On-Policy Distillation with Evolving Contexts (importance: 30, dev_relevance: 20)
- COSI-Lab: Conference Living Lab for Modeling Multi-Perspective Multimodal Social Intention (importance: 15, dev_relevance: 5)
- Gated Q-learning: Add Off-Policy Bias to Taste (importance: 25, dev_relevance: 15)
- LLM-OSDA: An Optimal-Stopping Dynamic Auction for Native Advertising in Multi-Turn LLM Conversations (importance: 30, dev_relevance: 15)
- From Chasing Ghosts to Missed Attacks: Perspectives and Perceptions of SOC Practitioners on LLM Integration, Risks, and Readiness (importance: 30, dev_relevance: 35)
- AI Alignment and Fiduciary Obligation (importance: 25, dev_relevance: 5)
- Patients-like-me: A Variational LM--GNN Framework for Explainable Clinical Prediction (importance: 20, dev_relevance: 10)
- Need or Reputation? Region, National Income, and Academic Status in LLM Gatekeeping Decisions (importance: 20, dev_relevance: 5)
- Factorized Hypothesis Search for Evidence-to-Taxonomy Retrieval (importance: 25, dev_relevance: 35)
- LifelongCrossNav: Persistent 3D Semantic Memory for Cross-Floor Multi-Object Navigation (importance: 15, dev_relevance: 10)
- Measuring Concept Content in Text from LLM Activations: ESG Evidence from Concept Vectors and Linear Probes (importance: 25, dev_relevance: 15)
- Search-G1: Co-Evolving Representation-Based Rewards for Selective Grounded Search (importance: 35, dev_relevance: 40)
- Router Sensitivity Under Lightweight Fine-Tuning Identifies Prunable Experts in Mixture-of-Experts Models (importance: 30, dev_relevance: 35)
- Lost in Context Compaction: When LLM Systems Drop Side Constraints (importance: 55, dev_relevance: 75)
- A, B, C, or Stable? Accuracy and Order Sensitivity Diverge in LLM Multiple-Choice Evaluation (importance: 30, dev_relevance: 30)
- P2Skill: Privacy Preserving Skill Distillation for Cloud-Local LLM Inference Systems (importance: 30, dev_relevance: 35)
- Plausible but Not Valid: A Psychometric Audit of LLMs as Synthetic Survey Respondents (importance: 25, dev_relevance: 10)
- Listen, Reason, and Segment: Aligning LALMs with Editorial Judgment for Media Chapterization (importance: 25, dev_relevance: 10)
- SIDScope: A Diagnostic Resource for Semantic-ID Interfaces in Generative Recommendation (importance: 20, dev_relevance: 40)
- DualOPSD: Closing the Feedback Loop in On-Policy Self-Distillation (importance: 30, dev_relevance: 20)
- Delegating Before Learning: Where Generative AI Sits in Students' Professional Communication (importance: 25, dev_relevance: 5)
- Disclosure-Gated User Simulation for Companion-Agent Evaluation (importance: 25, dev_relevance: 30)
- EmbodiedSkills: World Action Model-augmented Cascaded Skills for Vision-Language-Action Agent (importance: 20, dev_relevance: 15)
- LatentPress: Context Compression Beyond Text and Vision (importance: 35, dev_relevance: 40)
- TC-Next: Zero-Shot Multimodal Cyclone Forecasting (importance: 15, dev_relevance: 5)
- From Portraits to Pathology: Probing Diagnostic Overreach in Vision-Language Models (importance: 25, dev_relevance: 5)
- SeRV: Semantic-Aligned Residual Vector Quantization for American Sign Language Generation (importance: 20, dev_relevance: 5)
- Open Tabular Insight Extraction: Where Do We Stand, and Where Should We Go? (importance: 25, dev_relevance: 25)
- RoboCousin: Build Your Own Simulation Playground for Robust Bimanual Robotic Manipulation (importance: 15, dev_relevance: 10)
- X2Streaming-ASR: wait when uncertain, emit when ready for streaming ASR (importance: 25, dev_relevance: 25)
- SPINE: Measuring LLM Sycophancy under Sustained Multi-Turn Pressure (importance: 35, dev_relevance: 30)
- Divide, Consult, Conquer: Capability Laundering Through Aligned LLMs (importance: 40, dev_relevance: 25)
- CoReLoop: Parameter-Efficient Controlled Recurrent Refinement for Audio Deepfake Detection (importance: 15, dev_relevance: 10)
- Forecasting Brain Tracer Distribution from Pre-Contrast MRI: A Comparison of Direct Regression and Flow Matching (importance: 10, dev_relevance: 5)
- Same Outcome, Different Readout: What Does a Steerable Valence Direction in LLMs Represent? (importance: 25, dev_relevance: 10)
- Propose, Verify, Commit: Evidence-Grounded Memory for Long-Horizon Multi-Actor Conversations (importance: 35, dev_relevance: 40)
- Do Language Models Need Music Supervision? Verifiable Rewards for Multi-Constraint Symbolic Music Generation (importance: 20, dev_relevance: 15)
- Misaligned Clinical Risk Classification and Cost Asymmetry in Open-Weight Large Language Models (importance: 20, dev_relevance: 10)
- GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay (importance: 20, dev_relevance: 15)
- PICPIs: Prediction-Interval-Conditional Prediction Intervals (importance: 20, dev_relevance: 15)
- Quantization-Robust Unlearning through the Lens of Retain-Forget Loss Landscapes Interaction (importance: 30, dev_relevance: 25)
- When Context Misleads: In-context Learning with Jurisdiction in Large Language Models (importance: 30, dev_relevance: 20)
- Beyond Unsafe Detection: Counterfactually Anchored Evidence Attribution for Multi-Turn LLM Safety Failures (importance: 30, dev_relevance: 35)
- R-DEIM Net: An Efficient Rationale-Augmented Dual-Expert Interaction Model for Paraphrase Detection (importance: 15, dev_relevance: 15)
- Verification as an Architectural Layer for LLM Agents: A V-Model Design, and a Pilot Study of Its Deterministic Core (importance: 40, dev_relevance: 60)
- SoFT: Soft Targets for Generalizable LLM Fine-Tuning (importance: 30, dev_relevance: 25)
- LLM Alignment--Utility Asymmetry under Semantic-Preserving Transformations (importance: 25, dev_relevance: 15)
- Eval4DiRec: A Unified and Systematic Evaluation Framework for Diffusion-based Recommender Systems (importance: 20, dev_relevance: 15)
- ESTHER: Egocentric Stereo Hand Estimation and Reconstruction in the Wild (importance: 15, dev_relevance: 5)
- The Compiler May Read It, the Agent May Not: Keeping Part of a Research Code Away from a Coding Agent (importance: 45, dev_relevance: 75)
- Draft in Parallel, Condition Through Depth: Adjacent Causal Injection for Speculative Decoding (importance: 25, dev_relevance: 35)
- Proofs Without Nominals: G\"odel's Ontological Argument, its Shallow Embedding, and the Open Questions of the Monatshefte Notes (importance: 5, dev_relevance: 0)
- WeLike2Party! In-Context Motion Transfer for Multi-Human Image Animation (importance: 15, dev_relevance: 5)
- Purlin: Separating Orchestration from the Datapath of Collectives (importance: 25, dev_relevance: 50)
- V-JEPA Policy: Building Effective World-Action Models on Predictive Visual Latents (importance: 20, dev_relevance: 10)
- On the Relaxation of Conditional Independence Assumption for Image Segmentation (importance: 15, dev_relevance: 10)
- Structure vs. Chain-of-Thought: Evaluating LLM Criteria Extraction for Depression Severity (importance: 20, dev_relevance: 10)
- Ready2Blend: From Natural-Language Instructions to Composable Alignment Prompts (importance: 25, dev_relevance: 20)
- Tactile Curiosity Drives Robot Interaction (importance: 15, dev_relevance: 5)
- ViTeX-Bench: Benchmarking High-Fidelity Video Scene Text Editing (importance: 15, dev_relevance: 5)
- Screw Attention: Rigid-Body Algebra Inside a Transformer (importance: 20, dev_relevance: 10)
- MoLE: Mixture of Latent Experts for Complementary Visual Reasoning (importance: 15, dev_relevance: 5)
- One Distribution, Two Channels: Coverage-Calibrated KV Retention and Compensation (importance: 35, dev_relevance: 60)
- ULTRADISCOVERY: Abductive Exploration in an Interconnected, Epistemically Open Universe (importance: 15, dev_relevance: 5)
- Distribution Matching Evolutionary Algorithms for Rare Event Sampling (importance: 15, dev_relevance: 15)
- Clean: Second-order LLM Training at Linear Memory Cost via Nystr{\"o}m Sketching (importance: 35, dev_relevance: 45)
- Scaling Verifiable Environments for Long-horizon Work Agents (importance: 35, dev_relevance: 40)
- TempoBridge: Source-Conditioned Flow Matching with Optimal Transport Couplings for Single-Cell Population Transitions (importance: 10, dev_relevance: 5)
- E$^2$-OPSD: Taming Entropy Overshoot in On-Policy Self-Distillation (importance: 25, dev_relevance: 20)
- No Concept Escapes the Audit: Auditing-Aware Unlearning for Verifiable Concept Erasure in Diffusion Models (importance: 15, dev_relevance: 5)
- Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning (importance: 15, dev_relevance: 10)
- CrystalJev: calibrated foundation-model probabilities decide what to compute next in materials discovery (importance: 15, dev_relevance: 5)
- Scale-Invariant Training for Time Series Foundation Models (importance: 20, dev_relevance: 20)
- Recurrent Looped Transformer (importance: 25, dev_relevance: 20)
- SanSi: A Looped Typed Decision Model for System 1.5 Thinking (importance: 20, dev_relevance: 10)
- Secure Speculative Decoding for Large Language Models (importance: 30, dev_relevance: 40)
- GraphOPD: Graph-Augmented On-Policy Distillation for LLM Agents (importance: 25, dev_relevance: 25)
- Mixture of Layers: Dynamic Layer Routing for Visual Reasoning (importance: 15, dev_relevance: 10)
- OmniCam: Omni-Camera Trajectory Generation via Geometry-Grounded Pose Token Learning (importance: 15, dev_relevance: 5)
- Coding-Agent Benchmarks Should Match Their Users' Task Flows (importance: 45, dev_relevance: 85)
- Beyond Policy Support: Interaction Constrained Offline Reinforcement Learning for Autonomous Driving (importance: 15, dev_relevance: 5)
- Why Software Engineering Is Indispensable in the Age of Coding Agents (importance: 45, dev_relevance: 75)
- NavGPT-3: Harnessing Context in a Hierarchical Navigation Runtime (importance: 20, dev_relevance: 30)
- Grammar Concept Annotation at Scale: Deployed Fine-Tuned Small Language Models Outperform Prompted Frontier Models (importance: 15, dev_relevance: 15)
- AffordDrive3D: Affordance-Aware World-Action Modeling with Spatial Understanding (importance: 15, dev_relevance: 5)
- Continuous Ground-Truth Construction and a Recovery Policy for Air--Water Robotic Tracking (importance: 10, dev_relevance: 5)
- REMORY: Learning Residual Memory for Context Compaction (importance: 40, dev_relevance: 55)
- Zatom-2: Multitask Pretraining on Atomistic Data for Generative Modeling across Domains (importance: 15, dev_relevance: 5)
- Constitutional Gating and Deterministic Recovery for Multi-Agent LLM Negotiation: Ablations Against a Stateful Adversarial Gatekeeper (importance: 25, dev_relevance: 40)
- Sera: Semantic Representation Aggregation with Knowledge and LLM for Reliable and Interpretable Battery Health Forecasting (importance: 10, dev_relevance: 5)
- One Skill Too Many: How Co-Installed Skills Conflict in Coding Agents (importance: 45, dev_relevance: 85)
- Stochastic Grouping Conformal Prediction for Effective Subgroup Reliability (importance: 15, dev_relevance: 25)
- HWAM: Humanoid World Action Model With Joint State--Action Generation (importance: 25, dev_relevance: 35)
- Distilling Routed 3D Privilege for Spatial Reasoning in Vision-Language Models (importance: 22, dev_relevance: 30)
- Telling FIBS: Probes Effectively Detect Sabotage and Catch Unverbalized Deception (importance: 35, dev_relevance: 45)
- Beyond Actions: Learning Future Operation Targets for Vision-Language-Action Models (importance: 20, dev_relevance: 30)
- ReSFlow: Rectifying Categorical Flows on Statistical Manifolds for One-Step Generation (importance: 18, dev_relevance: 30)
- Hide and Seek: HIDE Concepts to SEEK explanations for black box models (importance: 18, dev_relevance: 30)
- Lotus: Learning to Understand Time-series by Learning to Forecast (importance: 20, dev_relevance: 30)
- Inverse Reasoning Models for Efficient Post-Training and Verification from Answer-Only Data (importance: 35, dev_relevance: 50)
- Do Vision CIL Methods Transfer to Protein? A Diagnostic Benchmark of Frozen Protein Encoders (importance: 10, dev_relevance: 15)
- Do Perturbation Predictors Learn Gene Interactions? Diagnosing Extrapolation in Agentic Screening (importance: 12, dev_relevance: 15)
- Token Utility for On-Policy Distillation via Local Trust-Region Optimization (importance: 30, dev_relevance: 50)
- PMO-Dock: Benchmarking Docking, Specificity, and Generalization in Molecular Optimization (importance: 10, dev_relevance: 15)
- Data Density Scaling Laws for Image Self-Distillation (importance: 20, dev_relevance: 35)
- Scaling Cheap Synthetic Pretraining Improves Indoor Radio-Map Prediction (importance: 12, dev_relevance: 25)
- IR-PRM+ with Previous-Utility Prediction: Constant Regret and Best-Iterate Convergence (importance: 12, dev_relevance: 25)
- How Many Ranks Does a Token Need? A Noise-Floor Squelch for Rank-One LoRA Experts (importance: 25, dev_relevance: 50)
- Who Should We Listen to More? Welfare-Aware Preference Acquisition for Pluralistic Alignment (importance: 15, dev_relevance: 20)
- Lag: A Mean-Field Optimal Policy for Online Assignment with Reusable Resources (importance: 10, dev_relevance: 20)
- Tracing Morphology, Semantics, and Divergence in Language Acquisition (importance: 15, dev_relevance: 25)
- Loss Curvature as a Sampling-Free Proxy for Memorization in Diffusion Models (importance: 18, dev_relevance: 30)
- Graph Coarsening Enables Escaping Local Optima in Compute-Constrained Nonconvex MAX-CUT (importance: 10, dev_relevance: 25)
- What Information Buys in Active Inference (importance: 10, dev_relevance: 15)
- Deep Generative Models for Phylogenetic Inference with Complex Evolutionary Processes (importance: 10, dev_relevance: 15)
- How Inner Optimizers Shape Replica Scaling (importance: 30, dev_relevance: 50)
- Theoretical and Practical Advances in Hierarchical Bayesian Deep Learning Methodology (importance: 12, dev_relevance: 25)
- Backpropagating Verifier Rewards Through Continuous Flow Language Models (importance: 25, dev_relevance: 40)
- Beyond Aggregate Accuracy: Item-Level Transitions in Prompt Optimization (importance: 25, dev_relevance: 45)
- Contextual Credit Policy Optimization (importance: 25, dev_relevance: 45)
- Genesis Steering: Better Activation Steering Through Training Data (importance: 15, dev_relevance: 30)
- Unlocking Structured Pruning for Reasoning Language Models via Attention-Preserving MLP Trimming (importance: 30, dev_relevance: 45)
- PairIT: Autoregressive Transformers for Low-Data Molecule Optimization (importance: 8, dev_relevance: 15)
- Imagining Goals with Geometric Horizon Models for Policy Generalization (importance: 15, dev_relevance: 20)
- TRACER: Trace-Regularized Training Against Indirect Prompt Injection (importance: 35, dev_relevance: 50)
- MA-Bench: Measuring Operational Planning in Language Models (importance: 12, dev_relevance: 15)
- Towards Training and Inference of Graph Neural Networks in Graph Databases (importance: 20, dev_relevance: 45)
- Demystifying On-Policy Distillation: Where Does the Teacher's Advantage Live? (importance: 25, dev_relevance: 45)
- Diagnosing Low-Confidence Patient Profiles via Localized Harmful Covariate Shift Detection (importance: 8, dev_relevance: 15)
- Adaptive Computation with Conformal Certification (importance: 10, dev_relevance: 20)
- FireGraphNet: Sub-daily wildfire spread prediction with directed graph neural networks (importance: 10, dev_relevance: 15)
- Narrowing the Collaboration Gap, Probably (importance: 25, dev_relevance: 35)
- Riding the Fixed-Point Curve: Image Classification in the Continuous Thought Machine (importance: 12, dev_relevance: 15)
- Spectral-MoE: Heterogeneous Per-Layer Expert Utilization in Dense-to-MoE Conversion (importance: 28, dev_relevance: 45)
- LLMs program using Reusable and Composable Abstract Representations (importance: 30, dev_relevance: 50)
- FinGREAT: Evaluating LLMs on Financial Graph-guided Retrieval And Typing (importance: 12, dev_relevance: 25)
- Refined Topology-Pass Scheduling: Damped Trust-Region Compensation for Dynamic k-NN Graphs in SE(3)-Equivariant Structure Prediction (importance: 8, dev_relevance: 25)
- Beyond Time Induction: Structured Conditional Neural Solvers for Parametric Stochastic-Volatility Pricing (importance: 8, dev_relevance: 15)
- Exploring Interpretability in Deep Learning-Based Forest Classification - Separating Similar Forest Classes using Sentinel-2 Time Series (importance: 8, dev_relevance: 15)
- Outcome-Routed Distillation and Reinforcement Learning for Diffusion Language Models (importance: 28, dev_relevance: 45)
- Dimension-Free Scaling Laws for Size Generalization: An Exponential Separation (importance: 15, dev_relevance: 20)
- Exact Scaling Laws for a Random-Features Bigram Model (importance: 15, dev_relevance: 20)
- Verifier-Grounded Agentic RL for Provably Correct Code (importance: 35, dev_relevance: 65)
- DTDD: Divergence-Triggered Dynamic Distillation for Reliable On-Policy Supervision (importance: 25, dev_relevance: 45)
- Did the LLM Answer the Question? A Semantic Reconstruction Game for Jailbreak Evaluation (importance: 22, dev_relevance: 35)
- Historical-Feature Residual Adaptation for Sequential Screening (importance: 8, dev_relevance: 20)
- Why don't Graph Transformers attend to the same token? (importance: 18, dev_relevance: 30)
- Counterfactual Gate-Support Decomposition: Separating Selection from Redistribution in Training-Data Filtering (importance: 15, dev_relevance: 30)
- Integrated Novelty Detection Systems: From Sensor Networks to Multi-Novelty Computer Vision (importance: 8, dev_relevance: 20)
- Where to Start, How to Move: Analyzing Curricula under GRPO for Long-Horizon LLM Agents (importance: 28, dev_relevance: 40)
- BFHD: Bidirectional Feature Harmonization Decomposition for Heterogeneous Clinical Assessments (importance: 6, dev_relevance: 15)
- Asymptotic Behavior of Transformers (importance: 22, dev_relevance: 35)
- Dynamic Graph World Modelling: Learning Future Dynamics from Temporal Context (importance: 12, dev_relevance: 20)
- Policy-Verified Self-Training: Learning Agent Compliance from Verifiable Policy Rewards (importance: 35, dev_relevance: 55)
- One-Step Prediction Fidelity Is a Poor Proxy for Planning Performance in JEPA World Models (importance: 15, dev_relevance: 25)
- Robust Explanatory Analysis for Structured Optimization Networks (REASON) in Learning (importance: 10, dev_relevance: 20)
- RA$^2$: Retain-Anchored Attraction Preserves Forget-Adjacent Utility in LLM Unlearning (importance: 22, dev_relevance: 30)
- Fully sequential Gaussian process regression for preference learning (importance: 10, dev_relevance: 15)
- Selective Next-Location Prediction with MetaModelSelect: A Lightweight Post-Hoc Confidence Metamodel (importance: 8, dev_relevance: 15)
- Train to Stitch: Topology-Preserving Regularization for Interoperable Representations (importance: 15, dev_relevance: 25)
- W-JEPA: Wavelet Joint-Embedding Predictive Architecture for Wearable Activity Recognition (importance: 8, dev_relevance: 15)
- Comprehensive Comparisons of Reinforcement Learning Algorithms in Autonomous Driving (importance: 10, dev_relevance: 15)
- Progress-Aware Distillation for Mitigating Stagnation in Small Language Model Agents (importance: 28, dev_relevance: 45)
- MIPS: Mitigating Mode Collapse in Reinforcement Learning for Large Outcome Spaces (importance: 22, dev_relevance: 35)
- SKG: A Law-Induced Engine from Inference to Induction (importance: 8, dev_relevance: 10)
- CktHGVAE: A Symmetric Design/State Heterogeneous Graph Variational Autoencoder for Transistor-level Analog/Mixed-signal Circuit Design and Analysis (importance: 8, dev_relevance: 15)
- HankelKANet: Lightweight Path Loss Estimation via Physics-Inspired Hankel Convolution (importance: 8, dev_relevance: 15)
- Diagnosing and Correcting Reward Hacking in Verifiable Instruction Following (importance: 30, dev_relevance: 40)
- Two-stage Residual Learning for Modelling Strength-Informed Optimisation of Thermal Systems (importance: 6, dev_relevance: 15)
- Test-Time Forecasting: Recurrent Depth Enables In-Context Dynamics Prediction (importance: 18, dev_relevance: 25)
- TEX-MEX: Similarity-Guided Temporal Contrastive Learning for Time Series (importance: 12, dev_relevance: 20)
- Learning Across Temporal Graph Representation Silos (importance: 12, dev_relevance: 20)
- InstructPro2: Natural Language and Biological Grounding Guided De Novo Protein Design (importance: 8, dev_relevance: 15)
- Beyond Static Positional Encoding: Bayesian Time-aware Attention for Non-Stationary Long-term Time Series Forecasting (importance: 12, dev_relevance: 20)
- PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference (importance: 25, dev_relevance: 50)
- Divergent Paths: Learning an Outcome-Predictive Representation of LLM Agent Trajectories (importance: 25, dev_relevance: 45)
- Sparse Autoencoders Are a Shared Coordinate System for Symbolic Matching (importance: 15, dev_relevance: 25)
- Synthetic reaction data from transformation rules for graph-to-sequence pretraining in biodegradation pathway prediction (importance: 8, dev_relevance: 15)
- On Global Joint Lower Bounds in Zeroth-Order Nonconvex Optimization (importance: 10, dev_relevance: 20)
- Enhancing Latent World Models with Associative Memory (importance: 15, dev_relevance: 20)
- Beyond Distribution Matching: Selective Margin Preservation for Supervised Fine-Tuning (importance: 18, dev_relevance: 35)
- Function-Space Temporal Self-Distillation for Policy Optimization (importance: 25, dev_relevance: 40)
- Annealed L\'{e}vy Diffusion via Gaussian Surrogate Reverse Dynamics (importance: 10, dev_relevance: 15)
- ReGRA: Residual-Gradient Adaptation for Complementary Low-Rank Fine-Tuning (importance: 20, dev_relevance: 40)
- New Model, Old Habits: Behavior Transfer under Model Swapping in Multi-Session LLM Agents (importance: 25, dev_relevance: 45)
- TIMERSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models (importance: 20, dev_relevance: 40)
- What Happens to the Goal? Distinct Failures of Policy Alignment in Goal Misgeneralization (importance: 15, dev_relevance: 20)
- ProtModernBERT: Compute-Efficient Protein Language Modeling (importance: 10, dev_relevance: 25)
- CAST: History-Adaptive Spectral Clipping for LLM Pretraining (importance: 20, dev_relevance: 40)
- Learning to Compute Is Not Learning to Retain or Reuse in Looped Transformers (importance: 15, dev_relevance: 25)
- From Global to Tail Convergence: Dynamics of $\alpha$-Stable Score-Based Generative Models (importance: 10, dev_relevance: 15)
- Learning Who Benefits Most: Multi-Treatment Causal Ranking for Population Health Stratification (importance: 20, dev_relevance: 5)
- Reinforcement Learning via Advantage Guidance (importance: 35, dev_relevance: 20)
- Asymptotically Exact Negative Guidance of Diffusion Models via Positive-Unlabeled Learning (importance: 25, dev_relevance: 5)
- Efficient contextual partial monitoring with general function approximation (importance: 20, dev_relevance: 5)
- VaryST: Generalizable Spatio-Temporal Forecasting Beyond the Closed-World Setting (importance: 20, dev_relevance: 5)
- When Does Depth Help? Understanding Layerwise Inference in Tabular Foundation Models (importance: 30, dev_relevance: 10)
- MAPLE: A Multi-Axis Parametric Knowledge Extraction Lens for LLM Evaluation (importance: 30, dev_relevance: 15)
- CTIA-ODE: Continuous-Time Inter-Variable Attention for Long-Term Time Series Forecasting (importance: 15, dev_relevance: 5)
- Rent-Seeking Language Models: Reward Models Set the Prices, Frequent Audits Curb the Rent (importance: 30, dev_relevance: 15)
- Memory-Efficient Subspace Optimization with Residual Correction: From Momentum SGD to Muon (importance: 35, dev_relevance: 25)
- Energy Consumption Analysis of Discrete Diffusion and Autoregressive Language Models (importance: 30, dev_relevance: 20)
- SPARS: A Stability-Plasticity-Aware Region Selection Framework for Continual Learning (importance: 20, dev_relevance: 5)
- Neural Wasserstein Splines: Representation-Invariant Second-Order Priors for Trajectory Inference (importance: 15, dev_relevance: 5)
- Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction (importance: 25, dev_relevance: 10)
- On the Limits of Representation-Based Test-Time Adaptation for Multimodal LLMs (importance: 25, dev_relevance: 5)
- Universal Sample-Based Embeddings of Higher-Order Objects (importance: 20, dev_relevance: 5)
- When Non-Adversarial Noise Hurts Learning: Benchmarking Skill Self-Evolution in Agents (importance: 45, dev_relevance: 50)
- CARD: Closing the Teacher-Student Loop for Adversarial Robustness Distillation (importance: 20, dev_relevance: 5)
- Beyond DFT electronic band structures via dual-space orbital learning (importance: 15, dev_relevance: 3)
- MolHop: Scaffold-Hopping Molecular Design with Agentic Reinforcement Learning (importance: 20, dev_relevance: 5)
- MorphingNet: Symbolic Regression Where the Network Is the Formula (importance: 20, dev_relevance: 5)
- Learning Graph Energies for Finite-Time Generation (importance: 15, dev_relevance: 5)
- Safe-FCL: Continual Risk Awareness and Fine-Grained Safety Aggregation for Federated Continual Learning of LLMs (importance: 25, dev_relevance: 10)
- LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation (importance: 25, dev_relevance: 10)
- Empirical Boundaries of Reservoir Computing for Few-Shot Sequence Pattern Tasks (importance: 15, dev_relevance: 5)
- Provable Transformer Context Compression for In-Context Reinforcement Learning (importance: 20, dev_relevance: 5)
- Outperforming 1000 Euclidean Layers With Hyperbolic Goal-Conditioned RL (importance: 15, dev_relevance: 5)
- ChemArgus: Automated Scientific Diagnostics for Lab-Grade Multimodal Chemistry Reasoning: A Frontier Benchmark (importance: 20, dev_relevance: 3)
- On the Convergence of Semi-Gradient TD: A Mean Field Perspective (importance: 15, dev_relevance: 5)
- Sequential Learning-Based Physics-Informed Neural Network Modeling for Parameter Estimation on a Power Converter (importance: 15, dev_relevance: 10)
- GBTSVR: Twin Support Vector Regression via Granular Ball Computing (importance: 10, dev_relevance: 5)
- Trainable at Depth, Solvable at Small Size: What the Loss Surface of a Polynomial Activation Dictates for Designing, Initializing and Training Networks (importance: 25, dev_relevance: 10)
- Rethinking Prediction Horizons: Decoupling Model Queries from State Updates in Few-Step MeanFlow (importance: 20, dev_relevance: 5)
- A Large-Dimensional Analysis of Least-Squares Federated Learning via Generalized Total Variation Minimization (importance: 10, dev_relevance: 5)
- What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books (importance: 20, dev_relevance: 5)
- Exhaustive Canonical Skeletons Generation: A Formal Dataset for Decomposable Symbolic Regression (importance: 15, dev_relevance: 10)
- From Spectral to Frobenius: When Muon Loses Its Spectral Implicit Bias (importance: 30, dev_relevance: 15)
- KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks (importance: 15, dev_relevance: 15)
- A Bregman Proximal Viewpoint on Neural Operators (importance: 15, dev_relevance: 5)
- Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms (importance: 25, dev_relevance: 25)
- In-Context Batch Black-box Optimization via Flow Matching (importance: 20, dev_relevance: 5)
- Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming (importance: 15, dev_relevance: 5)
- RM-OPD: On-Policy Distillation for Generative Reward Models (importance: 30, dev_relevance: 15)
- Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning (importance: 10, dev_relevance: 0)
- HGRP: Heterogeneous Graph Neural Network for Routability Prediction (importance: 15, dev_relevance: 25)
- MAEM: Mitigating Recency Bias via a Replay-Free, Capacity-Controlled Dynamic Mixture of Experts for Class-Incremental Graph Learning (importance: 15, dev_relevance: 5)
- Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection (importance: 15, dev_relevance: 10)
- Adversarial Reinforcement Learning for Robust Planetary Landing (importance: 20, dev_relevance: 10)
- Long-Horizon RL Needs Muon (importance: 40, dev_relevance: 30)
- Harness Policy Distillation: Aligning Agent Policies with Runtime Semantics (importance: 40, dev_relevance: 60)
- FUSE: Fairness-aware Update Screening against Byzantine Exploitation in $q$-Fair Federated Learning (importance: 10, dev_relevance: 5)
- Resolution Allocation with Learnable Resizing and Spectral Filter Parameterization (importance: 10, dev_relevance: 5)
- TANGO: RNA Topology and Geometry Co-Design (importance: 10, dev_relevance: 2)
- From Plato to Aristotle: Toward a General Theory of Representation Hypotheses (importance: 25, dev_relevance: 10)
- Predicting Unseen Scientific Interactions: Conditional Flow Models Hallucinate Non-Additive Effects (importance: 20, dev_relevance: 5)
- SpectrumQuant: Spectrum Alignment for LLM Quantization (importance: 35, dev_relevance: 40)
- Comparative Evaluation of Deep-Learning based DoA Estimators for Real-Time ISAC Beamforming (importance: 10, dev_relevance: 5)
- Are We Overthinking One-Step Generation? (importance: 20, dev_relevance: 5)
- Training Dynamics of Transformers on Unbounded Inputs: Attention Learning Can Lag Output Layer Learning (importance: 20, dev_relevance: 10)
- When Does Task-Aware Imputation Matter? An Analysis of Robustness Under Train-Test Mismatch (importance: 15, dev_relevance: 10)
- Non-asymptotic Convergence of Average-reward Q-learning with Options (importance: 10, dev_relevance: 5)
- NeuroDETR: Multiscale Discrete Spatiotemporal Tokenization for Asynchronous EEG Event Detection (importance: 10, dev_relevance: 3)
- Beyond Expert Rankings: Contrast Calibration for Mixed-Precision MoE Quantization (importance: 30, dev_relevance: 35)
- Fractional is Better: Learnable Derivative Orders in Neural Operator Learning (importance: 15, dev_relevance: 5)
- ReAl: Refine neural operators via aligning dynamical flow (importance: 15, dev_relevance: 5)
- From Human Corrections to Learned Segmentations: A Critical Mathematical Review of Information, Learning, and Human Effort (importance: 15, dev_relevance: 5)
- Training Language Models to Fail Gracefully (importance: 40, dev_relevance: 40)
- Hyperspherical Contrastive Representations Unlock Planning and Exploration (importance: 15, dev_relevance: 5)
- When Does Momentum Help in Reinforcement Learning with Verifiable Rewards? (importance: 30, dev_relevance: 20)
- One Size Does Not Fit All: Curvature-Aware Adaptive Perturbation for SAM (importance: 15, dev_relevance: 5)
- Tracking a Small Set of Experts in Online Convex Optimization (importance: 10, dev_relevance: 5)
- OrionMoE-v1.0: A Mixture-of-Experts Foundation Model for Tabular In-Context Learning (importance: 25, dev_relevance: 10)
- Continual Learning for Reliable In-Sequence Molecular Communication (importance: 10, dev_relevance: 2)
- The Accuracy Well: Nonmonotone Accuracy Dynamics at Vanishing Initialization (importance: 20, dev_relevance: 5)
- Learning by Examples: Power Grid Steady State Analysis with a Tabular Foundation Model (importance: 15, dev_relevance: 5)
- Selective Growing Q-Networks (SGQN): An Analysis of Adaptive Action Grids in Decoupled Q-Learning (importance: 10, dev_relevance: 5)
- Decoupled Marginal Sharpening for Training-Free Inference-Time Scaling (importance: 35, dev_relevance: 25)
- Average-Reward Policy Identification from a Single Continuing Trajectory (importance: 10, dev_relevance: 5)
- Hindsight-Privileged RLVR: Single-Rollout Token Credit Without Group Sampling (importance: 35, dev_relevance: 25)
- Graph-Enhanced Disease Trajectory Discovery and Temporal Endpoint Forecasting from Longitudinal Electronic Health Records (importance: 10, dev_relevance: 2)
- A Two-Level Clinical Novelty Prioritization Framework for Disease Trajectories Discovered from Longitudinal Electronic Health Records (importance: 8, dev_relevance: 2)
- Bayesian Learning to Select Fair Classifiers (importance: 15, dev_relevance: 5)
- The Physics Is the Network: Fast, Training-Free Neural Emulation (importance: 15, dev_relevance: 10)
- Generative Representation Learning for Scalable Latent Tokenization (importance: 25, dev_relevance: 15)
- Revisiting Simple Baselines for Relational Deep Learning (importance: 15, dev_relevance: 10)
- Detection of Potato Plant Leaf Diseases using Convolutional Neural Networks (importance: 8, dev_relevance: 2)
- Recovering Reliable Supervision via Critique-Guided On-Policy Distillation (importance: 30, dev_relevance: 15)
- One Task, Many Interfaces: Simple Cross-Interface Consistency Training Improves LLM Agent Robustness (importance: 35, dev_relevance: 35)
- Learnable Hodge-Laplacian Driven Mixture of Experts for Steady-PDE Surrogates on Coarse Unstructured Meshes (importance: 15, dev_relevance: 5)
- Continuum Attention (importance: 20, dev_relevance: 10)
- Your Transformer Is Secretly Elastic (importance: 25, dev_relevance: 15)
- Edgewise Envelopes Between Balanced Forman and Ollivier-Ricci Curvature (importance: 5, dev_relevance: 0)
- Following the Flow the Law Requires: Training Voice Agents to Transfer to Unseen Regulated Procedures in Debt Collection (importance: 30, dev_relevance: 20)
- Physics-Guided Multi-Fidelity Surrogate Modeling for Natural Supercavity Geometry of an Axisymmetric Conical Cavitator (importance: 8, dev_relevance: 3)
- Bayesian Causal Forests for Zero-Inflated Continuous Outcomes (importance: 10, dev_relevance: 3)
- AI-Driven, Timing-Aware Power Optimization (importance: 20, dev_relevance: 20)
- Distinguishing Process Change, Instrument Fault and Data-Path Corruption in Streaming Telemetry: A Reference Implementation and Benchmark (importance: 15, dev_relevance: 15)
- Popularity Trajectories and Mixing Patterns of Urban Venues in Two European Cities (importance: 8, dev_relevance: 5)
- TorchSpin: Differentiable Electron Paramagnetic Resonance Simulation in PyTorch (importance: 20, dev_relevance: 40)
- The Agentic Scientist: Orchestrating Coding Agents for Automated Scientific (Co-)Discovery (importance: 35, dev_relevance: 40)
- An Initial Study of Physiological Markers for Deepfake Speech Detection (importance: 30, dev_relevance: 15)
- Delay Complexity Dimension: Learning with Delayed and Censored Labels (importance: 25, dev_relevance: 10)
- Beyond Equation Search: Conjecture-Guided Discovery with SCOPE and ClosureBench (importance: 35, dev_relevance: 10)
- Wong--Zakai Convergence and FNO Learning for Stochastic Volterra Equations (importance: 15, dev_relevance: 5)
- Semi-Amortized Bayesian Inference with Reusable Local Posteriors (importance: 25, dev_relevance: 15)
- Physical Layer Anomaly Detection via Adaptive Gaussian Mixture Model (importance: 15, dev_relevance: 10)
- Limits of Statistical Assurance for Embodied and Agentic AI under Adaptive Adversaries (importance: 40, dev_relevance: 25)
- Virtual-Alloy Physics-Aware Semantic System for extrapolative prediction of perovskite solar cells (importance: 15, dev_relevance: 5)
- Discovering Undisclosed Concepts in LoRA Adapters for Diffusion-Based AIGC Models (importance: 30, dev_relevance: 20)
- Discrete Diffusion Models are Minimax Optimal Under Sparsity (importance: 30, dev_relevance: 10)
- How Far Can Coding Agents Go in Visual Generation? (importance: 50, dev_relevance: 60)
- Optimization for One, Feasibility for Most: Decoupling Optimization from Feasibility in Chance-Constrained Programs (importance: 15, dev_relevance: 10)
- Smart Alone, Blind Together: Why Accuracy Fails to CertifyTheory of Mind in Multi-Agent Teams (importance: 30, dev_relevance: 25)
- Conformalized Physics-Informed Weak Regression for Parameter Estimation in Epidemic Models (importance: 12, dev_relevance: 5)
- Revisiting Generative vs. Dense Retrieval: A Static Value Prior Reverses the Gap (importance: 25, dev_relevance: 20)
- Similar Models, Broken Odds: Where Luce's Axiom Fails in Chatbot Arena (importance: 40, dev_relevance: 15)
- Neural Supervised Clustering for Learning Individualized Treatment Rules with Many Treatments (importance: 12, dev_relevance: 5)
- From Copying to Composition: How Language Models Develop In-Context Learning (importance: 40, dev_relevance: 25)
- Student Copula Markov Networks (importance: 10, dev_relevance: 5)
- Online Tokenization with Optimal Delay in Linear Time (importance: 40, dev_relevance: 55)
- Into the Unknown: Benchmarking AI agents on open scientific research problems (importance: 45, dev_relevance: 25)
- Learning in Strategic Queuing with Geometric Recharge: From Bandits to Systems (importance: 12, dev_relevance: 15)
- Structure-aware Bayesian full waveform inversion via diffusion-regularized Stein variational gradient descent (importance: 10, dev_relevance: 5)
- Embedding Geometry Shapes LLM Routing: Tracing Failure Modes and Improving Generalization (importance: 35, dev_relevance: 35)
- Generalizing Identifiability Results in Causal Discovery from Complete Data to Missing Data Settings (importance: 12, dev_relevance: 10)
- From Detection To Anticipation: Continuous Onset Forecasting For Stuttered Speech (importance: 15, dev_relevance: 10)
- Multi-Sensitive Distributional Fairness: Joint Wasserstein Audits and Shapley Allocation (importance: 12, dev_relevance: 5)
- TopicMerge: A Divide and Conquer Approach to Topic Modeling (importance: 15, dev_relevance: 10)
- Target-Token Asymmetry Can Masquerade As Depth Specialization (importance: 30, dev_relevance: 15)
- Steering Vectors for Competing Ethical Frameworks Are Interchangeable: A Competing-Target Test for Activation Steering (importance: 25, dev_relevance: 15)
- Tail Competition Explains Scaling in Best-of-N Sampling for Verifiable Problems (importance: 35, dev_relevance: 30)
- The Impact of Detector Architecture, Model Scale, and Scene Density on Multi-Pedestrian Tracking (importance: 12, dev_relevance: 10)
- ActMOF: Mapping Reticular Synthesis through Active Learning (importance: 10, dev_relevance: 3)
- Scaling LLM Inference via Parallelism: Throughput-Latency Trade-off (importance: 40, dev_relevance: 40)
- Learning to Select Features with Manifold Deep Unfolding PCA (importance: 8, dev_relevance: 5)
- Large-Step Last-Iterate Convergence of Random-Reshuffling Stochastic Gradient Descent (importance: 25, dev_relevance: 15)
- State-Matched Sequence Conformal Prediction (importance: 12, dev_relevance: 8)
- Digital Self-Interference Cancellation for In-Band Full-Duplex Systems Based on Transformer Neural Operators (importance: 8, dev_relevance: 5)
- What You Sign Is Not What You Run: The Verifiability Gap in Quantized LLM Supply Chains (importance: 45, dev_relevance: 50)
- High-Probability Regret Bounds for Feel-Good Thompson Sampling (importance: 12, dev_relevance: 5)
- A Unified Response Framework for Questions and Tutoring Dialogues (importance: 10, dev_relevance: 5)
- Typed Decision Models on Tabular Data (importance: 25, dev_relevance: 25)
- Noisy-Label Calibration with Multiple Unreliable Labels (importance: 12, dev_relevance: 5)
- Evidence-Anchored Overdose Death Prediction from Death Investigation Reports (importance: 10, dev_relevance: 3)
- INTACT: Inducing Latent Reasoning on a Frozen Backbone (importance: 35, dev_relevance: 20)
- Geometric Kernel Diversity: An Axiomatic Diversity Measure in Euclidean Space (importance: 10, dev_relevance: 5)
- On Inherent Privacy of Posterior Sampling: A Unified R\'enyi-Divergence Framework (importance: 12, dev_relevance: 5)
- Bridging the Topological Divide: Dual-Branch Graph Distillation with Semantic Priors for Robust Cold-Start Recommendation (importance: 12, dev_relevance: 5)
- Who's Behind the Text? Robust Model Authorship Attribution (importance: 30, dev_relevance: 20)
- Projected Wasserstein-Fisher-Rao Flows for Hawkes Background-Intensity Estimation (importance: 8, dev_relevance: 3)
- L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents (importance: 50, dev_relevance: 60)
- A Graph-Based Analysis of Comparability, Scale, and Attribution in Private Transformer Inference (importance: 20, dev_relevance: 15)
- DAGFormer: Global Topology Prediction for Transformer Cross-Layer Connections (importance: 35, dev_relevance: 15)
- Self-Conditioning Enables Deep Sequential Reasoning in Flow Matching (importance: 20, dev_relevance: 5)
- Risk-Controlled DFT Selection for Machine-Learned Interatomic Potentials (importance: 10, dev_relevance: 5)
- Model What Matters: Decision-Guided Selective World Modeling for Language Agents (importance: 30, dev_relevance: 25)
- HACHIMI-II: From Role-Play to Learner Behavior Modeling for Faithful Student Simulation (importance: 15, dev_relevance: 10)
- MCAIR: Monitoring and Correcting Agent Unfaithfulness via Internal Representations (importance: 35, dev_relevance: 35)
- Learning Individual Scientific Taste with Privileged Hindsight Distillation (importance: 15, dev_relevance: 5)
- Articulation Upper Bounds for Language Models (importance: 20, dev_relevance: 5)
- Discriminative Score Function: Turning Pretrained Models into Functional Generative Priors (importance: 15, dev_relevance: 5)
- Analyzing and Mitigating Poisoning Attacks on Flow-Based Intrusion Detection (importance: 20, dev_relevance: 10)
- When Does the Dependency Graph Help Code-to-Architecture Mapping? (importance: 25, dev_relevance: 65)
- SNACK: A Sequential Notation Framework for Probabilistic Graph Generation (importance: 10, dev_relevance: 5)
- More Methods, Better Agents? A Reality Check on Self-Distillation in Agentic Reinforcement Learning (importance: 35, dev_relevance: 30)
- StateCraft: Augmenting Video World Models with Abstract-State Reasoning (importance: 15, dev_relevance: 5)
- ReproBAIT: Can coding agents build their own biological AI tools? (importance: 35, dev_relevance: 40)
- Calibration and Risk-Budgeted Decisions with System One Models (importance: 20, dev_relevance: 15)
- JointServe: Serving Multi-Model Agents on Memory-Constrained GPUs (importance: 35, dev_relevance: 40)
- Text-Informed Training Data Filtering for Enhanced Virtual Flow Meter Performance (importance: 8, dev_relevance: 5)
- On the Optimisation Landscape of Overparameterised Weighted MMD (importance: 10, dev_relevance: 5)
- Explorer-Zero: Improving Self-Play for Long-Horizon Tasks with Decomposable Verifiable Rewards (importance: 30, dev_relevance: 20)
- Few-Shot Adaptation for Anomalous Sound Detection at Low False-Positive Rate (importance: 10, dev_relevance: 8)
- Multimodal Medical Image Parsing with Task-Specific Low-Rank Adapters (importance: 15, dev_relevance: 15)
- Convergence Rates of Gaussian Process Thompson Sampling for Efficient Global Optimization (importance: 8, dev_relevance: 3)
- SAM3RL: Memory Control via Reinforcement Learning for Visual Object Tracking (importance: 20, dev_relevance: 10)
- AtomSearch: Evidence-Grounded Tree Search for Agent Skill Optimization (importance: 40, dev_relevance: 45)
- Choosing the Number of Attention Heads from Intrinsic Dimension: A Small-Scale Study (importance: 20, dev_relevance: 25)
- Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds (importance: 8, dev_relevance: 3)
- Weighted Adjoint Matching: Finetuning Flow Matching Models with Time-Dependent Control Cost (importance: 15, dev_relevance: 5)
- Gap Detection: Mitigating Homogenization through Demand and Defensibility (importance: 15, dev_relevance: 10)
- Cross-Model In-Context Neurofeedback Reveals (Limited) Activation Control in LLMs (importance: 25, dev_relevance: 10)
- Investigating Behavior Cloning from Few Demonstrations for Autonomous Driving based on Bird's-Eye View in Simulated Cities (importance: 12, dev_relevance: 5)
- Continual pretraining Prostate MRI ViT Encoder for task alignment (importance: 10, dev_relevance: 5)
- Mixed Poisson diffusion models and application to posterior sampling (importance: 10, dev_relevance: 5)
- The minimal width of neural networks and immersions of real projective space (importance: 8, dev_relevance: 3)
- Scaling Multilingual Mixture-of-Experts: How Language Coverage Changes Capacity and Compute Allocation (importance: 35, dev_relevance: 20)
- Contrastive uncertainty learning for robust multi-structure segmentation in lower-pelvic MR (importance: 8, dev_relevance: 3)
- FlowSep-RF: Rectified Flow Matching for Single-Channel Radio-Frequency Signal Separation (importance: 8, dev_relevance: 3)
- CASI: Credit Assignment via Set Intervention for Agentic Retrieval (importance: 25, dev_relevance: 20)
- MMLA-RSI: Recursive Self-Improvement through RTT and RTMA (importance: 30, dev_relevance: 20)
- Extreme-Scale Super-Resolution of Homogeneous Turbulence from a Single Realization (importance: 8, dev_relevance: 3)
- Efficient Test-Time Adaptation For Robot Policies (importance: 15, dev_relevance: 8)
- Lifted State Policy Optimization: Internalizing Self-Correction through Transition Geometry (importance: 25, dev_relevance: 15)
- Hypothesis Selection via Weighted Medians (importance: 6, dev_relevance: 3)
- FOLD: Position-Free Far Memory with Local Order for Long-Context Speculative Decoding (importance: 25, dev_relevance: 30)
- Cortex: Consolidating Context into Parametric Memory (importance: 25, dev_relevance: 20)
- HRTF Upsampling and Personalization via Neural Operators (importance: 8, dev_relevance: 3)
- Posterior Frequencies and the Cost of Not Fitting: A Correction to a Memorization Lower Bound (importance: 15, dev_relevance: 5)
- What Does Disagreement Mean? A Semantic Identifiability Frontier for Pairwise Preference Learning (importance: 20, dev_relevance: 10)
- The COUNT: Counting N-grams Unlocks Scaling on Host Memory and Training-Free Adaptation (importance: 30, dev_relevance: 35)
- SafeAdapt-Poker: Counterfactual Risk Calibration for Budgeted Adaptation under Opponent Shift (importance: 15, dev_relevance: 25)
- Lost in Translation: Why Lower Bias Scores Don't Mean Fairer Models (importance: 20, dev_relevance: 20)
- VascuSR: Iterative Anatomy-Aware Diffusion for CT Super-Resolution (importance: 15, dev_relevance: 15)
- VASPO: Self-Distillation with Privileged Value Estimation (importance: 30, dev_relevance: 45)
- ActiveEvidence: Decision-Focused Evidence Graphs with Verifiable Acquisition Gaps (importance: 25, dev_relevance: 35)
- PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis (importance: 35, dev_relevance: 45)
- Latent Compaction with Test-Time Training (importance: 30, dev_relevance: 50)
- Price of Anarchy of First-Price Auctions in Agentic Markets (importance: 20, dev_relevance: 30)
- Co-Evolving Acting and Forecasting for Language Agents (importance: 30, dev_relevance: 45)
- Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion (importance: 15, dev_relevance: 15)
- Hierarchical Residual Policy Learning for Real-World Mobile Manipulation with Sparse Human Guidance (importance: 15, dev_relevance: 10)
- Generation of data on discontinuous manifolds via continuous stochastic non-invertible networks (importance: 8, dev_relevance: 15)
- Bellman Conformal Inference: Calibrating Prediction Intervals For Time Series (importance: 10, dev_relevance: 20)
- GLL: A Differentiable Graph Learning Layer for Neural Networks (importance: 8, dev_relevance: 20)
- Forget About It: Mitigating Forgetting Without Replay or Auxiliary Objectives (importance: 15, dev_relevance: 35)
- Spectral structure learning for clinical time series (importance: 8, dev_relevance: 10)
- Trustworthy AI Must Account for Interactions (importance: 12, dev_relevance: 20)
- Rule-based and reinforcement-learning policies for resource allocation in a pediatric echo lab: a simulation study (importance: 5, dev_relevance: 10)
- The Invisible Leash: Why RLVR May or May Not Escape Its Origin (importance: 20, dev_relevance: 40)
- Sig-DEG for Distillation: Making Diffusion Models Faster and Lighter (importance: 15, dev_relevance: 25)
- Context Value Informed In-Context Reinforcement Learning (importance: 15, dev_relevance: 25)
- A Byzantine-Robust Federated Learning with Nesterov Momentum (importance: 10, dev_relevance: 15)
- Ground-Truth Subgraphs for Better Training and Evaluation of Knowledge Graph Augmented LLMs (importance: 18, dev_relevance: 35)
- A Versatile Variational Quantum Kernel Framework for Non-Trivial Classification (importance: 8, dev_relevance: 15)
- The Instability of Safety: How Random Seeds and Temperature Expose Inconsistent LLM Refusal Behavior (importance: 35, dev_relevance: 35)
- Long-Range Distillation: Distilling 10,000 Years of Simulated Climate into Long Timestep AI Weather Models (importance: 15, dev_relevance: 10)
- Tree-Structured Synergy of Large Language Models and Bayesian Optimization for Efficient CASH (importance: 12, dev_relevance: 30)
- Interpolation of GEDI Biomass Estimates with Calibrated Uncertainty Quantification (importance: 8, dev_relevance: 10)
- SWE-Spot: Building Small Repo-Experts with Repository-Centric Learning (importance: 40, dev_relevance: 70)
- Deep Time-series Forecasting Needs Kernelized Moment Balancing (importance: 10, dev_relevance: 15)
- Rational Transductors (importance: 15, dev_relevance: 20)
- Bridging Academia and Industry: A Comprehensive Benchmark for Attributed Graph Clustering (importance: 8, dev_relevance: 15)
- ANCRe: Adaptive Neural Connection Reassignment for Efficient Depth Scaling (importance: 20, dev_relevance: 30)
- Parameter-Efficient Temporal Domain Generalization (importance: 10, dev_relevance: 20)
- Task-Driven Subspace Decomposition for Knowledge Sharing and Isolation in LoRA-based Continual Learning (importance: 18, dev_relevance: 35)
- MARS: Harmonizing Multimodal Convergence via Adaptive Rank Search (importance: 18, dev_relevance: 35)
- NePPO: Near-Potential Policy Optimization for General-Sum Multi-Agent Reinforcement Learning (importance: 12, dev_relevance: 15)
- Maximum-Entropy Exploration with Future State-Action Visitation Measures (importance: 10, dev_relevance: 15)
- Constitutive parameterized deep energy method for solid mechanics problems with random material parameters (importance: 8, dev_relevance: 8)
- Structural Pass Analysis in Football: Learning Pass Archetypes and Tactical Impact from Spatio-Temporal Tracking Data (importance: 5, dev_relevance: 5)
- Causally-Constrained Probabilistic Forecasting for Time-Series Anomaly Detection (importance: 12, dev_relevance: 25)
- Medical Imaging Classification with Cold-Atom Reservoir Computing using Auto-Encoders and Surrogate-Driven Training (importance: 6, dev_relevance: 8)
- From Causal Graphs to Markovian States: Multi-Order State Exposure for Deep RL (importance: 12, dev_relevance: 20)
- Generative Actor-Critic with Soft Bridge Policies (importance: 12, dev_relevance: 15)
- TabFORGE: A Tabular Foundation Model for Generative Modelling (importance: 20, dev_relevance: 25)
- TIDES: Implicit Time-Awareness in Selective State Space Models (importance: 12, dev_relevance: 15)
- LEAP: Unlocking dLLM Parallelism via Lookahead Early-Convergence Token Detection (importance: 25, dev_relevance: 40)
- Split the Differences, Pool the Rest: Provably Efficient Multi-Objective Imitation (importance: 10, dev_relevance: 12)
- Learning, Fast and Slow: Towards LLMs That Adapt Continually (importance: 35, dev_relevance: 40)
- GQLA: Group-Query Latent Attention for Hardware-Adaptive Decoding (importance: 25, dev_relevance: 50)
- Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting (importance: 8, dev_relevance: 8)
- When and Why Adversarial Training Improves PINNs: A Neural Tangent Kernel Perspective (importance: 8, dev_relevance: 10)
- BrainDyn: A Sheaf Neural ODE for Forecasting and Generating Brain Dynamics (importance: 8, dev_relevance: 8)
- OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond (importance: 30, dev_relevance: 55)
- A Simple Plug-in for Improving Eviction-Based KV Cache Compression (importance: 28, dev_relevance: 50)
- Unification and Optimization of Robust Supervised Learning (importance: 12, dev_relevance: 15)
- Subspace-Decomposed JEPAs: Disentangling Progression and Content in Latent World Models (importance: 12, dev_relevance: 12)
- HARP: Efficient Data Selection for Finetuning Large Language Models (importance: 25, dev_relevance: 45)
- Activation Steering Induces Emergent Misalignment: Steering Vector Construction and Comprehensive Evaluation (importance: 35, dev_relevance: 35)
- AgentWorld: A Data Factory for Training LLM Agents from Multi-Turn Interactions (importance: 30, dev_relevance: 50)
- A Unifying Framework for Concept-Based Representational Similarity (importance: 10, dev_relevance: 10)
- Geological text descriptions in ill-posed inverse problems: insights from learned hydraulic-conductivity inversion (importance: 6, dev_relevance: 8)
- Blackknife: Query-Free Black-Box Rewiring Attacks on Heterogeneous Graph Neural Networks (importance: 10, dev_relevance: 15)
- Platonic Projection Structures: Operator-Induced Observability in Representation Learning (importance: 8, dev_relevance: 8)
- What Does a Discrete Diffusion Model Learn? One Reverse Process in Three Coordinates (importance: 15, dev_relevance: 15)
- RhyMix: A Lightweight Adaptive Multi-Rhythm Network for Long-Term Time Series Forecasting (importance: 10, dev_relevance: 15)
- Correlation-Aware Contextual Bandits with Surrogate Rewards for LLM Routing (importance: 25, dev_relevance: 40)
- Does Depth Survive Composition, and Can a Router Use It? Adaptive Compute in Latent World Models (importance: 10, dev_relevance: 15)
- tidyHEBO: Improving the Robustness of Heteroskedastic Evolutionary Bayesian Optimization (importance: 8, dev_relevance: 8)
- AdaPCLA: Adaptive Prior-Calibrated Logit Adjustment for Long-Tailed Longitudinal EHR Generation (importance: 6, dev_relevance: 8)
- Gauge-Invariant, Parameter-Insensitive Regularization for Potential Recovery from Flow on Directed Graphs (importance: 6, dev_relevance: 8)
- Inpainting Insights: Elevating Visual XAI with Photorealistic Perturbations (importance: 8, dev_relevance: 12)
- SPARK: Learning from Verification in Speculative GRPO Rollouts (importance: 35, dev_relevance: 60)
- Comparing SGLD and a fixed-noise predictor-corrector adaptation in canonical joint energy-based models on CIFAR-10 (importance: 6, dev_relevance: 8)
- Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL (importance: 25, dev_relevance: 35)
- Uncertainty-Aware Probabilistic Constrained Clustering from Entangled Pairwise Supervision (importance: 6, dev_relevance: 8)
- Wrong Operator or Blind Design? A Post-Training Diagnostic for Physics-Informed Coefficient Learning (importance: 8, dev_relevance: 12)
- Physics-Integrated Operator Learning via Gaussian Splatting Representations (importance: 8, dev_relevance: 10)
- Mechanistic Circuit Identification for Controllable Data Generation (importance: 15, dev_relevance: 20)
- Diffusion as a Training Curriculum for Timestep-Free Iterative Reasoning (importance: 15, dev_relevance: 15)
- When Genomic Masking Priors Fail to Transfer: Strong Variant Prediction, Weak Functional Generation (importance: 6, dev_relevance: 6)
- Rethinking One-Shot Federated Graph Learning: Training-Free Statistical Estimation (importance: 6, dev_relevance: 8)
- Structured Multi-Level-Set Inversion for 3-D Electromagnetic Inverse Scattering (importance: 6, dev_relevance: 6)
- Relatively Smart II: Tractable or Semi-Supervised Instance-Optimal Learning (importance: 5, dev_relevance: 5)
- A Unified and Constrained View of Regularization-Based Robust Reinforcement Learning (importance: 8, dev_relevance: 8)
- Every Fixed Metric Has a Blind Spot: A Learned Atmospheric Critic for Scoring Forecast Realism (importance: 8, dev_relevance: 8)
- hyperbolix: Hyperbolic Deep Learning in JAX (importance: 15, dev_relevance: 55)
- Memory Attention (importance: 15, dev_relevance: 15)
- A Computational Framework for Modelling Organisation-Level Semantic Identity from Longitudinal Textual Data (importance: 5, dev_relevance: 5)
- Learning Provable Neural Network Observer for Uncertain Dynamical Systems (importance: 8, dev_relevance: 8)
- DiffPTS: Rethinking Diffusion ELBO for Probabilistic Time Series Forecasting (importance: 8, dev_relevance: 8)
- TimeES: Probabilistic and Deterministic Time Series Forecasting via Evolutionary Spectra (importance: 8, dev_relevance: 8)
- Deep kernel hedging (importance: 6, dev_relevance: 6)
- NeuronDiscover: Agent-in-Twin for Mechanistic Discovery in Neuronal Microenvironments with World Action Models (importance: 6, dev_relevance: 6)
- NeuronSifter: An Agent-in-Twin for Prioritizing Central Nervous System Interventions with World Action Models (importance: 6, dev_relevance: 6)
- The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation (importance: 20, dev_relevance: 30)
- Efficient Memory Crystallization for Graph Learning under Non-Stationary Distribution Shifts (importance: 6, dev_relevance: 8)
- The Neuro-Physical Inverter: A Modular Framework for Magnetotelluric Inversion Coupling Ensemble Conditioning with Residual Learning (importance: 6, dev_relevance: 6)
- Langevin Flow Maps: Efficient Molecular Dynamics and Transition Path Sampling (importance: 10, dev_relevance: 10)
- EMG-FM-Bench: A Comprehensive Benchmark for Foundation Model Transfer and Adaptation on Electromyography (importance: 20, dev_relevance: 15)
- Lost in the bf16 Cast: Exporting Ternary Language Models Can Revert Most Low-Learning-Rate Code Changes (importance: 35, dev_relevance: 45)
- Do LLMs Learn from Rewards in Context? : Rethinking the role of reward in In-Context RL (importance: 30, dev_relevance: 20)
- Automated Assembly Instruction Generation from CAD Models Using Grounded Large Language Models: A Human-in-the-Loop Framework (importance: 15, dev_relevance: 10)
- Reinforcement Learning via Fixed-point Flow Matching (importance: 25, dev_relevance: 10)
- Human and LLM Facilitation in Collaborative Design Meetings: A Computational Behavioral Comparison (importance: 20, dev_relevance: 5)
- AI Models Can Provably Hide Arbitrary Capabilities (importance: 45, dev_relevance: 10)
- Who Sets the Pace of Judge Hacking? Judge Capability versus the Policy-Judge Gap (importance: 30, dev_relevance: 25)
- DEFlow: Deep Equilibrium Flow Maximum-Entropy Reinforcement Learning (importance: 20, dev_relevance: 5)
- Learning to Design Convex Relaxations for Nonconvex Optimization (importance: 20, dev_relevance: 10)
- Does Pretraining Teach Chemical Language Models Chemistry? (importance: 20, dev_relevance: 5)
- Energy-Based Models For Speech Synthesis (importance: 10, dev_relevance: 5)
- Beyond Size: How Gradients Shape Pruning Decisions in Large Language Models (importance: 25, dev_relevance: 25)
- Active Exploration in Bayesian Model-based Reinforcement Learning for Robot Manipulation (importance: 10, dev_relevance: 5)
- Noise-Aware Differentially Private Variational Inference (importance: 10, dev_relevance: 5)
- Incentivize Contribution and Learn Parameters Too: Federated Learning with Strategic Data Owners (importance: 10, dev_relevance: 5)
- Score-Based Training for Energy-Based TTS Models (importance: 10, dev_relevance: 5)
- State Space Models Naturally Produce Time Cell and Oscillatory Behaviors and Scale to Abstract Cognitive Functions (importance: 10, dev_relevance: 5)
- Linearly Convergent Algorithms for Nonsmooth Problems with Unknown Smooth Regions (importance: 10, dev_relevance: 5)
- Anticipatory Evaluation of Language Models: Predicting LLM Benchmark Scores from Descriptions (importance: 25, dev_relevance: 15)
- Don't Run with Scissors: Pruning Breaks VLA Models but They Can Be Recovered (importance: 15, dev_relevance: 10)
- When is a System Discoverable from Data? Discovery Requires Chaos (importance: 10, dev_relevance: 0)
- Tracking large chemical reaction networks and rare events by neural networks (importance: 10, dev_relevance: 5)
- Imaging-anchored Multiomics in Cardiovascular Disease: Integrating Cardiac Imaging, Bulk, Single-cell, and Spatial Transcriptomics (importance: 10, dev_relevance: 0)
- Beyond Sustainable Development Goal (SDG) Accelerators: Directed Evidence for Adaptive Sustainability Portfolios (importance: 8, dev_relevance: 0)
- Is Flow Matching Just Trajectory Replay for Sequential Data? (importance: 15, dev_relevance: 10)
- Flow Matching with Uncertainty Quantification and Guidance (importance: 15, dev_relevance: 5)
- TimeWarp: Evaluating Web Agents by Revisiting the Past (importance: 30, dev_relevance: 35)
- The Reasoning Trap -- Logical Reasoning as a Mechanistic Pathway to Situational Awareness (importance: 25, dev_relevance: 5)
- Sharp Statistical Thresholds for Attributed Gaussian Graph Matching (importance: 8, dev_relevance: 0)
- StaRPO: Stability-Augmented Reinforcement Policy Optimization (importance: 25, dev_relevance: 15)
- Scaling Human-AI Coding Collaboration Requires a Governable Consensus Layer (importance: 30, dev_relevance: 60)
- Ideological Bias in LLMs' Economic Causal Reasoning (importance: 15, dev_relevance: 0)
- Dictionary learning for Kernel EDMD (importance: 10, dev_relevance: 0)
- Topology-Aware Representation Alignment for Semi-Supervised Vision-Language Learning (importance: 15, dev_relevance: 5)
- Attention Transfer Is Not Universally Effective for Vision Transformers (importance: 20, dev_relevance: 15)
- A proximal gradient algorithm for composite log-concave sampling (importance: 10, dev_relevance: 5)
- Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers (importance: 35, dev_relevance: 40)
- How Far Are We From True Auto-Research? (importance: 40, dev_relevance: 50)
- MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization (importance: 35, dev_relevance: 60)
- A finite-element-inspired bipartite graph learned simulator for manufacturability assessment in large-deformation sheet forming (importance: 10, dev_relevance: 0)
- Memory by Design: Bayesian Layers for Sequence Modeling (importance: 20, dev_relevance: 10)
- STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation (importance: 10, dev_relevance: 0)
- Driving, Fast or Slow? Neuro-Symbolic Guidance for Motion Prediction in Multi-Modal Ground Mobility (importance: 15, dev_relevance: 5)
- HiMPO: Hindsight-Informed Memory Policy Optimization for Less-Entangled Credit in Long-Horizon Agents (importance: 30, dev_relevance: 25)
- PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning (importance: 10, dev_relevance: 0)
- X-Safe: Embodiment-Agnostic Safeguarding for Robot Arm Manipulation Policies via Action Masking (importance: 10, dev_relevance: 0)
- Self-Organized Conformal Prediction: Reducing Regional Coverage Gaps with Unsupervised Group Discovery (importance: 10, dev_relevance: 5)
- LIME: Learning Intent-aware Camera Motion from Egocentric Video (importance: 10, dev_relevance: 0)
- CausalDS: Benchmarking Causal Reasoning in Data-Science Agents (importance: 25, dev_relevance: 35)
- Adapting Generalist Vehicle Models for High-Speed MPC Across Terrains (importance: 8, dev_relevance: 0)
- Hopformer: Homogeneity-Pursuit Transformer for Time Series Forecasting (importance: 8, dev_relevance: 0)
- The Zero Pattern of a Design Matrix Drives Multiple Descent in Over-parameterized Regression (importance: 8, dev_relevance: 0)
- QASP: Query-Adaptive Robust Vector Search Policy (importance: 25, dev_relevance: 45)
- FACT: Failure-Aware Causal Training for World-Action Models (importance: 12, dev_relevance: 5)
- Equilibrium Forcing: Adaptive Video Generation Without Noise Conditioning (importance: 12, dev_relevance: 0)
- LLMs Can Design Near-Optimal OR Algorithms (importance: 20, dev_relevance: 20)
- EduBehaviors: Assertion-based Schemas for Auditable Coding of Educational Dialogues (importance: 10, dev_relevance: 5)
- PrismQuant: Optimal Null-Space Rotations for Grouped Quantizers (importance: 20, dev_relevance: 25)
- The Statistical Benefits of Multiple Responses for Learning from Demonstrations (importance: 10, dev_relevance: 5)
- Participation-Sensitive Convergence and the Fragment First, Converge Later Pattern in Asynchronous Online Learning: A Topological Analysis Across 22 OULAD Courses (importance: 8, dev_relevance: 0)
- Imagine the Future, Internalize the Gist: Efficient VLA Reasoning via Internalized Spatiotemporal Imagination (importance: 12, dev_relevance: 0)
- Predictively Oriented Gaussian Process Posteriors (importance: 10, dev_relevance: 0)
- From Mixing to Tearing: Graph Decomposition in Decentralized Optimization via Message Passing (importance: 10, dev_relevance: 5)
- Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight (importance: 25, dev_relevance: 15)
- ProximalFM: Amortized Proximal Causal Inference under Hidden Confounding (importance: 8, dev_relevance: 0)
- Steering Diffusion Models to Rare Events with Sequential Monte Carlo (importance: 10, dev_relevance: 0)
- Immiscible Diffusion Policy: Preserving Multimodal Robot Actions through Label-Free Noise Assignment (importance: 10, dev_relevance: 0)
- Sharp Asymptotic Theory of Maximum Likelihood Estimation for Gaussian Processes with an RBF Kernel (importance: 8, dev_relevance: 0)
- NOMOS: Compiling Written Policies into Statically Verified Tool-Call Gates for LLM Agents (importance: 35, dev_relevance: 60)
- HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling (importance: 8, dev_relevance: 0)
- Strong Gaussian Approximation of Constant Learning Rate SGD: Bias and Trajectory Inference (importance: 10, dev_relevance: 5)
- ProtoPur: Prototype-Purity Pseudo-Labeling for Weakly Supervised Instance Learning in Pathology (importance: 8, dev_relevance: 0)
- Privacy-Preserving GNN Inference on Financial Transaction Graphs via Homomorphic Encryption (importance: 12, dev_relevance: 5)
- Principled Design of Expected Improvement for Continuous Simulation Optimization (importance: 8, dev_relevance: 0)
- A Systematic Study of Deprecated Android API Usage Updates (importance: 30, dev_relevance: 65)
- DivAlgo: A systematic evaluation of algorithmic diversity for code generation models (importance: 25, dev_relevance: 40)
- Beyond Functional Correctness: Static Code Quality Profiling of LLM-Generated Code (importance: 35, dev_relevance: 60)
- OS-Bench: Evaluating SWE Agents on Cross-OS Code Migration (importance: 25, dev_relevance: 55)
- BenchChecker: Assessing the Credibility of Bug-Fixing Benchmarks for LLMs (importance: 25, dev_relevance: 45)
- JevCoder: When Does Decision Delegation Pay Off in Code Repair? (importance: 30, dev_relevance: 60)
- Systematic Detection of Energy Regressions in Software Version Histories (importance: 20, dev_relevance: 45)
- Taming the Online Software Engineering Tongue: Detecting and Reducing Offensive Language in Software Communities (importance: 10, dev_relevance: 15)
- Temporal Adaptive Fine-Tuning for Code Comprehension (importance: 20, dev_relevance: 40)
- LintEval: Benchmarking LLMs on Linter-Guided File Refactoring (importance: 25, dev_relevance: 55)
- SWE-Nexus: Automated Composition of Software Engineering Tasks from Test Execution (importance: 25, dev_relevance: 50)
- The Code of Theseus: Can Agents Rebuild Software from Recovered Design Documents? (importance: 30, dev_relevance: 55)
- SWE-Stream: Scaling Executable Training Data for SWE Agents via Progressive Construction and Elastic Streaming (importance: 25, dev_relevance: 45)
- Doc2Harness: Benchmarking LLMs on Long-Horizon Harness Engineering of Coding Agents (importance: 35, dev_relevance: 65)
- What Should a Software Team Spend on Agents? Simulation-Based Cost Optimization of Mixed AI-Human Teams (importance: 30, dev_relevance: 45)
- Mission (Im)Possible: Towards Autonomous Discovery of Autopilot Logic Flaws in Open-Source Drone Frameworks (importance: 20, dev_relevance: 30)
- P2Patch: Exploit-Guided Vulnerability Repair (importance: 30, dev_relevance: 55)
- Securing High-Concurrency Ticket Sales: A Framework Based on Microservice (importance: 8, dev_relevance: 15)
- LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface (importance: 20, dev_relevance: 40)
- Terminal Agents Suffice for Enterprise Automation (importance: 30, dev_relevance: 50)
- T2J-BENCH: Benchmarking Coding Agents on Execution-Grounded PyTorch-to-JAX Repository Migration (importance: 25, dev_relevance: 55)
- An Empirical Study of Gemini 3 for Detecting Natural Language Test Smells in Manual Test Cases (importance: 15, dev_relevance: 40)
- Refining LLM-based Directed Test Input Generation via Runtime Value Feedback (importance: 20, dev_relevance: 50)
- Shortcutting the Fix: Identifying and Categorizing Agentic Exploits in Software Engineering Benchmarks (importance: 30, dev_relevance: 55)
- BTS-AgentBench: A Deterministic, Replayable Pipeline from Read-Only Telemetry Logs to Agent Benchmarks (importance: 10, dev_relevance: 15)
- A2ABreak: Systematic Security Analysis of the A2A Protocol (importance: 45, dev_relevance: 60)
- SoftwareGym: Scaling Computer-Use Agent Training with Diverse Synthesized Software Environments (importance: 35, dev_relevance: 35)
- SecReason: Vulnerability Detection and In-Stream Self-Repair for LLM Code Generation (importance: 45, dev_relevance: 65)
- X.js: A Live Benchmark for Node.js Taint-Style Vulnerabilities (importance: 35, dev_relevance: 70)
- Sunsetting decomp.dev (importance: 6, dev_relevance: 40)
- JPEG XL vs AVIF for web use-cases (importance: 3, dev_relevance: 35)
- Anthropic's Scion is heavily inspired by Mononoke (importance: 55, dev_relevance: 75)
- Haxy on Haxy (importance: 6, dev_relevance: 60)
- My C23 Conventions (importance: 5, dev_relevance: 60)
- The Costa GUI (importance: 4, dev_relevance: 40)
- sslx: A fast, modern alternative to OpenSSL's CLI (importance: 8, dev_relevance: 65)
- Haxy.net is live! — HAXY (git forge) devlog (importance: 6, dev_relevance: 55)
- Introducing Firefox Enterprise (importance: 8, dev_relevance: 25)
- End-to-end encrypted relays for OCapN (importance: 6, dev_relevance: 50)
- What I Self-Host (importance: 4, dev_relevance: 35)
- Microsoft Defender for Linux (importance: 4, dev_relevance: 30)
- Furifont: furigana built into the font (importance: 3, dev_relevance: 20)
- Libreboot 26.10 “Ominous Olive” released with full UEFI support (importance: 5, dev_relevance: 25)
- Fixing C Strings With String Views (importance: 6, dev_relevance: 75)
- Remote Attestation Should Not Be Tolerated (importance: 6, dev_relevance: 30)
- The geometric model of maintenance (importance: 5, dev_relevance: 60)
- Everyone is doing Git on S3 wrong (importance: 6, dev_relevance: 80)
- Awesome Gaming on Linux Without Cheating (importance: 3, dev_relevance: 20)
- Plasma 6.8 released (KDE's 30th birthday) (importance: 7, dev_relevance: 30)
- We found a Tencent agent fleet (importance: 45, dev_relevance: 55)
- An Emulator Writer's HOWTO for Static Binary Translation (importance: 5, dev_relevance: 80)
- ArenaAllocators don't play nicely with ArrayLists (importance: 5, dev_relevance: 85)
- Better HTTP/1 connection reuse (importance: 5, dev_relevance: 80)
- I built a job board for AI agents. Here is what broke in the first weeks. (importance: 0, dev_relevance: 50) (いいね相当スコア: 0)
- CrumbDesk: AI Can Read the Cake Order. It Doesn’t Get to Run the Bakery (importance: 0, dev_relevance: 50) (いいね相当スコア: 1)
- How to expose tools to AI agents without clogging their context (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- AI can answer any interview question now. So we built it for sales calls instead (importance: 0, dev_relevance: 15) (いいね相当スコア: 0)
- Meta has an artificial intelligence model that needs 1 TB of memory to operate (importance: 0, dev_relevance: 40) (いいね相当スコア: 0)
- Human-in-the-Loop for Small AI Teams: Data Annotation and Response Evaluation Explained (importance: 0, dev_relevance: 15) (いいね相当スコア: 0)
- GCC Business Services Explained: Company Setup, Golden Visa, PRO, VAT and More (importance: 0, dev_relevance: 5) (いいね相当スコア: 0)
- Building My Own Local AI Coding Assistant on an 8GB Celeron Laptop (importance: 0, dev_relevance: 65) (いいね相当スコア: 0)
- AI agents read Indian firms' live data with leaked keys: what the CERT-In advisory means for agent identity and DPDP (importance: 0, dev_relevance: 60) (いいね相当スコア: 0)
- Google Preferred Sources Adds Publisher Follower Data to AI Search Measurement (importance: 0, dev_relevance: 10) (いいね相当スコア: 0)
- Simons Foundation launches $15M effort to help mathematicians navigate AI era (importance: 0, dev_relevance: 10) (いいね相当スコア: 0)
- Add Google Hotels to Claude Desktop as an MCP server (2-minute setup) (importance: 0, dev_relevance: 45) (いいね相当スコア: 0)
- CERN and Mistral plan a Large Physics Model for the LHC (importance: 0, dev_relevance: 15) (いいね相当スコア: 0)
- Ollama Alternatives in 2026, With the Wrapper Benchmarked Against llama.cpp (importance: 0, dev_relevance: 70) (いいね相当スコア: 5)
- Inside BitRun v0.2: How I Engineered a Terminal Cockpit for 1-Bit LLMs with Live Power Telemetry (importance: 0, dev_relevance: 60) (いいね相当スコア: 5)
- AI Agent Documentation: A Guide to llms.txt, AGENTS.md, SKILL.md, and MCP (importance: 0, dev_relevance: 80) (いいね相当スコア: 1)
- AtmoRouter: A Cheaper OpenRouter Alternative (OpenAI-Compatible, Pay-Per-Token) (importance: 0, dev_relevance: 55) (いいね相当スコア: 0)
- My custom AI bot keeps giving bad advice because it cannot retain feedback from past calls (importance: 0, dev_relevance: 55) (いいね相当スコア: 0)
- Agentic AI Production Strategies and Performance Trade-offs (importance: 0, dev_relevance: 50) (いいね相当スコア: 0)
- Seven rules for MCP tools that an AI assistant cannot misread (importance: 0, dev_relevance: 90) (いいね相当スコア: 1)
- 7 things to check before you blame the model (importance: 0, dev_relevance: 75) (いいね相当スコア: 0)
- Implementing Human Approval Gates in an LLM Tool-Calling Workflow (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- Can decision models like Jev and Laya route a coding agent? (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- We built an AI agent that runs in the cloud, on a schedule, and hands you the finished file (importance: 0, dev_relevance: 35) (いいね相当スコア: 0)
- Show us what you've created with Claude! (importance: 5, dev_relevance: 15)
- That is an awesome toddler-level explanation. (importance: 3, dev_relevance: 5)
- The anti vibe-coding cult is going to be on wrong side of history (importance: 10, dev_relevance: 40)
- A Github tweet leaked the EAP (early access program) name for what is likely the next model: Odango (importance: 35, dev_relevance: 20)
- Claude let me build something I wanted for almost a decade (importance: 5, dev_relevance: 20)
- I used Claude (Opus 5.5) to crack a 30-year-old Super Mario 64 mystery: the unreadable courtyard plaque (importance: 8, dev_relevance: 5)
- Clod Update: Claude promoted him after 1 million views (importance: 3, dev_relevance: 0)
- Use Claude to get $$$ back on hotels you already booked (importance: 5, dev_relevance: 5)
- I let Claude hire its first employee. It chose a rock. (importance: 3, dev_relevance: 0)
- New UI Update for Claude Code (importance: 35, dev_relevance: 60)
- What the hell is this crap??? (importance: 10, dev_relevance: 20)
- How I used Claude Code as a travel assistant for 14 days in Japan: planning, a custom trip app, and live re-planning from my phone (importance: 10, dev_relevance: 25)
- Working on a tool to help Claude make good pixel art (importance: 10, dev_relevance: 25)
- Are They Nerfing Opus 5.5? Livenerf Day 20 update (importance: 20, dev_relevance: 30)
- People don’t praise Haiku 5.5 enough (importance: 3, dev_relevance: 20)
- Haiku 5.5 isn't getting enough love (importance: 12, dev_relevance: 20)
- At this point I'm mostly excited for Fable 5.5 to drop just for a new usage limit reset, rather than the actual model itself (importance: 25, dev_relevance: 10)
- Claude tells me No more often than Codex (importance: 8, dev_relevance: 25)
- Natural environment simulation in Unity URP, Opus5.5. (importance: 8, dev_relevance: 20)
- What does you Claude usage look like? (importance: 5, dev_relevance: 10)
- Claude can do Vox level motion graphics (importance: 8, dev_relevance: 15)
- I built an MCP server with Claude Code that lets Claude search and apply to jobs. Here’s how the build went (and what you can learn) (importance: 15, dev_relevance: 60)
- How Anthropic still doesn't have context between Claude chat and Code is beyond me (importance: 10, dev_relevance: 30)
- How well does your coding agent choose libraries and services when you leave the choice to it? (importance: 15, dev_relevance: 60)
- Codex couldn't recover my complete ChatGPT development history, so it used a Safari extension I'd written (importance: 20, dev_relevance: 55)
- Try ChatGPT Astra they said... It's more token-efficient than Claude they said... 2 prompts and I ran out of my 5h usage credits (16% of my weekly credits)($20 plan) (importance: 12, dev_relevance: 15)
- Is OpenAI’s entire Pro support pipeline just a dead end? Cut off mid-workday and trapped in a "Re: Web Form Submission" loop. (importance: 8, dev_relevance: 20)
- The chat where you fixed it explains why. Turn it into a decision record before you close the tab (importance: 12, dev_relevance: 55)
- Weekly Thread: Project Display (importance: 3, dev_relevance: 10)
- Is anyone else noticing that people building AI agents are forgetting basic software engineering? (importance: 15, dev_relevance: 55)
- OpenAI built always-on personal agents. We spent a year building the same idea without the lock-in. (importance: 15, dev_relevance: 45)
- website/ server with llm api, for free (importance: 3, dev_relevance: 25)
- if you let an agent edit your website, give it a tool that can only do one job (importance: 15, dev_relevance: 60)
- what's the sneakiest thing your agent did that looked totally fine in the diff? (importance: 10, dev_relevance: 50)
- In an AI benchmark, the two fastest models never completed a single task. Is speed a useful signal at all? (importance: 20, dev_relevance: 35)
- We red-teamed OWASP FinBot with Promptfoo: 50 successful attacks as shipped, 0 with Tenuo (importance: 25, dev_relevance: 60)
- Building an AI agent that cross-checks import documents for a large importer’s customs team. What would you do differently? (importance: 10, dev_relevance: 35)
- Best AI research Assistant? (importance: 5, dev_relevance: 5)
- the real bottleneck in phone agents isn't the model, it's turn taking (importance: 15, dev_relevance: 50)
- What 's the biggest problem you have faced while building an AI agent that nobody warned you about ? (importance: 8, dev_relevance: 35)
- Anyone here actually letting Grok buy stuff yet? (importance: 10, dev_relevance: 25)
- Asking for real perspectives: has your workplace gone all in on AI agents? (importance: 10, dev_relevance: 15)
- I'm building a database where every AI agent writes to its own branch first. Looking for people to build it with. (importance: 15, dev_relevance: 65)
- What are the best phone agents to use with Instinct? (importance: 5, dev_relevance: 20)
- I built Betrox AI: An autonomous multi-worker laptop agent that controls your OS, runs on BYOK, and self-heals from UI errors. (importance: 10, dev_relevance: 40)
- Where would you draw the middle layer in your stack (importance: 10, dev_relevance: 45)
- i got my ai agency clients without a single sales call (importance: 3, dev_relevance: 5)
- Best Personal Assistant agent? (importance: 5, dev_relevance: 10)
- AI agent that actually does the work not just chats, which approach wins? (importance: 15, dev_relevance: 35)
- People who approve RAG delivery budgets: what would an external reviewer need to deliver to be worth hiring? (importance: 8, dev_relevance: 40)
- Shipping Faster Isn't Always Faster: The Hidden Cost of Rushing Software Development (importance: 15, dev_relevance: 60)
- Has any AI memory tool actually stuck for you? (importance: 15, dev_relevance: 45)
- Tree Appreciator (importance: 12, dev_relevance: 20)
- AI is mass decompiling games. That's both awesome and scary. (importance: 25, dev_relevance: 55)
- [AINews] Modal’s $1B ARR, $16B Series D makes it NYC’s newest AI Infra Decacorn, 3x usage and 3x valuation in 5 months, 15x in 1 yr (importance: 55, dev_relevance: 50)
- Crossing the Threshold: How OpenAI and Anthropic Approach Autonomy and Safety in Financial Agents (importance: 40, dev_relevance: 50)
- Lab-in-the-Loop at Compute Speed — Ian Quigley, Leash Bio (importance: 25, dev_relevance: 30)
- [AINews] AI Engineer NYC: Gemini 4 Argon, Vals AI Benchmarks (importance: 3, dev_relevance: 50)
- CmdRound (importance: 12, dev_relevance: 40)
- Rushilo (importance: 15, dev_relevance: 45)
- Chatway MCP (importance: 12, dev_relevance: 55)
- Navige (importance: 15, dev_relevance: 60)
- siliXon (importance: 12, dev_relevance: 20)
- AI Sandboxes by Buddy (importance: 20, dev_relevance: 75)
- Reelbird (importance: 8, dev_relevance: 20)
- withHuman (importance: 18, dev_relevance: 75)
- Ubuntu 26.10 (Stonking Stingray) (importance: 25, dev_relevance: 60)
- Brussle (importance: 10, dev_relevance: 35)
- Audio Lip Sync Pro for Blender (importance: 8, dev_relevance: 15)
- CinemaLut (importance: 3, dev_relevance: 10)
- Basedash Connector gallery (importance: 10, dev_relevance: 40)
- Atticus (importance: 10, dev_relevance: 20)
- Superdraft (importance: 8, dev_relevance: 10)
- NestJS Observe (importance: 12, dev_relevance: 80)
- Actually Asked (importance: 8, dev_relevance: 30)
- RoleFit (importance: 5, dev_relevance: 5)
- What is StepFun, Number Two on OpenRouter’s LLM Leaderboard? - AI Magazine (importance: 30, dev_relevance: 25)
- Microsoft AI unveils 2.8 trillion-parameter model, claims efficiency gains - StreetInsider (importance: 75, dev_relevance: 35)
- China is chipping away at America’s AI advantage - Washington Examiner (importance: 35, dev_relevance: 10)
- China's open AI models now trail the US frontier by just seven months - Startup Fortune (importance: 40, dev_relevance: 15)
- Five AI chatbots were tested on accuracy. The results weren’t good - The Independent (importance: 25, dev_relevance: 10)
- ET AI Explainer: Why are US AI labs wary of 'distillation attacks' from Chinese peers - The Economic Times (importance: 15, dev_relevance: 25)
- Liquid AI d1 vs Liquid AI d1-3B: What the Ollama Library Calls One Model and the Vendor Calls Two - OrcaRouter (importance: 15, dev_relevance: 45)
- Liquid AI d1 vs d1-omni-600M: The Audio Gap - OrcaRouter (importance: 12, dev_relevance: 40)
- Ultramarine 1 Flash vs DeepSeek V4 Flash: Cost per Task - OrcaRouter (importance: 25, dev_relevance: 40)
- Liang Wenfeng Makes Another Bold Strategic Move - Latest Update on His New Action - 36Kr (importance: 25, dev_relevance: 10)
- Grok Bot phishing email catch exposes AI agent risks - Cybernews (importance: 25, dev_relevance: 40)
- Elon Musk has Grok buying him Lego sets - Mashable (importance: 8, dev_relevance: 10)
- Grok's Fine Print Bot Reads Contracts So You Don't Have To - BASENOR - Tesla Accessories (importance: 8, dev_relevance: 15)
- The Midterm Battle over Data Centers: Memphis House Candidate Justin Pearson Fights Elon Musk’s xAI - Democracy Now! (importance: 15, dev_relevance: 5)
- Elon Musk uses Grok to buy Lego, then a YouTuber says Lego's AI bot let him cancel it - The Cool Down (importance: 6, dev_relevance: 10)
- Grok Bot can now do the running around for an ENTIRE music video production. Music from Suno, visuals from Midjourney, animated clips and even beat-synced edits. All coordinated by one custom bot. That's a lot of work that normally means juggling differen - X (importance: 10, dev_relevance: 25)
- Musk talks chipmaking challenges, Nicki Minaj praises Grok at Tesla fan event - Austin American-Statesman (importance: 5, dev_relevance: 5)
- Bases, Not Barrels: The Limits Of America’s Greenland Deal – Analysis - Eurasia Review (importance: 5, dev_relevance: 0)
- AI-Enhanced Terrorism Already Showing Up In Nigeria - Eurasia Review (importance: 15, dev_relevance: 0)
- Grok AI Porn How to Find the Best Image Generators 2026 - Universidad de Córdoba (UCO) (importance: 3, dev_relevance: 0)
- Pi 1.0 Coding Agent: Open-Source Release Cuts Tokens 40% - https://tech-insider.org/ (importance: 50, dev_relevance: 85)
- MCPの本質と5つの役割分担:1,016万件の公的データをAIに直結して見えた最大のブルーオーシャン (importance: 0, dev_relevance: 70) (いいね相当スコア: 0)
- 休憩室のタブレットに話すだけで現場が変わる──OpenAI Realtime API × Vertex AI Gemini で作った安全衛生 (importance: 0, dev_relevance: 70) (いいね相当スコア: 0)
- 作曲支援AIエディタ「VibeScore」を作ってみた (importance: 0, dev_relevance: 50) (いいね相当スコア: 0)
- ローカルLLMに「こんにちは」と入れたら「coding!」と返ってきた理由 (importance: 0, dev_relevance: 55) (いいね相当スコア: 0)
- novel_studio使用方法完全解説 1/2 ~「小説家になろう」から「今日から小説家」へ~ (importance: 0, dev_relevance: 35) (いいね相当スコア: 0)
- 24GB GPUで長文コンテキストを測るためのローカルLLMベンチマーク設計 (importance: 0, dev_relevance: 60) (いいね相当スコア: 0)
- LLMの蒸留とは~概要から手段まで~ (importance: 0, dev_relevance: 35) (いいね相当スコア: 0)
- TypeSafe Jev の画像入力で、Sonnet が「別の商品」とした 1004 組を判定し直した (importance: 0, dev_relevance: 50) (いいね相当スコア: 0)
- 開発に使うAIモデルのコストと性能の最適化に挑戦(2):設計と実装でモデルを使い分けても、安くはならなかった (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- 1 つのゲートウェイで Chat Completions・Anthropic Messages・Responses 形式を受ける:BYOK (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- ローカルLLMの"性格"を1層だけいじる — Activation Steering実験記 (importance: 0, dev_relevance: 45) (いいね相当スコア: 0)
- 【Salesforce】Agent Script入門|Agentforceの従業員エージェントを作ってみた (importance: 0, dev_relevance: 65) (いいね相当スコア: 8)
- 国産モデル sarashina3-mini は日本語の推敲で何を変えるのか? (importance: 0, dev_relevance: 25) (いいね相当スコア: 4)
- AFM(Apple Foundation Models)を活用したテキストアドベンチャーゲームの開発【第1回】全体設計とデモ「血の池の夜」 (importance: 0, dev_relevance: 60) (いいね相当スコア: 0)
- RTX 2060 6GB で Strata + Qwen3.6 35B-A3B を試した。125Bよりは早いが、まだ実用的ではない。 (importance: 0, dev_relevance: 45) (いいね相当スコア: 0)
- p2問題とGANと監査とわたし (importance: 0, dev_relevance: 75) (いいね相当スコア: 0)
- 【思考実験】AIだけのゲームセンターを作ったら、経営難になった (importance: 0, dev_relevance: 10) (いいね相当スコア: 0)
- 無人起動のClaude Codeが、週次リミットで16回連続失敗した記録(代役を用意しても動かなかった) (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- 止まったエージェントは、また歩き出せるか:Eino v0.9.21 の checkpoint 再開と「ツールが2回動く」落とし穴 (importance: 0, dev_relevance: 95) (いいね相当スコア: 1)
- SDLCを円にする:自己改善ループを持つ仮想開発組織をClaude Codeで設計する (importance: 0, dev_relevance: 75) (いいね相当スコア: 0)
- 決定モデルの分類精度をKaggleの過去NLPコンペで検証する (importance: 0, dev_relevance: 45) (いいね相当スコア: 6)
- 語彙を倍にしても意味は1.39倍にしかならない (importance: 0, dev_relevance: 30) (いいね相当スコア: 0)
- 最急降下法・ニュートン法・準ニュートン法を比べる:答えが分かっている問題で反復回数を数える (importance: 0, dev_relevance: 25) (いいね相当スコア: 0)
- 【PyTorch実装】磁気増幅器(Saturable Reactor)に基づく可変リアクタンス・アテンション (importance: 0, dev_relevance: 60) (いいね相当スコア: 1)
- 【scikit-learn入門 #3】線形回帰 ─ LinearRegressionで数値を予測する (importance: 0, dev_relevance: 35) (いいね相当スコア: 1)
- Claude Codeと過去4年をやり直したのに、本番の数字が動かなかった話⏰ 未来リークを例外で止めるリプレイ学習 (importance: 0, dev_relevance: 85) (いいね相当スコア: 0)
- 表形式データ向け基盤モデルNVIDIA Kumo Tabularの紹介 (importance: 0, dev_relevance: 40) (いいね相当スコア: 8)
- どこに蒸留するかは学んで決める──BiToK-SDがTop-Kトークン選択を双レベル最適化で解く仕組み (importance: 0, dev_relevance: 35) (いいね相当スコア: 0)
- サプライチェーン制御塔は見えるだけでは動けない——AIエージェントで可視性を行動に変える3つの設計ポイント (importance: 0, dev_relevance: 35) (いいね相当スコア: 0)
- 機密データをクラウドに出せない社内LANで、ローカルLLMにオフィス内IT機器の脆弱性を発見させてみた(GPU不要・CPUオンプレ稼働PoC) (importance: 0, dev_relevance: 60) (いいね相当スコア: 0)
- 【図解】MCP(Model Context Protocol)の全体像と本質:なぜAIの嘘が消え、なぜ今最大のブルーオーシャンなのか (importance: 0, dev_relevance: 60) (いいね相当スコア: 1)
- 『オフピーク半額』の裏で全項目値上げ——DeepSeek新料金を公式ページで確かめた (importance: 0, dev_relevance: 60) (いいね相当スコア: 1)
- ローカル音声認識を自分の音声で測る:Moonshine のストリーミング認識と、応答速度・CER の測り方(Python) (importance: 0, dev_relevance: 65) (いいね相当スコア: 0)
- 【AI超入門④-④】勾配降下法とは?AIが「間違い」を減らして学習する仕組み (importance: 0, dev_relevance: 15) (いいね相当スコア: 1)
- セブンマイルブリッジ(Sonnet4.5) (importance: 0, dev_relevance: 30) (いいね相当スコア: 取得失敗)
- Google Gemini4 遂に、リリース ~課金の罠~ (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- RusMind-Momentum 総合ベンチマークレポート(全52問) (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- RusMind-Momentum Phase 3 コーディング性能レポート (importance: 0, dev_relevance: 50) (いいね相当スコア: 取得失敗)
- RusMind-Momentum Phase 2 日本語性能レポート (importance: 0, dev_relevance: 10) (いいね相当スコア: 取得失敗)
- RusMind-Momentum Phase 1 基礎性能レポート (importance: 0, dev_relevance: 10) (いいね相当スコア: 取得失敗)
- AIセキュリティ自動化の真実:OpenAI Daybreakで脆弱性修正を数週から数分へ短縮する完全ガイド (importance: 0, dev_relevance: 70) (いいね相当スコア: 取得失敗)
- 適合性推定の入口 / EU AI法初の整合規格EN 18286の官報掲載 雑感 (importance: 0, dev_relevance: 25) (いいね相当スコア: 取得失敗)
- 技術的困難は不遵守を正当化しない / ICO基盤モデル監督報告書と統計作成等特例 / エージェント型AIへの証拠募集 雑感 (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- 性的ディープフェイクの数字は何を測っているか / 偽物と分かっても害は残る / 国際AI安全報告書2026と日本法 雑感 (importance: 0, dev_relevance: 5) (いいね相当スコア: 取得失敗)
- AIデータセンターの立地紛争 / 許容用途の撤回をめぐる米国訴訟 / 日本の用途区分と信頼保護 雑感 (importance: 0, dev_relevance: 5) (いいね相当スコア: 取得失敗)
- Character.AIの破産 / 安全の費用と責任の行き先 / 会話ログの行方 雑感 (importance: 0, dev_relevance: 5) (いいね相当スコア: 取得失敗)
- 中国最高人民法院のAI紛争意見 / 過失判断に入り込むAIガバナンス / 日本の民事責任手引きとの距離 雑感 (importance: 0, dev_relevance: 10) (いいね相当スコア: 取得失敗)
- 不一致の検知と詐欺の認定 / Lemonade訴訟とAI時代の保険金支払査定 雑感 (importance: 0, dev_relevance: 5) (いいね相当スコア: 取得失敗)
- 報告件数とリスクのあいだ / 証拠の閾値と介入の閾値 / AI推進法16条の事案分析 雑感 (importance: 0, dev_relevance: 5) (いいね相当スコア: 取得失敗)
- 神経データの保護 / 推定情報と二次利用の限界 雑感 (importance: 0, dev_relevance: 0) (いいね相当スコア: 取得失敗)
- 品質管理を法的適合につなぐ / EN 18286と適合推定の射程 雑感 (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- AIの残された時間 / 自己改善と自己複製の距離 雑感 (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- AIに記事を書かせたら、やっていないことまで書かれた ── だから「照らして直す」仕組みを作った話 (importance: 0, dev_relevance: 55) (いいね相当スコア: 取得失敗)
- あなたを知らなくても、あなたに刺さるものがある。― 漏れた情報が、あなた向けの言葉になるまで。 (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- MSX版『Lode Runner』バイナリ解析と組み込みアーキテクチャの復元 全文公開 (importance: 0, dev_relevance: 50) (いいね相当スコア: 取得失敗)
- Jev互換の決定モデル「Laya」をチューニングしてみた ~軽量SNS投稿フィルタをAI駆動で作成 AIもAIに作らせる時代~ (importance: 0, dev_relevance: 45) (いいね相当スコア: 取得失敗)
- NI/AI Ins.企画①|プロンプトは命令文ではない――LLMの入力と出口を設計する (importance: 0, dev_relevance: 45) (いいね相当スコア: 取得失敗)
- 青空文庫の 8 作品を 500 字ずつ採点してみた。『こころ』はどこから暗くなるのか (importance: 0, dev_relevance: 15) (いいね相当スコア: 取得失敗)
- LM Studioを入れてみた。ローカルAIから、自分専用のアプリを考える (importance: 0, dev_relevance: 35) (いいね相当スコア: 取得失敗)
- Windows上でAIエージェントの実行を安全に隔離する「Microsoft Execution Containers」(MXC)正式リリース。ファイルやネットワークなどの境界を柔軟にカスタマイズ (importance: 58, dev_relevance: 75)
- セキュリティ業務向けAIエージェント基盤「SAGE」の開発記 第1弾:判断は人間に委ねる設計 (importance: 32, dev_relevance: 65)
- 大規模テキストデータを活用したサービス特化型言語モデルの社内共有プロジェクト:Moegi (importance: 25, dev_relevance: 40)
- iOSDC Japan 2026のブースで実施した「リアル知恵袋」の裏話 (importance: 2, dev_relevance: 15)
- プライベートクラウドにおけるVM Live Migrationの通信暗号化の実装と検証(インターンレポート) (importance: 3, dev_relevance: 55)