AI News Digest 2026-10-22

台本で使った記事

特集

DuckDB 2.0.0 LTS「Cinnamon Teal」登場
DuckDB 2.0.0 LTS "Cinnamon Teal"

[01:39〜]

Google Cloudが Gemini・Claude に加え OpenAI モデルも提供開始、3社揃う唯一の主要クラウドに
Google Cloud now offers OpenAI models alongside Gemini and Claude, becoming the only major provider with all three

[12:37〜]

DeepSeek DeepInferはRTX 5090では動かない ― 代わりに何を動かすべきか
DeepSeek DeepInfer Won't Run on Your RTX 5090. Here's What to Run Instead

[18:30〜]

開発者コーナー

TypeScript 7.1ベータ、型検査を速くし埋め込み言語のAPIを追加

[27:30〜]

中堅コーナー

AIが準リーマン予想を証明 ― その後に何が起きるのか
AI Proved the Quasi-Riemann Hypothesis. What Happens Now?

[32:41〜]

OpenAI、低資源言語の実装とCOBOL移行で精度を公開

[36:05〜]

AIツール紹介コーナー

侵害されたLiteLLM 1.107.0、PyPI上で認証情報窃取ツールとバックドアをインストール
Compromised LiteLLM 1.107.0 on PyPI Installs a Credential Stealer and Backdoor

[39:32〜]

速報コーナー

Cloudflare・VoidZero、TypeScriptコンパイラのRust移植「tsrs」を発表

[46:02〜]

Google Cloud、ロンドンのゾーンが停電で約8時間停止

[47:04〜]

本日のPHPセキュリティリリースと、PHPセキュリティ強化の取り組み
Today's PHP Security Release, and How We're Strengthening PHP Security

[47:57〜]

Google Cloud、TPU・GPUクラスタ向けキャッシュ対応ルーティングのllm-d 1.0を提供開始
Google Cloud Ships llm-d 1.0 With Cache-Aware Routing for TPU and GPU Fleets

[49:00〜]

Discord、Twitch埋め込み経由の1クリックRCEに5500ドルの報奨

[50:01〜]

Mistral AIが音声エージェントSDK付きのVoxtral 3(音声認識・音声合成)を公開
Mistral AI Releases Voxtral 3: Speech-to-Text, TTS and a Voice Agent SDK

[51:01〜]

CodeQL、macOS 27とXcode 27でトレース解析が使えない

[51:56〜]

エージェント規模のGitインフラへの橋渡し
The bridge to agent-scale Git infrastructure

[52:59〜]

GoogleもAmazonに続きAI開発者向けツールでNvidiaの牙城に挑む
Google has joined Amazon in attacking Nvidia's stronghold with AI app developers

[53:58〜]

参考記事一覧を表示

Google Cloud now offers OpenAI models alongside Gemini and Claude, becoming the only major provider with all three (importance: 70, dev: 40)
OpenAI acquires Orchestra Bio to build AI agents that speed up drug development (importance: 50, dev: 20)
Eight Bits of Trust; eMudhra's CA keys cracked in milliseconds (importance: 25, dev: 60) (いいね相当スコア: 0)
Today's PHP Security Release, and How We're Strengthening PHP Security (importance: 55, dev: 80)
Conversational Gemini navigation comes to Apple CarPlay (importance: 35, dev: 5)
Mistral AI Releases Voxtral 3: Speech-to-Text, TTS and a Voice Agent SDK (importance: 50, dev: 70)
Google Open-Sources TorchTPU, a Native PyTorch Backend for Cloud TPUs (importance: 4, dev: 85)
AWS、PyTorchをTrainiumでそのまま動かすTorchNeuronを公開　PrivateUse1の標準バックエンドとしてApache 2.0、10月20日にパブリックベータ (importance: 4, dev: 90)
DeepSeek DeepInfer Won't Run on Your RTX 5090. Here's What to Run Instead (importance: 65, dev: 85) (いいね相当スコア: 1)
Cloudflare・VoidZero、TypeScriptコンパイラのRust移植「tsrs」を発表　TS 7.0比最大8倍速・メモリ約7割減、フォークではなくTypeScriptチームと協力 (importance: 60, dev: 95)
Cloudflare Releases Clairic, a System One Extraction Model on Workers AI (importance: 35, dev: 60)
Inception's Mercury 3 diffusion model claims to beat GPT-6 Luna and Claude Haiku 5.5 at a fraction of the wait (importance: 45, dev: 30)
Achieve Productive AI Capacity Faster and More Predictably with NVIDIA DSX POD (importance: 25, dev: 20)
China Criticizes U.S. Ban on DeepSeek A.I. in Government Systems - The New York Times (importance: 30, dev: 5)
OpenAI can't yet explain why its coding agents switch off misalignment monitoring (importance: 65, dev: 50)
Patch the Planet: global response coordination as agents uncover critical vulnerabilities (importance: 35, dev: 55)
Trump administration loses bid to block NAACP lawsuit against Musk's xAI over data center - TradingView (importance: 20, dev: 5)
SpaceX’s AI business is beating expectations (importance: 25, dev: 5)
Hacker Group Used Claude, Grok to Automate Cyberattacks on Latin American Banks, The Information Reports - marketscreener.com (importance: 50, dev: 20)
OpenCode Security Breach: Distillation Attack Hit Console Users With Linked AI Providers. How to Find OpenCode on Developer Machines and Respond - StepSecurity (importance: 55, dev: 80)
XAI Floating Rate & Alternative Income Trust Announces Final Results of Tender Offer - The Manila Times (importance: 2, dev: 0)
Figma co-founder Dylan Field thinks MCP is not a perfect protocol (importance: 30, dev: 60)
Python機械学習の実践書4選 (importance: 0, dev: 45) (いいね相当スコア: 0)
DuckDB 2.0.0 LTS "Cinnamon Teal" (importance: 70, dev: 95)
Compromised LiteLLM 1.107.0 on PyPI Installs a Credential Stealer and Backdoor (importance: 80, dev: 95)
Physicist Christopher Stubbs joins Anthropic's Long-Term Benefit Trust (importance: 20, dev: 5)
v2.1.306 (importance: 25, dev: 90)
v2.1.305 (importance: 38, dev: 90)
How AT&T is rethinking in-house legal work with OpenAI (importance: 25, dev: 10)
ChatGPT Ads expands to 26 new countries as over 100,000 advertisers have used the platform (importance: 35, dev: 5)
Using thousands of AI agents to uncover new leads in biological data (importance: 40, dev: 25)
How we built ChatGPT’s novel ad stack for user trust and relevance (importance: 35, dev: 60)
Franchisee reconciles cash in 94% less time with ChatGPT Work (importance: 15, dev: 5)
Terminal X matches senior investor judgment 88.6% with GPT-6.1 (importance: 20, dev: 25)
Southeast Asia’s digital economy to hit $368 billion (importance: 3, dev: 0)
Empowering frontline maternal care with AI-assisted ultrasound (importance: 15, dev: 5)
Solving a 31-year-old grand challenge in mathematics – the KLS conjecture (importance: 30, dev: 5)
Called to serve: Tech, research, and positive impact (importance: 8, dev: 5)
NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required (importance: 30, dev: 40)
How Specializing the Vocabulary of SLMs Can Cut Inference Cost by 77% in Real Applications (importance: 35, dev: 55)
Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster (importance: 35, dev: 35)
FairStress: A deeper evaluation of fairness in LLM decision-making (importance: 25, dev: 20)
ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model (importance: 35, dev: 55)
AnyLanguageModel 1.0: One Year of Local Models on Apple Platforms (importance: 35, dev: 80)
MPS in PyTorch 2.15 (importance: 40, dev: 85)
The bridge to agent-scale Git infrastructure (importance: 50, dev: 75)
How 3 Java Champions are using GitHub Copilot for increased productivity (importance: 25, dev: 50)
Introducing full transaction detection for Cloudflare WAF (importance: 35, dev: 55)
Cloudflare deploys protections against php2shell vulnerability (importance: 45, dev: 70)
How Commerce.com and Cloudflare are getting storefronts ready for AI agents (importance: 40, dev: 55)
JetBrains Plugin Developer Conf 2026: AI, New APIs, and More (importance: 20, dev: 70)
Small Talk with Ulises Olave, CTO and Co-Founder of Fitia (importance: 5, dev: 10)
Coding-Agent Benchmarks Should Match Their Users’ Task Flows (importance: 45, dev: 80)
Why We Still Need an IDE (importance: 25, dev: 80)
Your Agent Runs in a Docker Sandbox. Now It Can Join a Team. (importance: 25, dev: 60)
From baseline to Secure Software Factory: securing the systems that build your software (importance: 30, dev: 60)
Design Proteins with RFD4-Proteina Using Reward-Guided Inference-Time Search (importance: 20, dev: 30)
Inside NVIDIA Sentry: Secure AI Agents at Runtime (importance: 30, dev: 60)
How NVIDIA OpenShell Reference Architecture Secures and Governs AI Agent Workloads at Scale (importance: 25, dev: 55)
Developing an AI Factory Digital Twin with an Agent-Assisted Workflow (importance: 15, dev: 40)
Accelerating Mamba-2 with High-Performance Forward Pass Kernels in NVIDIA CUDA Tile (importance: 30, dev: 85)
NVIDIA Run:ai Is Now Open Source as Part of NVIDIA DSX AI Factory Platform (importance: 45, dev: 70)
The new economics of enterprise AI (importance: 15, dev: 20)
AI adoption by industry shows what other sectors can learn (importance: 15, dev: 15)
A visit to Samsung’s massive new semiconductor hub (importance: 25, dev: 10)
Judge rejects Trump’s wild claim EPA can dismiss data center pollution lawsuits (importance: 15, dev: 0)
The $15 billion ad-powered AI startup your doctor may already be using (importance: 20, dev: 5)
Meta drops the word "train" from its commercial TOS after AI backlash (importance: 20, dev: 5)
Ukraine’s drone rescue of wounded soldier puts much touted idea into practice (importance: 5, dev: 0)
Meta tries to scapegoat staff who downloaded torrents to spare execs (importance: 20, dev: 0)
Meta Muse exposes users' personal and business secrets when they share conversations (importance: 45, dev: 5)
AI promises productivity, but for managers, it’s mostly a headache for now (importance: 15, dev: 10)
Anthropic pitches AI safety bill that may block a model if 50+ people could be killed (importance: 45, dev: 5)
Hugging Face CEO warns that Netflix’s OpenAI hack documentary isn’t accurate (importance: 15, dev: 5)
Anthropic’s cloud partnerships are getting messy (importance: 40, dev: 10)
Google has joined Amazon in attacking Nvidia’s stronghold with AI app developers (importance: 50, dev: 75)
ChatGPT now lets you lock your chats (importance: 30, dev: 5)
Eric Schmidt reportedly owns about half as much of Anthropic as CEO Dario Amodei (importance: 12, dev: 0)
Meta’s Muse hits 10M US downloads in 40 days, a milestone reached by only 11 other apps (importance: 20, dev: 0)
Ex-Nvidia-turned-VC Amulya Vishwanath: ‘Physical AI is in its LEGO era’ (importance: 10, dev: 5)
Plants can speak using electricity, so this startup built an AI model to decipher their language (importance: 10, dev: 5)
Eka is worth $1.45B thanks to a physical AI model that has experts raving (importance: 25, dev: 10)
Yelp can now call on your behalf to check wait times at a restaurant (importance: 20, dev: 5)
Otter AI raises $320M from General Catalyst’s growth fund for its meeting tool (importance: 25, dev: 10)
Mamdani says feds have abdicated on AI, urges New Yorkers to push back (importance: 15, dev: 0)
Judge dismisses Sam Altman’s sister’s sexual abuse lawsuit (importance: 10, dev: 0)
Meta pauses Muse sharing after researchers find AI chats on the web (importance: 45, dev: 10)
Spotify pulled 50 million AI slop and spam tracks from its recommendations (importance: 25, dev: 0)
Adobe Lightroom mobile can now open closed eyes in photos with a simple tap (importance: 15, dev: 0)
CoreWeave’s debt is becoming the whole AI industry’s problem (importance: 40, dev: 10)
Article Series: Agentic AI Architecture (importance: 30, dev: 75)
Google SynthID Detector Simplifies Watermark Detection and Expands Support to Other AI Providers (importance: 35, dev: 30)
TornadoVM 7 Enables GPU-Native Workloads From Java Applications (importance: 30, dev: 90)
Presentation: Beyond the Prototype: Scaling Framework Agnostic AI Agent Infrastructure with Ray (importance: 25, dev: 80)
How Wikipedia’s Editors Keep the Online Encyclopedia Human-Written (importance: 15, dev: 10)
Could Treating AI as Conscious Make It More Dangerous? (importance: 20, dev: 5)
Deepmind founder Hassabis shortens AGI timeline from five to two to three years (importance: 35, dev: 5)
Mistral and Nvidia want to speed up industrial design with physics AI (importance: 30, dev: 25)
Claude Science tackles deadly scorpion stings by designing protein binders from scratch (importance: 45, dev: 10)
Elevenlabs releases two new speech-to-speech models designed for real-time AI conversations (importance: 40, dev: 40)
ITエンジニア1265人の調査、AIを毎日使う層ほどキャリア不安が強い (importance: 35, dev: 25)
Google Cloud、ロンドンのゾーンが停電で約8時間停止 (importance: 60, dev: 80)
露出したCairnの操作画面から、AIによる侵入作業が判明 (importance: 65, dev: 60)
TypeScript 7.1ベータ、型検査を速くし埋め込み言語のAPIを追加 (importance: 65, dev: 95)
Meta、端末上のエージェント向け30BモデルMuse Glimmerを公開 (importance: 60, dev: 50)
Discord、Twitch埋め込み経由の1クリックRCEに5500ドルの報奨　force_manifest_nodeからElectronサンドボックス脱出まで連鎖 (importance: 50, dev: 75)
CodeQL、macOS 27とXcode 27でトレース解析が使えない (importance: 45, dev: 85)
GitHub、異常時は公開リポジトリのHTTPS取得に認証を要求 (importance: 0, dev: 70)
Apache Camel、HIGH 13件を含む38件のCVEを修正 (importance: 0, dev: 85)
Safari、OS 27で識別解決インフラへの通信を遮断 (importance: 40, dev: 45)
Chrome 155、unloadハンドラを既定で発火させない (importance: 0, dev: 80)
MiniMax、エージェント向けM3.1 FlashをAPIで提供 (importance: 50, dev: 65)
OpenAI、低資源言語の実装とCOBOL移行で精度を公開 (importance: 65, dev: 65)
Google Cloud Ships llm-d 1.0 With Cache-Aware Routing for TPU and GPU Fleets (importance: 60, dev: 85)
Exa 3.0 Search API Tops Perplexity, Brave and Parallel on 8 Benchmarks (importance: 50, dev: 75)
Meet Drex Max: Nace AI’s Multimodal Decision Model for Computer-Use Agents with a Score 67.59 (importance: 35, dev: 45)
Navalia: Advancing Cowork Agents through Verifiable Task Synthesis and Long-Horizon Post-Training (importance: 25, dev: 30)
Large language model and reinforcement learning based autonomous construction and progress updates (importance: 12, dev: 10)
Pareto-Optimal RTL Code Generation via Multi-Objective Reinforcement Learning with Large Language Models (importance: 35, dev: 70)
MedPAO-Assist: Edge-Deployable, Modality-Aware Agentic AI for Radiology Report Structuring (importance: 20, dev: 25)
DataPraxis: Scaling Analytics Tasks via Verified Workspace Evolution (importance: 30, dev: 40)
Industrial Agentic AI: LLM-Driven New-Generation Agents for Smart Manufacturing (importance: 20, dev: 20)
Fix, Not Just Resample: Bounded Compiler-Guided Repair in Lean 4 (importance: 35, dev: 60)
MAMath-Bench: Process-Aware and Robust Evaluation of Multimodal Multi-Agent Mathematical Reasoning (importance: 20, dev: 15)
LightCRS: Lightweight Multi-Domain Conversational Recommendation System (importance: 15, dev: 15)
When Planning Problems Become Hard for LLMs: Separating Structural Complexity from Semantic Interpretability (importance: 25, dev: 30)
SPEM: Structure-Aware Procedural Memory for Efficient Skill Use in LLM Agents (importance: 35, dev: 60)
Does Multi-Agent Collaboration Reduce Hallucination in Clinical Tasks? (importance: 20, dev: 15)
Teaching AI Agent to Play Poker Strategically Through Preference Optimization (importance: 15, dev: 25)
M-Plan: Comprehensive Planning For LLM Agents with Memory Arrangement (importance: 25, dev: 35)
Physical AI: Concepts, Techniques, and Challenges (importance: 20, dev: 15)
ReuF2F: Benchmarking Coding Agents on Formalized Undergraduate Research Problems (importance: 25, dev: 40)
A general-purpose System One decision model for clinical records (importance: 35, dev: 20)
Omni-Woodpecker: Hallucination Correction via Evidence-Prior Rebalancing for Omni LLMs (importance: 15, dev: 10)
PROVISO: Provenance-Informed Action Selection for Secure LLM Agents (importance: 35, dev: 50)
PORA: Learning When More Reasoning Pays Off for LLM Agents (importance: 25, dev: 35)
EDMA: Entropy-Driven Multimodal Answering (importance: 12, dev: 15)
MoSAIK: Motion Search with Artificial Intelligence in K-space (importance: 15, dev: 5)
Learned Control Variates for Rare Grid-Outage Estimation at Electric Vehicle Charging Stations (importance: 10, dev: 5)
LSA-PAST: Layer-Specific Adaptation and Prototype Transport for Compositional Incremental Learning (importance: 10, dev: 10)
Sparse RL Relaxation for LLM Reasoning (importance: 20, dev: 20)
Magnolia: A Persistent-Memory Agent Framework for Computational Chemistry (importance: 15, dev: 20)
EGI-BENCH: An Enterprise General Intelligence Benchmark Suite (importance: 30, dev: 50)
TxnQA: Benchmarking Analytical Reasoning and Evidence Grounding over Transaction Event Sequences (importance: 10, dev: 15)
A latent control model for realistic rodent motion (importance: 12, dev: 10)
TEX: Test-Time Scaling Testing Agents via Execution-based Cross-Validation (importance: 35, dev: 65)
Vulnerability Analysis for Safe Multi-Agent Reinforcement (importance: 10, dev: 5)
Stronger artificial intelligence breakthroughs hasten their own exhaustion (importance: 25, dev: 10)
When Does an Added Planning Capability Create Operational Value? An Observable Evidence Standard (importance: 12, dev: 15)
Evolving Self-tuning Benchmarks for Long-Horizon Tool Use (importance: 25, dev: 40)
Unsupervised Tool Evolution for Coding Agents in Data-Lake Analytics (importance: 30, dev: 60)
Modeling Authentic Research Trajectories: A Graph View of Research Ideation (importance: 12, dev: 10)
An LLM-Based Agent for Legal Compliance Assessment: Architecture, Memory, and Evaluation in a Canadian and Quebec Context (importance: 12, dev: 15)
REM: What to Store, When to Recall, What to Forget in Experience Memory for Coding Agents (importance: 30, dev: 55)
Correcting Attribution Patching with a Few Exact Measurements (importance: 15, dev: 15)
When PTQ Calibration Helps or Hurts: Dose-Driven Collapse and Validation-Gated Abstention in Quantized Text Embedders (importance: 15, dev: 25)
Unified Sharp Bounds on Probabilities of Causation (importance: 5, dev: 5)
ProMemAgent: Long-Horizon Proactive Assistance with Context-Aware Prospective Memory from Sensory Perceptions (importance: 15, dev: 15)
More Memory, Worse Agents: Error Reproduction and Anti-Persistence in LLM Agents (importance: 30, dev: 50)
KPHarness: Co-Evolving Knowledge and Probing for Long-Horizon Gaming Harnesses (importance: 15, dev: 15)
Harness Self-Evolution for Long-Horizon Terminal Agents (importance: 30, dev: 55)
Ask to Distill: Learning to Synthesize Data for On-Policy Distillation (importance: 15, dev: 20)
Recoverability and Verifier Strength: Two Measured Preconditions for Unattended Agent Self-Improvement (importance: 35, dev: 50)
Runtime Control for AI Agents: Closing the Evaluation--Execution Gap in Operational Workflows (importance: 25, dev: 40)
Securing Agentic Commerce Through Behavior-Aware Fraud Detection and Cybersecurity for Trusted AI-Agent Payment Authorization (importance: 12, dev: 15)
The More the Better? A Comparison of Single- and Multi-Agent LLM Setups on a Financial Review Task (importance: 20, dev: 35)
Structured-Token Attention for Fine-Grained mmWave Liquid Identification (importance: 8, dev: 5)
When Health Assistants Agree Too Readily: A Benchmark of Time-Critical Self-Diagnosis (importance: 25, dev: 5)
MemDev-Bench: Does Memory Help Code Agents?Benchmarking Memory Systems over Real Repository Histories (importance: 30, dev: 60)
Automated Multi-Agent Organization Design: From Architecture Search to Organizational Learning (importance: 20, dev: 35)
Act on Your Behalf, Beyond Your Authorization: Benchmarking LLM Agent Authorization Alignment (importance: 20, dev: 35)
RaspAssistant as an Agentic Framework for Automating Adsorption Simulation Research in Porous Materials (importance: 12, dev: 20)
When LLMs Manipulate: Measuring Behavioral, Social, and Temporal Effects (importance: 15, dev: 10)
The Show-Up Attack: When Subagents Win Over Their Supervisor (importance: 30, dev: 50)
DuplexLedger: Benchmarking Transactional Memory in Full-Duplex Speech Agents (importance: 15, dev: 20)
Beyond Experience: Knowledge-Guided Evolution for NPU Kernel Generation (importance: 20, dev: 60)
Meta Expert: Cross-Layer Expert Sharing for Storage- and Retrieval-Efficient MoE (importance: 15, dev: 25)
From Encoding to Generation: Towards Empathetic Large Language Models (importance: 12, dev: 5)
TriGeoBench: A Multimodal Mathematical Geometry Benchmark with Figures and TikZ (importance: 10, dev: 10)
When Does Feedback Help? Auditing Learned Voltage-Clamp Experimental Design (importance: 10, dev: 5)
SCODEL: Shared Canonical Ontology Definitions for Food Entity Linking and Ontology-Grounded Recipe Knowledge Graphs (importance: 8, dev: 15)
CUA-Recovery: Benchmarking Error Recovery in Long-Horizon Computer-Use Agents (importance: 30, dev: 40)
Does Multi-Agent Collaboration Pay When Agents Serve Users? Testing Agent Scaling in Dual-Control Settings (importance: 25, dev: 35)
When Can LLM Reasoning Reliably Reuse Intermediate Results? (importance: 20, dev: 35)
Inferring continuous cognitive traits from limited neuroimaging data using pairwise convolutional neural networks (importance: 8, dev: 5)
Reusing Multimodal Agents: Efficient Evaluation of Sound Designs in Games (importance: 12, dev: 15)
Temporal Evidence Allocation For Longitudinal Clinical Retrieval: A Position And Technical Report (importance: 12, dev: 20)
Pattern Intelligence: Unsupervised, LLM-Interpreted Discovery of Systematic Underpayment and Denial Patterns in Healthcare Revenue Cycle Management (importance: 15, dev: 20)
EASformer: Environment-Action-State Alignment for Industrial Generative Ranking (importance: 12, dev: 25)
E8-QAT: Efficient 2-Bit Quantization-Aware Training with Implicit Lattice Codebooks (importance: 25, dev: 50)
How Fine-Tuning Objectives Shape Layer-Wise Information in LLM Hallucination Detection (importance: 20, dev: 30)
When Does Experience Memory Help Closed-Loop LLM Driving Agents? A Controlled Evaluation with Resampled Warmup Pools, Token-Matched Fillers, and Audited Baselines (importance: 15, dev: 15)
How Language Models Organize Task Information (importance: 15, dev: 15)
RSI Control: Can AI Models Perform Covert Sabotage in the Self-Improvement Process? (importance: 40, dev: 10)
Give the Model a Finger: Developmentally Aligned Evaluation of Counting in Vision-Language Models (importance: 15, dev: 25)
CrystalAR: Autoregressive Modeling for Symmetric Crystal Generation (importance: 10, dev: 5)
RITE: Evaluating End-to-End Resilience to Silent Tool Errors in E-Commerce Agentic Systems (importance: 25, dev: 40)
GRADUS: Discovering and Recovering Agent Lessons from the Environment's Latent Syllabus (importance: 20, dev: 30)
Toward Agentic AI for the 6G Physical Layer: From Algorithm Discovery to Verified Baseband Silicon (importance: 25, dev: 50)
LatBridge: Efficient and Stable Model Replacement in Latent-Communication Multi-Agent Systems (importance: 20, dev: 30)
Privacy-preserving inference on hosted models: a learned input-embedding transform for frozen Qwen3.8-27B (importance: 45, dev: 55)
COMPASS: Routing via Capability-Oriented Model Profiling with Adaptive Search, Selection, and Tool Use (importance: 42, dev: 60)
SAPD: Learning from Agent Experience without Environment Interaction (importance: 38, dev: 50)
Physical Operating-Condition-Guided Vision-Language Matching for Cavitation Diagnosis in Industrial Pipeline Systems (importance: 15, dev: 15)
Automatic Differentiation with Constraints in Forward and Inverse Problems (importance: 20, dev: 40)
Staged Continual Adaptation of Multimodal Foundation Models for Japanese Financial Documents (importance: 35, dev: 40)
MMEnterprise-Bench: Evaluating Enterprise Agents over Evolving Multimodal Records (importance: 35, dev: 45)
ACE-Compression: An Ability-Centric Evaluation of Compressed Large Language Models (importance: 35, dev: 50)
Quantum-JEPA: Quantum Born Priors for Latent-Action JEPA World Models (importance: 25, dev: 25)
From Comparison to Selection: Enhancing Cross-Scale Generalization of Pointer Networks Trained with Reinforcement Learning for Capacitated Drone Delivery Routing (importance: 22, dev: 30)
HypTI: Hyperbolic Geometry-Aware Multimodal Feature Fusion Framework for Dementia Detection Based on Tabular-Image Data (importance: 15, dev: 25)
Lost in Recovery: Demystifying Recovery Behavior of Computer-Use Agents via Goal-Oriented Action Replay (importance: 40, dev: 55)
When Do Humans and Machines Truly Complement Each Other? Formal Conditions and Empirical Evidence for Human-AI Synergy (importance: 30, dev: 30)
Building Telecom Language Models with RSOD: Domain Data, Specialist Post-Training, Memory-Conscious Distillation, and Networking Evaluation (importance: 30, dev: 40)
Dancing Diffusion: Learning Continuous Programs for Image Generation and Editing (importance: 25, dev: 20)
TrajEvolve: Trajectory Evolution for Reinforcement Learning with Hindsight Credit Assignment (importance: 35, dev: 45)
Slot-JEPA: Object-Structured Prediction Interfaces for Latent World Models (importance: 25, dev: 20)
IECG: Weakly-Supervised Interpretability-Enhancing Foundation Model for Electrocardiogram (importance: 20, dev: 20)
InsightRAG: Aggregating Distributed Scholarly Judgments for Scientific Survey Generation (importance: 40, dev: 50)
Polyscene: Scenario Planning as Diversity Planning (importance: 20, dev: 20)
A Semi-Analytical Inference Framework for Structural Causal Models (importance: 20, dev: 30)
Robust Representation Learning for Spiking Neural Networks based on Conditional Information Bottleneck (importance: 15, dev: 15)
Interruption-Aware Recurrent Reinforcement Learning for UAS Navigation Under Temporary Overrides (importance: 20, dev: 20)
MaD-RSI: Model-as-Data for Recursive Self-Improvement of Agent Harnesses (importance: 45, dev: 70)
ChartEvo: Self-Evolving Agent Harnesses for Long-Tail Chart Reasoning (importance: 40, dev: 50)
Is Sycophancy a Form of In-Context Updating? (importance: 35, dev: 30)
S-Agent 2.0: Spatial Reasoning with a General Coding Agent (importance: 35, dev: 50)
FALCA: Falsification-Aware Reinforcement Learning with Hierarchical Credit Assignment for Long-Horizon Deep Search Agents (importance: 35, dev: 40)
RACE: RoPE-Aligned Attention Compression Based on Query-Key Activation Energies (importance: 40, dev: 65)
DIVE: Benchmarking Full-Duplex Interaction with Visual Engagement in Continuous Streams (importance: 30, dev: 30)
Stratified Sparse Attention: Reducing the Sparse-to-Dense Gap at Extreme Sparsity (importance: 40, dev: 60)
Toolflation: Seller Agents Distort Tool Markets Through Spontaneous Generative Engine Optimization (importance: 45, dev: 50)
Beyond Trajectory Statistics: Community-Consistent Human Mobility Generation via Evolving User-Location Graphs (importance: 15, dev: 15)
Claims of emergence in AI should be grounded in information decomposition (importance: 30, dev: 25)
Exploring Temporal Context-Aware Physical Grounding in Latent World Models (importance: 25, dev: 20)
Proof-Carrying Contracts: Commitments AI Agents Cannot Break (importance: 40, dev: 45)
DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems (importance: 35, dev: 50)
Beyond 4 bits: Data-Centric Post-training Quantization and Recovery (importance: 35, dev: 55)
Humans Correct Their Judgments Through Debate, but Weaker AI Models Do Not (importance: 30, dev: 15)
Agentic Ensemble with Vision-Language Models for Visual Anomaly Detection (importance: 25, dev: 35)
OptScientist: Automated Discovery of Optimization Laws (importance: 25, dev: 20)
Evolving Semantic Indexes for Generative Retrieval (importance: 30, dev: 40)
Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition (importance: 40, dev: 45)
Beyond the Sequentiality Bottleneck: A Hybrid Diffusion-Autoregressive Framework for Efficient Multimodal Reasoning (importance: 30, dev: 35)
What Do RTL Repair Benchmarks Actually Measure? (importance: 30, dev: 50)
Evolving Agent Interfaces (importance: 40, dev: 55)
Value-Rich, Differently Organized: Measuring Value Expression and Organization in AI-Generated Debates (importance: 20, dev: 15)
Semantic Observability for Autonomous Systems Diagnostics (importance: 35, dev: 60)
Adaptive Multi-Token Prediction: How far should we go? (importance: 35, dev: 55)
Large Language Models for Agent-Based Modelling: Current and possible uses across the modelling cycle (importance: 15, dev: 20)
INCARBench: A Benchmark for Scientific Configuration in VASP INCAR by Large Language Models (importance: 20, dev: 25)
Fine-Tuning Small Language Models for Reliable VASP INCAR Generation (importance: 20, dev: 30)
Aptamers as a sandbox for lab-in-the-loop agentic molecular discovery (importance: 30, dev: 10)
Measuring Consensus in AI Agent Networks (importance: 25, dev: 30)
Disentangling Depression from Cognitive Decline in Elderly Speech Using Concurrent Clinical Assessments (importance: 15, dev: 10)
PW-Pruning: Phase-Aware Expert Selection for Speech-Centric Omni-Modal Models (importance: 25, dev: 35)
MOSS: A Production Framework for Monitoring and Diagnosing Semantic Failures in Search and Recommendation Systems (importance: 35, dev: 55)
Uncertainty Gates the Gradient: Adaptive Updates for Generalized Few-Shot 3D Segmentation (importance: 15, dev: 15)
Steerable Pluralistic Policy Alignment from Anonymized Crowd Feedback (importance: 20, dev: 15)
Enhancing analytical photon transport in turbid media through physics-guided neural network corrections (importance: 12, dev: 10)
Multilingual Brain Decoding Reveals Shared Cortical Semantic Geometry (importance: 15, dev: 5)
Expert-Guided Rubrics as Dense Reward Signals for Reinforcement Learning in Non-Verifiable Domains (importance: 30, dev: 35)
How Intervention History Changes AI Agent Sabotage on Subsequent Tasks (importance: 30, dev: 30)
GRaSP: Latency-Aware Optimization for Geo-Distributed Retrieval-Augmented Generation (importance: 30, dev: 50)
Beyond Policy Discrepancy: Capability-Aware Multi-Teacher On-Policy Distillation (importance: 25, dev: 30)
FoundPDE: A generative foundation model for data-driven symbolic discovery of partial differential equations (importance: 20, dev: 15)
SZED at ArGuard Shared Tasks: Multi-Task Label-Attention and Cross-Modal Fusion for Arabic Hateful Meme Detection (importance: 10, dev: 15)
Dynamic Dual-Feedback Conformal Inference for Time Series Forecasting (importance: 20, dev: 25)
3D-MoE: Complementarity-Guided Hardware-Software Co-Design for MoE Decode on 3D-Stacked NMP (importance: 25, dev: 35)
CoASFL: Coupled Consistency Optimization for Asynchronous Split Federated Learning (importance: 15, dev: 20)
HierDTA: Hierarchical Structure-Aware Representation Learning for Drug--Target Affinity Prediction (importance: 15, dev: 10)
Exploration via Activation-Space Steering for Reinforcement Learning with Verifiable Rewards (importance: 25, dev: 30)
Reliable Estimation of Differential Privacy under Selection Bias (importance: 20, dev: 35)
BR-MoE: Budgeted Semantic Release for Privacy-Preserving Traffic Analytics (importance: 15, dev: 25)
Robustness of Internal Signals for Hallucination Detection in Vision-Language Models (importance: 25, dev: 30)
Cross-Lingual Reusability of LLM Pipelines: Financial Extraction, Multilingual Reasoning, and Semantic Screening (importance: 30, dev: 40)
Discrete Neural Interlingua for Ontology-Agnostic EHR Modeling (importance: 20, dev: 25)
Joint Constraint Aggregation for Qualifier-Sensitive Retrieval-Augmented QA (importance: 20, dev: 35)
Reconstruction-Optimized Expert Pruning for Mixture-of-Experts Language Models (importance: 25, dev: 40)
Mind Your Boundaries: Boundary-Aware Model Obfuscation against Cross-node Attack (importance: 20, dev: 35)
Ontic state recovery through prediction (importance: 10, dev: 10)
Representing short-term movement to reveal animal identity and learning progress (importance: 10, dev: 10)
ICLHerd: Herding-Based Data Selection for LLMs with In-Context Learning as a Proxy (importance: 25, dev: 40)
Model Exposure on One Device Endangers Others: Rethinking Edge Neural Network Security across Devices (importance: 25, dev: 35)
Understanding the Benefits of Scaffolds through Computational Separations (importance: 35, dev: 40)
Visual Task Modeling and Automated Planning for Humanoid Robots in Education of Individuals with Autism Spectrum Disorder (importance: 12, dev: 15)
On the Fragility of Tasks under Composition: Implicit Budgeting in Language Models (importance: 30, dev: 35)
CARLAR Influence: A Structured Hallucination Risk in Multimodal Cheque Extraction and an Exploratory Compressed-Image Countermeasure (importance: 25, dev: 30)
LoWAgent: A Low-Resource Multilingual Benchmark for Web Agents (importance: 25, dev: 40)
Language Guidance for Supervised Vision Training: An Empirical Study of Generalization (importance: 20, dev: 20)
Profile-DP: Differential Privacy for User-Controlled Profile Release in Personalized LLM Inference (importance: 25, dev: 35)
MMedAgent-Plus: Scaling Multimodal Medical Agents across Modalities, Tasks and Workflows (importance: 20, dev: 25)
AutoRedClaw: Automated Red-Teaming for Claw-Style Agents (importance: 40, dev: 50)
Target-Aware Curation of Hallucination Stress Tests for Large Language Models (importance: 20, dev: 30)
ValiAssist: Scaling LLM-generated Programming Feedback Validation by Demonstration (importance: 20, dev: 30)
When Models Know but Can't Linearly Say: Relational Encoding Limits Linear Vulnerability Classification in Code LLMs (importance: 30, dev: 55)
Theorem prover Arend (importance: 15, dev: 30)
Auditing Safety Preference Supervision with Recorded Derivations (importance: 20, dev: 20)
Certifying Repeatable Entry-Replay Witnesses for COBOL-to-Java Migration (importance: 35, dev: 60)
VPFlow: Value-Predictive Planning with Imperfect World Models for Offline Goal Reaching (importance: 20, dev: 20)
When Does Hate Transfer? A Large-Scale Study of Cross-Lingual Hate Speech Detection (importance: 22, dev: 25)
Generalizing Multimodal Foundational Models in Context-Driven Class Incremental Learning (importance: 20, dev: 30)
Evaluating Context Compaction on the Questions That Come Later (importance: 40, dev: 65)
Global Security Blind Spots: Translation as a Safety Bypass in Low-Resource Languages (importance: 35, dev: 30)
Steering Vision-Language Models for Visual-Contextual Privacy in Location Disclosure (importance: 25, dev: 25)
Refuses When Asked, Complies When Acting: A Runtime Safety Check for Robot Control Loops (importance: 40, dev: 45)
Exposing Vulnerability Monoculture Through Systemic Adversarial Robustness (importance: 25, dev: 30)
LiBRA: Logit Fingerprints for Byzantine-Robust Client Selection in Federated Learning (importance: 15, dev: 35)
On the Limits of Scale: Inverse Scaling in Vision-Language Models (importance: 25, dev: 25)
Predictive Closure: A Set-Level Characterization of JEPA's Feature Selection (importance: 15, dev: 20)
Trail-Weighted Influence Scores for Approximate Outcome Prediction in Bayesian Networks (importance: 12, dev: 25)
Fundamental Impossibility Results in AI Safety: Trading Off between Desiderata (importance: 25, dev: 20)
The Router Already Knows (importance: 25, dev: 40)
MLIPS: Multi-Layer Indexer Pool Sharing (importance: 25, dev: 65)
Scaling Graph Reasoning in Language Models with Learnable Graph Positional Encodings (importance: 20, dev: 35)
Beyond Scalar Leaderboards: Adaptive Sampling for Pareto Frontier Identification in Multi-Objective LLM Arenas (importance: 20, dev: 35)
Projected Distillation: On Learning to Solve Hard Problems from Close, Correct Targets (importance: 22, dev: 40)
Training-Free Continual Learning for Multimodal Forecasting via Prompt Evolution (importance: 20, dev: 35)
Privacy-Preserving Personalized Federated Learning (PPP-FL) for Medical Imaging (importance: 15, dev: 25)
Beyond the Prompt: Recovering Hidden Concepts in Text-to-Image Models (importance: 18, dev: 20)
Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models (importance: 25, dev: 25)
Self-Evolving Agents through Privileged Retrospective Distillation (importance: 28, dev: 50)
Cross-Visibility-Gap Identity Persistence for 4D Scene Graphs (importance: 12, dev: 20)
Progressive Subset Selection and Cascaded Consistency for Unsupervised Vision-Language Model Adaptation (importance: 15, dev: 25)
HealBench: A Fine-Grained Benchmark for Evaluating Psychological Comfort in Text-to-Image Generation (importance: 15, dev: 15)
Adversarial Benchmarks and BenchDoor Attacks: Exploiting Benchmark Incentives for Backdoor Attacks (importance: 25, dev: 35)
Less is More: Concise Retrieval and Critic-based Re-ranking for Composed Image Retrieval (importance: 12, dev: 20)
Can Native Safety Travel? Reusing Safety Pretraining Across Models via On-Policy Distillation (importance: 25, dev: 35)
AnchorRM: Parallel Intra-Batch State Reconstruction with Anchor-Based Residual Memory for Temporal Graph Neural Networks (importance: 12, dev: 35)
CityPaint: A Free Lunch from 2D Diffusion for City-Scale 3D Texturing (importance: 15, dev: 15)
AnomalyJev: Towards Evaluating Image Anomaly Explanations via Claim-Level Verification (importance: 15, dev: 25)
Learned Readout Tokens for Extracting Large Language Model Confidence (importance: 22, dev: 35)
AudiPsy: Evidence-Grounded Multimodal Reasoning over Spoken Counseling Conversations for Depression Assessment (importance: 12, dev: 15)
TRCM: Topology-Aware Route-Consistency Modeling for Robust Neural Vehicle Routing (importance: 10, dev: 20)
Factual Access States in Diffusion Language Models: From Mechanism to Knowledge Editing and Retrieval (importance: 15, dev: 25)
Benefit and Cost of Random-Token Replacement under Repeated Data: A Known-Entropy Decomposition and Evidence from Ten Real Corpora (importance: 20, dev: 40)
WORTH Benchmark: Evidence Discovery under Not-Fully-Cooperative Information -- A Two-Episode Pilot (importance: 12, dev: 20)
Auditing the Product, Not the Recursion: The Evaluation Evidence Behind Recursive Self-Improvement Claims (importance: 30, dev: 40)
ABench: Adversarily Benchmarking Large Language Models across Taxonomized Values (importance: 15, dev: 20)
Learning Inter-Graph Interactions Between Heterogeneous Substructures of Chemical Systems (importance: 12, dev: 15)
Local Allow, Global Deny: Why Per-Action Authorization Does Not Compose in LLM Agents (importance: 40, dev: 60)
A Head Camera Is (Almost) All You Need: An Electronics-Free UMI Glove Labelled and Observed by One Moving Camera (importance: 15, dev: 20)
BCT: Bayes-Consistent Tempering for Flow Language Models (importance: 18, dev: 35)
Corpus Quality Improves, Retrieval Utility Does Not: An Ablation of Preprocessing and Refinement in a Speech-to-QA Pipeline (importance: 15, dev: 35)
A Unified Attack Framework against Statistical Generative Watermarks for Large Language Models (importance: 30, dev: 30)
Pairwise Learning Increases Data Efficiency in Deep Neural Networks (importance: 15, dev: 25)
CovNav: Conformal Prediction-Guided Adaptive Inference for Object Goal Navigation (importance: 12, dev: 25)
VLA-Retain: Benchmarking Cross-Dimension Effects of Vision Tuning in VLAs (importance: 12, dev: 20)
Beyond Calibration: Temporal LLM Confidence Assessment Across Languages and Local Settings (importance: 18, dev: 30)
RLSplat: Resource-Linear Budget Control for Feed-forward 3D Gaussian Splatting (importance: 12, dev: 20)
World2HOSI: A Multi-Expert Harness for Long-Horizon Human-Object-Scene Interaction Generation (importance: 12, dev: 20)
Anytime-Valid Runtime Monitoring of Learned Barrier Certificates (importance: 12, dev: 20)
Hearsay or Fact? How Framing and Elicitation Shape LLM Probability Updates (importance: 15, dev: 25)
S$^2$-MAS: A Society-Scale Multi-Agent Simulation Framework with 500 Million Household Persona Agents (importance: 20, dev: 35)
How Much Do $\ell_1$ Attacks Need Concentrated Perturbations? (importance: 10, dev: 20)
Radar-Based 3D Human Pose Estimation via Motion-Aware Pose Initialization and Observation-Guided Refinement (importance: 10, dev: 15)
RELAY: Scaling Historical Build Restoration through Rule Reuse and Agentic Repair (importance: 35, dev: 65)
SLM-SS: Token-Based Speech Separation with Speech Language Models (importance: 12, dev: 20)
RAG-NIMBF: A Non-Invasive, Modular Benchmarking Framework for Automated RAG Configuration Analysis (importance: 25, dev: 55)
SMART: Semantic Segment Mapping and Assessment of Recognition Transcripts with Contextual Rubrics (importance: 12, dev: 20)
VISH-ASR: Viterbi-Guided Selection of Hotwords for Speech-LLM Contextual ASR (importance: 12, dev: 25)
Hallucinations of Large Language Models in Cross-Language Code Translation: Phenomena and Mitigation (importance: 30, dev: 70)
Physical-Variable-Informed Deep Learning for Robust Diagnosis in Industrial Fluid Transport Systems (importance: 10, dev: 15)
The Shape of the DPO Hyperparameter Landscape: What to Measure, How It Moves with Scale, and When Not to Transfer It (importance: 25, dev: 45)
CollabSciReM: Aligning Language Models for Scientific Conversations via Science-Aware Reward Decomposition (importance: 15, dev: 30)
Polarity Blindness in LLM Safety: A Systematic Study of Negation, Antonym Substitution, and Adversarial Framing (importance: 25, dev: 30)
APRIL: Ambiguity-Aware Expert Routing for Respiratory Sound Domain-Incremental Learning (importance: 10, dev: 15)
Recast the Past: Stable Long Video Generation with Full Context Memory (importance: 12, dev: 15)
Learning Tactile-Aware Hand Representations from Human Hand-Object Interaction Data (importance: 12, dev: 15)
On-Policy Transition Distillation (importance: 18, dev: 35)
Assessing Representation Semantics Under Language Pretraining in Fine-Grained Classification (importance: 10, dev: 20)
Slots Are Not Speakers: Training-Free Speaker Linking for Frozen Streaming Diarizers (importance: 12, dev: 30)
Auditing Where Concept Bottleneck Accuracy Comes From on CUB (importance: 12, dev: 15)
Quantifying the Threat: A Bayesian Probabilistic Framework for Heavy Metal Health Risks among Children in the Densu Basin, Ghana (importance: 8, dev: 10)
Same Judgment, Different Yes: Measuring Multilingual Judgment Profiles in Language Models (importance: 15, dev: 25)
PULSE: Personalized Opportunistic Learning for Wearable Devices (importance: 10, dev: 20)
MoDDiT: Simple Token Routing in Diffusion Transformers (importance: 12, dev: 25)
From Complex Histories to Simple Predictors: Predictive Dimension in Dynamic Graph Learning (importance: 12, dev: 30)
BAIT: Can LLMs Find What Biases a Reader? A Benchmark for Cognitive Bias Triggers in News (importance: 20, dev: 25)
ClawMMU: Protection-Coherent Virtual Memory for Stateful LLM Agents (importance: 35, dev: 55)
De-Diffusing WAMs: Action-Sufficient Direct Prediction for Robot Control (importance: 12, dev: 20)
Causal Discovery for Linear Models with Latent Confounding via Multi-domain Identifiable Representation Learning (importance: 10, dev: 20)
From Code Generator to Engineering Companion: A Plant-aware Agent for PLC Programming (importance: 20, dev: 40)
Africa-Papers: A 10-Billion-Token Multi-Domain Academic Corpus for African Foundation Language Models (importance: 15, dev: 25)
Temporal Evidence-Parameter Co-Alignment for LLM Reasoning over Temporal Knowledge Graphs (importance: 15, dev: 25)
Perceives but Doesn't Brake: A Perception-Action Binding Gap in Reasoning Driving VLAs (importance: 15, dev: 20)
Offline Makes Online Better: Calibrated Cache Compression for Video Generation (importance: 15, dev: 35)
Adaptive Pullback: Reweighting Supervision at Penalty Jumps in On-Policy Distillation (importance: 18, dev: 40)
Reframing Object-Centric Reinforcement Learning: From Absolute to Relative Representations (importance: 10, dev: 20)
Parameter-Efficient Adaptation of Time Series Foundation Models for Electrical Load Forecasting (importance: 12, dev: 25)
JAXbounce: A Continuous-Control Pong-Like Benchmark for Active Inference Beyond Low Dimensional Environments (importance: 10, dev: 30)
Region-Guided Food-Package Evidence Extraction and Abstention-Aware Draft R3337 Screening with Selective VLM Review (importance: 12, dev: 25)
Local Inference of LLMs on Long Contexts: A Data Systems Perspective (importance: 28, dev: 55)
TPO: Tri-level Distributionally Robust Learning for OOD Direct Preference Optimization (importance: 15, dev: 35)
PI-MLLM: Psychophysics-Inspired Multimodal Large Language Model for Oddly Satisfying Video Emotion Analysis (importance: 10, dev: 15)
VGL-stGen: Learning a shared multimodal model for lung histology and spatial transcriptomics (importance: 15, dev: 15)
GALA: Geometry-Aware Language Model for Controllable Object Arrangement (importance: 10, dev: 15)
Local Work, Global Progress: A Convergence Theory of AI Agent Collaboration (importance: 25, dev: 45)
ABACUS: A Benchmark for Image Editing Models to Assess Object Category and Counting Under Scene Conservation (importance: 12, dev: 20)
Identifying Dependent Components from Multi-domain Linear Mixtures (importance: 10, dev: 15)
Beyond a Single Sentiment Score: Topical and Affective Dimensions of Emoji Meaning on Twitter (importance: 15, dev: 35)
Gated Prototype Transfer for Zero-Support Medical Vision Language Model (importance: 20, dev: 25)
RayCo4D: Ray-Conditioned Continuous Motion Model for Monocular 4D Dynamic Reconstruction (importance: 18, dev: 20)
CountVideoBench: A Benchmark to Evaluate Prediction Biases on the Object Counting task by Video-Language Models (importance: 22, dev: 35)
Context-Aware Prompt Injection Detection via Small Language Models in RAG Pipelines: nokast-secureRAG (importance: 30, dev: 60)
Don't Skip the Baseline: A Controlled Comparison of Dynamic Expert Skipping in MoEs (importance: 25, dev: 50)
VisualEvolve: A Skill Self-Evolution Framework for Visual Reasoning (importance: 25, dev: 40)
MAS-Guard: Unified Guardrails from Individual Agents to Multi-Agent LLM Systems (importance: 30, dev: 60)
The Instruction Tax on Factual Recall: Base Models Can Beat Their Instruction-Tuned Counterparts (importance: 25, dev: 45)
Rewards for Budget-Aware Multi-Turn Agents That Adapt to Arbitrary Test-Time Constraints (importance: 28, dev: 55)
Brain2Vec: A Self-Supervised 3D Foundation Model for Structural Brain MRI (importance: 18, dev: 20)
Code2Video-Compass: A Comprehensive Benchmark for General-Purpose Code-Based Video Generation across T2V, I2V, and V2V (importance: 20, dev: 40)
DISTRACT: Degradation-Indexed Study of Textual Reference Across Competing Targets (importance: 15, dev: 25)
WorldSR: Harnessing World Knowledge Search for Grounded Image Super-Resolution (importance: 18, dev: 20)
After the Attacker Leaves: Benchmarking Carryover Attacks on LLM Agents (importance: 35, dev: 65)
Racer: Hunting GPU Kernel Bugs with Symbolic Equivalence Checking (importance: 30, dev: 75)
A Digital Twin to Improve Cyber-Social Impact (importance: 12, dev: 25)
Plausible Future, Wrong Choice: Steering World-Action Models toward Intended Future (importance: 15, dev: 20)
Every Model Leaves a Trace: Multi-Resolution LLM Fingerprinting (importance: 18, dev: 35)
Model- and Layer-Level Closed-Loop Forecasting with SSM-Based Spiking Neural Networks (importance: 12, dev: 25)
BitMTP: When Multi-Token Prediction Meets Low-Bit Large Language Models (importance: 25, dev: 60)
ECG for laboratory values:A comprehensive physiological and technical analysis (importance: 12, dev: 15)
VLM-Guided Contact Retargeting and Generative Partner Modeling for Humanoid-Human Interaction (importance: 15, dev: 15)
Binary Discrete Diffusion: Candidate-Set Diffusion for Logic and Planning Tasks (importance: 18, dev: 35)
Learning Computational Structure for Sample-Efficient Prediction (importance: 15, dev: 25)
When Do LLMs Help with Feature Selection? A Systematic Empirical Study (importance: 20, dev: 40)
Large Flow Language Models (importance: 25, dev: 40)
Spatially Informed and Direction Sensitive Representation Learning for Spatial Transcriptomics (importance: 12, dev: 20)
SceneRig: An Agentic System for Simulation-Ready 3D Scene Reconstruction from Single Images (importance: 18, dev: 30)
Capable Teacher, Myopic Scorer: Reward-Free Teacher Repair for Agentic On-Policy Distillation (importance: 25, dev: 55)
Turning TS2Vec into a Foundation Model for Time Series Classification (importance: 15, dev: 30)
Benchmarking Arabic Dialects for Math Problems (importance: 12, dev: 15)
MindView: Benchmarking Generative Multi-View Spatial Intelligence (importance: 12, dev: 15)
Industrial ReMEmbR: Scaling Spatiotemporal Memory to Robot Fleets (importance: 18, dev: 35)
UniWALM: Unified World Action Language Model for Embodied Robot Control (importance: 15, dev: 20)
GBClaw: An agentic AI framework for atomistic grain boundary research and mechanistic discovery (importance: 18, dev: 30)
Joints, Camera, Action! Co-Generating 3D Motion and Video for Arbitrary Rigs (importance: 15, dev: 20)
Gibson Env: Real-World Perception for Embodied Agents (importance: 10, dev: 20)
GenoMAS: Guided Multi-Agent Programming for Scientific Discovery from Genomic Data (importance: 15, dev: 35)
How Far Are LLMs from Symbolic Planners? An NLP-Based Perspective (importance: 18, dev: 35)
LLM-Guided Evolution of Symbolic Acceleration Inequality Families for MILP (importance: 15, dev: 30)
Single Agent Robust Deep Reinforcement Learning for Bus Fleet Control (importance: 10, dev: 20)
EMMA: Enhancing Robotic Control Robustness through a Generative Visual Model (importance: 12, dev: 15)
VRPAgent: LLM-Driven Discovery of Heuristic Operators for Vehicle Routing Problems (importance: 12, dev: 30)
Mitigating Model Merging Conflicts via Knowledge-Aware Task Vector Purification (importance: 14, dev: 40)
The Reward Model Selection Crisis in Personalized Alignment (importance: 12, dev: 30)
Agents-as-Peers: Bias-Controlled LLM Evaluation as Multi-Agent Peer Assessment (importance: 18, dev: 40)
Tabular Foundation Models Can Learn Association Rules (importance: 12, dev: 30)
From Debate to Decision: Conformal Social Choice for Safe Multi-Agent Deliberation (importance: 14, dev: 35)
Non-Stationarity Breaks Permutation Surrogates in Multi-Agent Reinforcement Learning: Diagnosis and Remedies (importance: 10, dev: 20)
Mental Health AI Safety Claims Must Preserve Temporal Evidence (importance: 15, dev: 20)
Learning Transferable Latent User Preferences for Human-Aligned Decision Making (importance: 12, dev: 25)
Beyond a Single Direction: Chain-of-Thought Disrupts Simple Steering of Refusal (importance: 15, dev: 35)
PReMISE: Policy Rubrics as Measurement Specifications for LLM Judges (importance: 18, dev: 50)
Who Can Change Course? An Integrative Review of AI Sovereignty and Collective Learning Capacity (importance: 10, dev: 10)
The Shape of Wisdom: How Language Models Converge on an Answer (importance: 12, dev: 25)
Teaching the Way, Not the Answer: Privileged Tutoring Distillation for Multimodal Policy Optimization (importance: 12, dev: 25)
Agentic vs. Parametric World Models: Hybrid Planning for Reliable Language Agents (importance: 18, dev: 45)
Ask the World Before Acting: Environment Probing for Calibrating Agent World Models (importance: 18, dev: 45)
Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents (importance: 18, dev: 45)
AIMO Interpretability Challenge (importance: 15, dev: 30)
IDEAgent: Agentic Quality-Diversity Search for Research Idea Generation (importance: 12, dev: 20)
Coordinated Networking for On-Device Agent-Augmented Real-Time Communication (importance: 14, dev: 35)
IndustryForge-27B: A Domain-Enhanced Multimodal Foundation Model for Industrial CAD (importance: 14, dev: 30)
ECHO: An Auditable Dual-Path Memory System for Long-Horizon Agents (importance: 22, dev: 50)
Ethical LLM-Assisted Research: A Framework for Responsible Delegation, Verification, and Epistemic Value (importance: 12, dev: 20)
Differential Functionalism: Jacobians Shape the Structure of Phenomenal Experience (importance: 8, dev: 5)
Unified Agentic Video Editing Across Levels of Complexity and Creativity (importance: 12, dev: 25)
ER-EDF: A Psychology-Grounded Emotion Regulation Framework for Speech Empathetic Dialogue Generation in Large Audio-Language Models (importance: 10, dev: 20)
Beyond Episodic AI: Cognitive Field Networks for Biologically Inspired Persistent Cognition (importance: 10, dev: 25)
AtomWorld-Mem: Memory-Restored World States for Long-Horizon Atomistic Evolution (importance: 12, dev: 15)
What Does the Rank Buy? A Spectral and Distributional Analysis of Low-Rank Adaptation (importance: 18, dev: 50)
APOLO: Automatic Prompt Optimization for Ontology Learning (importance: 12, dev: 35)
Diffusion Subgoal Planning for Long-Horizon Offline Goal-Conditioned Reinforcement Learning (importance: 10, dev: 20)
Reasoning with Continuous Latent Diffusion (importance: 18, dev: 35)
Penumbra: Sample-Efficient Adversarial Search for Regulatory Obligations (importance: 15, dev: 35)
Not Self-Decidable: LLMs Cannot Draw the Boundary of What an Agent Verifier Can Check (importance: 16, dev: 45)
MultiWorldBench: Do Independently Controlled Views Describe One Shared World? (importance: 10, dev: 20)
FDE-Bench: Evaluating End-to-End Delivery from Underspecified Real-World Business Requests (importance: 15, dev: 40)
What Survives When Activation Steering Is Used as Memory (importance: 12, dev: 30)
MADDIE: Multi-Agent Drug Discovery Engine (importance: 14, dev: 25)
Adaptive-Group Process Rewards for GUI Agents (importance: 15, dev: 35)
Thinker-WM: A Unified World Model with Shared Multimodal Representations for Embodied Intelligence (importance: 12, dev: 20)
AgentStripes: An Evidence-Graded Maturity Model for AI Agents (importance: 20, dev: 50)
Recognition Is Not Grounding: Benchmarking Multilingual Financial Evidence Extraction (importance: 12, dev: 25)
Memorizing Who, Knowing How: Benchmarking Personal and Procedural Memory for Persistent Assistants (importance: 18, dev: 45)
Beyond Large Language Models: A Mental World Model Framework for Mental Modeling, Simulation, Evolution, and Interpretability (importance: 10, dev: 10)
AM-Bench: A Unified Taxonomy and Evaluation Suite for Agentic Misalignment (importance: 18, dev: 40)
HABIT-Bench: Evaluating Habit Induction from Longitudinal Weak Evidence in Agent Memory (importance: 14, dev: 35)
HSPR: Recovering Software-Agent Benchmark Scores via Hierarchical Posterior Inference (importance: 18, dev: 50)
VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse (importance: 12, dev: 15)
From Proposal to Execution: Modeling Interaction Semantics in Shared Autonomy (importance: 10, dev: 15)
VibeTraining-Bench: Data-Trainer-Harness Co-Design for Iterative System Improvement (importance: 22, dev: 55)
Revision-Based Learning from Interactive Experience and Feedback (importance: 18, dev: 55)
From Outcomes to Components: A Counterfactual Evaluation Framework for LLM-Proposed Repairs (importance: 14, dev: 40)
Taskonomy: Disentangling Task Transfer Learning (importance: 8, dev: 10)
Typhoon: Gradient-Based Task-Specific Masking for Masked Language Models (importance: 8, dev: 20)
Entropic perspective on AI resilience, data-minimal learning and optimal control (importance: 6, dev: 10)
Jacobian Descent for Multi-Objective Optimization (importance: 10, dev: 25)
Deep learning-based multi-level airway collapse site localization on obstructive sleep apnea patients based on snoring sounds (importance: 8, dev: 5)
Help Me Help You: The Aggregate Value of Source and Target Data in Transfer Learning (importance: 15, dev: 10)
Synergistic Action and Style Priors for Learning Legged Locomotion (importance: 15, dev: 15)
PALATE: Peculiar Application of the Law of Total Expectation to Enhance the Evaluation of Deep Generative Models (importance: 15, dev: 20)
MedHal: a Synthetic Dataset for Medical Hallucination Detection (importance: 25, dev: 25)
A Survey of Generative Categories and Techniques in Multimodal Generative Models (importance: 15, dev: 10)
WoW-Bench: Evaluating Fine-Grained Acoustic Perception in Audio-Language Models via Marine Mammal Vocalizations (importance: 15, dev: 10)
Attention-Aligned Reasoning for Large Language Models (importance: 25, dev: 25)
Bayesian Nonparametric Dynamical Clustering of Time Series (importance: 10, dev: 10)
Online Continual Learning on Intel Loihi 2 via a Co-designed Spiking Neural Network (importance: 15, dev: 10)
SpectralTrain: A Universal Framework for Hyperspectral Image Classification (importance: 10, dev: 15)
From Crawled to Published: A Publisher-Controlled Web Architecture for an AI-Native Internet (importance: 25, dev: 45)
When does Subgroup Balance Matter? Predicting Performance Sensitivity from Latent Separability (importance: 10, dev: 10)
OPTIMA: Optimal One-shot Pruning for LLMs via Quadratic Programming Reconstruction (importance: 30, dev: 40)
Reinforced Linear Genetic Programming (importance: 10, dev: 15)
Activation Sensitivity as a Unifying Principle for Post-Training Quantization (importance: 25, dev: 35)
Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling (importance: 15, dev: 10)
TTCS: Test-Time Curriculum Synthesis for Self-Evolving (importance: 20, dev: 15)
Graph-Theoretic Analysis of Phase Optimization Complexity in Variational Wave Functions for Heisenberg Antiferromagnets (importance: 5, dev: 5)
CHIMERA: Compact Synthetic Data for Generalizable LLM Reasoning (importance: 25, dev: 20)
Evaluation format, not only model capability, shapes measured triage failure in consumer health AI (importance: 20, dev: 10)
Understanding and Defending VLM Jailbreaks via Jailbreak-Related Representation Shift (importance: 15, dev: 10)
Anchoring and Rescaling Attention for Semantically Coherent Inbetweening (importance: 10, dev: 5)
Detection Is Cheap, Routing Is Learned: Why Refusal-Based Alignment Evaluation Fails (importance: 30, dev: 10)
Interpretable PM2.5 Forecasting for Urban Air Quality: A Comparative Study of Adaptive Time-Series Models (importance: 8, dev: 15)
Abstract Compression of Conversational Audio Context for LLM-based ASR (importance: 20, dev: 20)
LLaVA-CKD: Bottom-Up Cascaded Knowledge Distillation for Vision-Language Models (importance: 15, dev: 20)
Dynamic Cross-Modal Prompt Generation for Multimodal Continual Instruction Tuning (importance: 12, dev: 15)
CounterCount: Diagnosing Knowledge Bias in How Vision-Language Models Count Attributes (importance: 12, dev: 5)
Dynamic Model Merging Made Slim (importance: 20, dev: 30)
DEFLECT: Temporal Counterfactual Preference Learning for Delay-Robust Asynchronous VLAs (importance: 15, dev: 10)
Rethinking Cross-Layer Information Routing in Diffusion Transformers (importance: 15, dev: 15)
NVIDIA Spectrum-X: Hardware-Accelerated Multiplane Networking for Large-Scale AI Training (importance: 30, dev: 45)
What a One-Hour Interview Does Not Tell an LLM About You: Interviews Elicit Traits, Not Decision Policies (importance: 20, dev: 10)
Same Evidence, Different Answers: Canonical-Context On-Policy Distillation for Multi-Turn Language Models (importance: 20, dev: 20)
Flow Control: Steering Vision-Language-Action Models with Simple Real-Time Inputs (importance: 15, dev: 10)
Service-Induced Congestion in Memory-Constrained LLM Serving (importance: 25, dev: 45)
KITE: Decoupling Kinematics and Interaction for Zero-Shot Cross-Embodiment Manipulation (importance: 12, dev: 10)
Yuvion VL: A Multimodal Foundation Model for Adversarial Content and AI Safety (importance: 25, dev: 15)
For Coding Agents, the Repository Matters (importance: 55, dev: 80)
Probe, Don't Prompt: A Hidden-State Probe for Metadata Filtering in Multi-Meta-RAG (importance: 12, dev: 25)
When Synthetic Speech Is All You Have: Better Call GRPO (importance: 15, dev: 10)
DiffEEG: A Self-Supervised Denoising Diffusion Model for Learning EEG seizure-related Representations (importance: 10, dev: 10)
DECODEM: Data Extraction from Corporate Organizational Documents via Enhanced Methods (importance: 15, dev: 20)
Reducing belief in conspiracy theories as they unfold using large language models (importance: 20, dev: 5)
Harnessing Magnitude-Only and Complex Measurements for Improved Dynamic MRI Reconstruction with Learned Priors (importance: 8, dev: 10)
DA-WAM: Decision-Aligned Future Latents for Driving World Models (importance: 15, dev: 10)
Difference-in-Differences on a Censored Rating Scale Can Manufacture an Effect: Evidence from a Pre-Registered LLM-Judge Audit (importance: 20, dev: 35)
Symmetries and Causality: Causal Effect Identification Beyond IID Data (importance: 8, dev: 5)
When Models Edit Too Much: On the Fidelity of Minimal Code Edits (importance: 35, dev: 75)
Knowledge Acquisition During Pre-training? Large Language Models Learn Better With Auxiliary Views (importance: 15, dev: 10)
An Empirical Study of Development-Tuned Representation Sensitivity for Medical QA Data Selection (importance: 12, dev: 25)
Decentralized Optimal Equilibrium Learning Over Dynamic Networks (importance: 5, dev: 5)
EMERGE: Resolution-Agnostic Point Cloud Generation with Equivariant Graph-Based Diffusion (importance: 8, dev: 5)
OPIS: An Input-Grounded Benchmark for Multi-Object Memory in Video World Models (importance: 12, dev: 5)
Learning Native Reflection in Unified Models with Interleaved Reinforcement Learning (importance: 12, dev: 10)
Binarization Flattens the Score Space (importance: 15, dev: 35)
Constructing Challenging Browser-Use Tasks by Controlled Environment Interventions (importance: 20, dev: 35)
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision (importance: 15, dev: 15)
$S^3$: Spectral Null-Space Swap Makes Reasoning Models Efficient (importance: 15, dev: 15)
Retargeting Motions to Diverse Skeletons via Learnable Flattening (importance: 5, dev: 5)
Exploring More, Reasoning Better: Stepwise Risk-Sensitive GRPO for Diffusion Language Models (importance: 8, dev: 5)
FERPO: Forward Entropy-Regularized Policy Optimization (importance: 8, dev: 5)
BridgeCast: Bridging Ocean Wave Forecasts to Reanalysis via Flow Matching with Exogenous Variables (importance: 8, dev: 5)
EDISCO: Equivariant DIScrete Diffusion for Euclidean Combinatorial Optimization (importance: 8, dev: 5)
TAPDreamer: Transferable Adversarial Patches for World Action Models (importance: 10, dev: 5)
Composing What Each Teacher Learned: Delta Distillation for Multi-Teacher On-Policy Learning (importance: 15, dev: 15)
CRISP: Fixing Flying Pixels in Latent LiDAR Generation via Diffusion Decoding (importance: 8, dev: 5)
WeGR: Towards Unified Generative Retrieval via Behavior-aware Tokenization and Preference Alignment (importance: 10, dev: 10)
StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions (importance: 15, dev: 15)
Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents (importance: 25, dev: 50)
A Global View of How LLMs Value Human Time (importance: 10, dev: 5)
Falsification-proof non-market allocation mechanisms (importance: 5, dev: 5)
ELARSpeech: A 30K-Hour Dataset Across 675 Underrepresented Languages for Long-Tail Multilingual Speech Technologies (importance: 10, dev: 10)
Does On-Policy Distillation Need to Complete Every Rollout? (importance: 20, dev: 25)
Training-Free On-Policy Distillation (importance: 20, dev: 20)
Knowledge That Is Linear Before Fine-Tuning Generalizes After It (importance: 12, dev: 10)
Batched Contextual Reinforcement II: Does RLVR Need Long Rollout Windows? (importance: 12, dev: 10)
A Stronger Teacher Within: Strengthening MoE Models via On-Policy Bidirectional Distillation (importance: 12, dev: 10)
Learning Administrative Boundary-Aware Location Encoders via Reverse Geocoding (importance: 8, dev: 10)
HumanSkill-4D: Training-Free Language Querying of Animatable Gaussian Avatars in Space and Time (importance: 8, dev: 5)
Muse Glimmer 30B: An On-Device Agentic Model (importance: 40, dev: 55)
Stable Signer V3: 4D Deformable Primitive Splatting Is All You Need for Sign Language Production (importance: 8, dev: 5)
Multimodal-Conditioned Latent Diffusion Models for Spatiotemporal Brain MRI Forecasting (importance: 8, dev: 5)
Secure Agentic Coding through Counterexample-Grounded Feedback (importance: 30, dev: 75)
Policy Search in Code Space for Multi-Agent Pathfinding (importance: 20, dev: 40)
TamedFlow: Trajectory-Aware Distillation for Teacher-Assisted Diffusion Sampling (importance: 8, dev: 5)
Rethinking LLM Fine-Tuning via Weight Space Reparameterization: Preserving Safety during Downstream Adaptation (importance: 15, dev: 15)
WeAct: Dynamic Weight Switching as Agent Actions (importance: 20, dev: 30)
Web Agent Agentic Reinforcement Learning Decision Model Under Multi-Cost and Failure Risk (importance: 15, dev: 25)
PolyGraph: Plan-Buffered CUDA Graphs for Adaptive Speculative Decoding (importance: 15, dev: 55)
Forecasting Patient Trajectories from the Electronic Health Records with Action-Conditioned World Models (importance: 8, dev: 5)
SAGE-CP: When Marginal Coverage Hides Route-Conditional Failures in Adaptive VLM Ensembles (importance: 10, dev: 10)
Announced Breaks Separate Conformal Reliability from Frequency Calibration (importance: 5, dev: 5)
Zero-Shot Tabular Learning via Hypergraph Reasoning with LLMs (importance: 10, dev: 10)
Improving Fake Profile Detection Through Multimodal Learning and Research-Guided Feature Integration (importance: 8, dev: 5)
miniTE: Diagnosing and Bridging the Execution Gap in Native NVFP4 Reinforcement Learning (importance: 15, dev: 35)
Organizing LLM Inference for Aggregate Predictions from Behavioral Data (importance: 10, dev: 15)
Factorization-Guided Training and Inference for Diffusion-Drafted Speculative Decoding (importance: 15, dev: 35)
Beyond Prediction: A Component-Level Study of Concept Recovery in In-Context Learning (importance: 10, dev: 10)
SECURE-AI: Multi-Signal Uncertainty Quantification for Trustworthy Acoustic Cough Detection (importance: 8, dev: 5)
Pack and Extract: An Effective Graph-Based Approach for Feature Selection (importance: 15, dev: 25)
PeekKV: The First Content Token Already Knows Which Heads Matter (importance: 30, dev: 85)
Fast Construction of Inner Sphere Trees Using Machine Learning (importance: 15, dev: 35)
A Group-Aware Comparative Study of Classical Machine Learning and Deep Learning Methods for Sensor-Based Human Activity Recognition (importance: 20, dev: 30)
ReTTT: Reflection-Driven Test-Time Training via Experience-Based Parametric Memory (importance: 35, dev: 60)
How Data Scales in Agentic Reinforcement Learning: Laws and Synthesis Strategies (importance: 40, dev: 50)
Parameter misspecification in physics-informed neural networks: A closed-form audit and its limits (importance: 15, dev: 25)
DS-Net: A Dynamic Split Network for Edge-Cloud Collaborative Inference (importance: 25, dev: 45)
Compaction-Aware Reinforcement Learning For Ascend Triton Kernel Generation (importance: 35, dev: 70)
LatLM: Line-Parallel Language Models with Deferred Commitment (importance: 25, dev: 35)
DISCERN: Identifiable Disease Evidence for Generalizable Medical Multimodal Learning (importance: 15, dev: 15)
Physics-Aware Self-Supervised Pretraining on Multi-Domain Raman Spectra (importance: 12, dev: 15)
UltraData-Code: From Raw Repositories to Algorithmically Dense and Task-Oriented Code Data (importance: 30, dev: 65)
SPADE-KD: Peer-to-Peer Collaborative Learning under Heterogeneous Settings (importance: 20, dev: 30)
TWIN-MoE: Two-stage Intra-Expert Imputation Network for Incomplete Multimodal Learning (importance: 22, dev: 30)
Dynamical Kernel Attention: Estimating the Present from a Record of the Past (importance: 25, dev: 35)
Do Knowledge and Reasoning Oriented Benchmarks Induce Stable Representation Signatures in LLMs? (importance: 22, dev: 30)
Visual Causal-chain Bookmarking: Retrieval-Weighted Credit Assignment for Reinforcement Learning from Delayed Reward (importance: 18, dev: 25)
Client-Aware Security Adaptive Early Exit Vision Transformers for Heterogeneous Federated Learning (importance: 15, dev: 20)
What Remains Without Test Labels? Re-evaluating a Graph-Level Anomaly Detection Benchmark (importance: 18, dev: 25)
Alliance of Stochastic Routing and Reinforcement Fusion for Multi-Model Reasoning (importance: 22, dev: 35)
GLOW: General Linear Models Optimized with Ward's Method (importance: 12, dev: 15)
Looped GNNs Beyond Fixed Depth: Graph Size Generalization and Parameter-Efficient Adaptation (importance: 20, dev: 30)
Graph-Regularized Synthetic Control for Stable Counterfactual Estimation under Donor Redundancy (importance: 12, dev: 15)
On the Selectivity of Generative Models in Structure-Based Drug Design (importance: 20, dev: 15)
Data-Driven Resource Aggregation for Scalable Digital Twin and Simulation Model Generation (importance: 15, dev: 30)
Inside the Loop: A Mechanistic Study of Weight-Tied Transformers on Depth-Bound Algorithmic Tasks (importance: 28, dev: 35)
Continual Test-Time Learning with Evolution Strategies (importance: 25, dev: 35)
Budgeted Bayesian Bandits for Future Utilization (importance: 12, dev: 15)
Margin Certificates for Voting Classifiers under Mixing Dependence (importance: 10, dev: 10)
Bootstrap Stochastic Interpolant Tree for Inference-time Search Under Sparse Reward (importance: 20, dev: 25)
On Reparameterizing the Score Function (importance: 25, dev: 35)
Hidden Trajectories Don't Lie: Correcting Corruption Drift for Test-Time Adaptation (importance: 18, dev: 25)
Dynamic Neural Optimal Transport for Subset Alignment (importance: 12, dev: 15)
Dataset Inference for Data Provenance and Privacy Auditing in Tabular Foundation Models (importance: 25, dev: 25)
An important Bottleneck in LLM Optimizers: Principle-Level Local Optima (importance: 25, dev: 35)
Inducing Fourier Geometry: A Smoothness Prior Imposes Approximate Translation Equivariance in Multimodal Stellar Embeddings (importance: 8, dev: 10)
Constructive Universal Approximation by Infinite-Dimensional Neural Networks via Gaussian-Weighted Ridgelet Transform (importance: 8, dev: 5)
Predicting CVSS and CWE from CVE Descriptions with Fine-Tuned Small Language Models (importance: 30, dev: 50)
Torch-DPhys: A Torch-Native, MOOSE-Compatible Framework for Differentiable Physics Simulation and Hybrid Machine Learning Integration (importance: 22, dev: 70)
Muon Operates Beyond the Edge of Stability (importance: 25, dev: 30)
Latent State Utilization in Probabilistic Circuits (importance: 15, dev: 20)
Morpheus-2.0: Multiscale Probabilistic Forecasting with Dual-Axis Attention (importance: 22, dev: 30)
PAC Learning with Query Feedback (importance: 10, dev: 8)
Power-Law Correlations Without Apparent Temperature-Tuned Criticality in Fixed-Tree Hierarchical Languages (importance: 12, dev: 10)
Uniform convergence at optimal rate and scale (importance: 8, dev: 5)
The Certification Exponent: Instance-Optimal Sequential Certification of Policy Improvement from History Dependent Logs (importance: 15, dev: 20)
Measuring the Task Value of Precision Upgrades in Closed-Loop Agents (importance: 25, dev: 35)
Exact Risk Certificates for Pairwise-Judged Reranking (importance: 12, dev: 20)
Revisiting Smoothed Online Learning: Competitive Ratio, Dynamic Regret with Switching Cost, and Best of Both Worlds (importance: 8, dev: 5)
Cross-Dimension and Cross-Modality Relational Tensor Representation Learning for Multimodal Health Data Modeling (importance: 12, dev: 10)
Transport What Shares the Conduction Pathway: Normal-Anchored SPD Geometric Transport for Source-Free Patient Adaptation of Holter and Patch ECG (importance: 12, dev: 10)
Transformers as Adaptive Kernel Learners across Tasks for In-Context Learning (importance: 15, dev: 10)
Near-Optimal Regret for Semi-Bandit Learning of Monotone Stochastic Problems (importance: 8, dev: 5)
Generative Retrieval for Reverse Target Fishing with a Proteome-Scale Benchmark (importance: 12, dev: 10)
Measuring Empowerment from Offline Data for Intrinsically Motivated Exploration (importance: 15, dev: 25)
VBO-MI: A Differentiable Black-Box Optimization Framework via Variational Mutual Information Estimation (importance: 12, dev: 15)
CoTs remain monitorable when being unfaithful requires computation: Replicating and Extending the Necessity Effect Across Eleven Models (importance: 30, dev: 30)
TideKV: Delayed KV Eviction via Layer-Wise Chunked Prefill for Memory-Efficient Long-Context LLM Inference (importance: 25, dev: 55)
HiPPO Functional Memory: Coordinate-Queryable Attention for Irregular Time Series (importance: 15, dev: 20)
BEM: Learning Bregman Divergences for Concept Erasure in Flow Matching Models (importance: 15, dev: 15)
What "Form'' of Uncertainty Does Conformal Prediction Quantify? (importance: 12, dev: 12)
TopoCompile: Molecular Generation with Exact Global Count Constraints (importance: 10, dev: 10)
ProbePO: Sparse Probes, Redistributed Credit for Critic-Free LLM Reinforcement Learning (importance: 25, dev: 35)
Structured Latent Dynamics for Long-Horizon Embodied World Models (importance: 15, dev: 15)
From Score to Diagnosis: Search-Independent Post-Diagnosis for Symbolic Regression (importance: 12, dev: 15)
A Multi-Step Inertial ADMM Algorithm for Federated Learning (importance: 10, dev: 10)
Theory-Guided Instance-Dependent Early Stopping via Second-Order Loss Dynamics (importance: 15, dev: 25)
No-Regret Revenue Maximization in Sequential Market Making Under Inventory Constraints (importance: 8, dev: 5)
Representations Have Trajectories: Characterizing Temporal Representation Dynamics in Neural Networks (importance: 15, dev: 15)
DeliMF: Gated Cross-Attention Fusion of Structured Time Series and Clinical Text for Early Intensive Care Delirium Prediction (importance: 12, dev: 10)
AdamPM: Memory-Efficient LLM Optimization via Moment Pooling (importance: 25, dev: 45)
TCMScience: An Autonomous Scientist for Traditional Chinese Medicine (importance: 15, dev: 25)
How Much Do Activation Monitors Benefit from Original Model Computations? (importance: 20, dev: 25)
Uncovering Semantic Hierarchies in Text-Attributed Graphs via Variational EM-based LLM-GNN Synergy (importance: 12, dev: 15)
EdgeKernelBench: Unified LLM-Driven Kernel Generation for Edge Backends (importance: 25, dev: 55)
Contrastive Adversarial Training for Robust Graph Neural Networks under Label Poisoning (importance: 12, dev: 12)
Native Probability Outputs for Clinical Triage: An Evaluation of Calibration and Decision Accuracy (importance: 12, dev: 10)
Tensor Network Datamodels for Training Data Attribution (importance: 12, dev: 12)
fxBench: Evaluating and Understanding Formula Suggestions in Spreadsheets (importance: 20, dev: 35)
How Does Muon Work in Private Deep Learning? Scalable Differentially Private Training via Orthogonalization (importance: 15, dev: 20)
Gaussian Process Bandits with Paired Auxiliary Observations (importance: 10, dev: 10)
From Overparameterised Expressions to Parsimonious Models: Fisher-Geometric Pruning for Symbolic Regression (importance: 10, dev: 10)
Paired-Outcome Learning for Selective Spacecraft Attitude Capture under Torque Limits (importance: 8, dev: 15)
Global Convergence in Deep Networks via the Second Law of Thermodynamics (importance: 10, dev: 8)
Process-Aware Event Log Augmentation for Process Prediction under Labeled Data Scarcity (importance: 10, dev: 15)
First-Order LoRA Training with Phone NPUs (importance: 22, dev: 35)
Pair-Aware Optimization: Coordinating Exploration and Refinement for Jailbreaking LLM Agents (importance: 22, dev: 30)
Compose the Horizon: Parameter-Shared Blockwise Forecasting under a Sequential-Compute Trade-off (importance: 10, dev: 10)
Recurrent Temporal Restricted Boltzmann Machines as Unsupervised Encoders for Biomechanical Movement Sequences (importance: 8, dev: 8)
A Statistical Viewpoint on Adam, Shampoo and Muon (importance: 25, dev: 35)
An Impossibility Result for ETF Target Geometry, and an Equinorm Remedy for Imbalanced Analytic Continual Learning (importance: 10, dev: 10)
LLM-Based Evaluation Planning for Heterogeneous Optimization (importance: 12, dev: 15)
FlowMaker: From Tasks to Cost-Efficient Generative AI Workflow Deployments (importance: 20, dev: 40)
RAFT: Redundancy-Aware Input Filtering for Temporal Graph Neural Network Training (importance: 10, dev: 25)
Sample-Efficient Reinforcement Learning with General Utilities via Neural Dual Critics (importance: 10, dev: 8)
Distribution-Based Regularization for Reward Fine-Tuning Diffusion and Flow Models (importance: 12, dev: 12)
A Symmetry-Matching Approach to Blind-Spot Reduction in Sparse Autoencoders (importance: 15, dev: 20)
Arctic-SWE: Revisiting the Long Horizon Agent Training by Modeling-System Co-Design (importance: 35, dev: 70)
CARTA: Self-Supervised Reusable Transaction Embeddings via Heterogeneous Tokenization and Cross-Attention (importance: 10, dev: 15)
TrustAgent-RFF: Agentic Diagnostic Planning for Reliable RF Fingerprinting under Domain Shift (importance: 22, dev: 35)
MLFont: Machine Learning-Driven Font Subsetting and Range-Compact Unicode-Range Slicing for Low-Latency Web Typography (importance: 30, dev: 65)
T-MUON: Tensorizing Muon via t-SVD for Large Language Model Pre-training (importance: 35, dev: 45)
Robust Direct Preference Optimization under Preference Label-Flip Attacks (importance: 28, dev: 40)
ControlJEPA: Principled Trajectory Regularization via Lyapunov Tube Loss (importance: 25, dev: 35)
CasePlay: Self-Play Reinforcement Learning from Case Reports for Medical Reasoning (importance: 25, dev: 30)
Label-Space Mismatch Distorts Cross-Cohort Evaluation in Breast Cancer Molecular Subtyping (importance: 15, dev: 20)
PRECOCIAL: Pretraining with Code Curriculums via Iterative Search for Accelerated Learning (importance: 25, dev: 35)
Certified Sparse Attention via Perturbed Retrieval (importance: 35, dev: 60)
Block Recursive Transformers for Structured Parameter Sharing (importance: 28, dev: 40)
UtoMe: Observation-Uncertainty-Guided Token Merging for Weather Foundation Models (importance: 15, dev: 25)
UpLoop: Upcycling Large Language Models with Gated Loop (importance: 32, dev: 40)
Conjugate Compilation: Self-Improving Agentic Co-Compilation for Speculative Decoding (importance: 35, dev: 65)
Latent World Models Should Plan with Distances Along the Latent Manifold (importance: 22, dev: 30)
Multifidelity Monte Carlo Estimator for High-Dimensional Quantities of Interest (importance: 5, dev: 15)
Decentralized Online Exp-Concave Optimization (importance: 8, dev: 12)
LLM Agents as Suppliers of Physical System Models: Calibration and Acceptance in Modelica (importance: 30, dev: 50)
Synergy-Aware Residual Flow Matching for Single-Cell Perturbation Prediction (importance: 15, dev: 15)
Beyond the Teacher: Understanding and Preserving Distillation Overshoot in Agentic On-Policy Distillation (importance: 35, dev: 50)
PACT: NVFP4 Data-Flow LoRA Fine-Tuning on Consumer GPUs (importance: 35, dev: 65)
HCFM: Helmholtz Component Flow Matching for Time Series Anomaly Detection (importance: 20, dev: 30)
Policy Is Not Capability: Some Safety Benchmarks Measure Which Definition of "Unsafe" You Trained On (importance: 28, dev: 35)
The Optimal Sample Complexity of Pointwise Replicable Agnostic Learning (importance: 8, dev: 10)
Auto-MD-bench: Evaluating Autoresearch Agents for Real-World Medical Tasks (importance: 22, dev: 30)
CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow? (importance: 40, dev: 60)
Conformal Prediction-Enabled Sequence-level Policy Optimization for Reasoning Language Models (importance: 30, dev: 40)
Self-organizing reinforcement learning: a swarm approaches the explore-exploit optimum (importance: 10, dev: 15)
Polynomial Learning of Positive ReLU Networks under Gaussian Inputs (importance: 15, dev: 15)
Memory as Belief: Bayesian Evolution of Memory for Self-Evolving LLM Agents (importance: 35, dev: 45)
ClinSDT: LLM-Encoded Clinical Semantic Guidance for Sepsis Treatment via Decision Transformer (importance: 20, dev: 25)
Trackable Skill Library Lifecycle Management for Agentic Reinforcement Learning (importance: 35, dev: 50)
FHKAN: Fractional Hermite Kolmogorov-Arnold Networks for Data-Driven and Physics-Informed Learning (importance: 15, dev: 20)
Learning without Overwriting: Structural Knowledge Fusion for Continual Learning (importance: 18, dev: 20)
Indistinguishable Agents: Training Agent Privacy as Twin-World Indistinguishability (importance: 30, dev: 35)
M-JEPA: Predictive Self-Supervised Learning for Molecular Graphs with Scaffold-Shift Evaluation on Tox21 (importance: 15, dev: 20)
World Models, Trees and Foundation Models for Retail Forecasting: Evidence from Four Public Data Sets (importance: 20, dev: 25)
Sampling and Curvature in First-Exit Comparisons for Momentum SGD (importance: 6, dev: 15)
Shallow Probes Are Sufficient for Near-Oracle Personalization Depth Allocation in Federated Learning (importance: 15, dev: 25)
ViCoR: Estimating Visual Necessity via Counterfactual Residuals for Multimodal Medical Data Selection (importance: 18, dev: 25)
NeuroAct: Advancing EEG Analysis and Decoding via Dual-Loop Autonomous Research (importance: 25, dev: 35)
DROD: Dual-Ratio Ordered Dropout for Federated Continual Learning on Heterogeneous Devices (importance: 15, dev: 20)
Improving the Expressive Power of Hybrid Sequence Models via Multi-Hop Retrieval (importance: 25, dev: 35)
BatchParScale: Sparse Batch-Mixing for Sequence Models (importance: 25, dev: 40)
Emergent Sparsity: How Many Effective Variables Survive Coarse-Graining in AlphaEarth Embeddings? (importance: 10, dev: 15)
EasyResearch: Honest Regression, Forecasting and Signal Classification for Researchers Who Do Not Program (importance: 18, dev: 35)
EG-GRPO: Evidence-Guided Step-level Credit Assignment for Deep Search Agents (importance: 30, dev: 45)
Data-driven forecasting of tropical cloud forest phenology: A dynamical systems and machine learning hybrid approach (importance: 5, dev: 10)
A discrete-continuous-time stochastic framework for attention-deficit/hyperactivity disorder-related symptom modeling and pharmacological intervention via methylphenidate (importance: 5, dev: 5)
World Model Interpretability: From Decoding Representations to Understanding Latent Worlds (importance: 22, dev: 25)
Learning When to Abort in Transformer Planning: A Mechanistic Study (importance: 22, dev: 30)
Where Groups Split: A Probe Prior for Nonzero Within-Group Variance in RL Post-Training (importance: 25, dev: 40)
Follow the Mixed Groups: Ridge-Tracking Gaussian Sampling for Zero-Variance Web Rollouts (importance: 25, dev: 40)
A Distribution-Free Separation Between Learned and Fixed Features (importance: 10, dev: 10)
Predicted Class Correction: Conformal Thresholds for Classes with No Calibration Data (importance: 15, dev: 18)
CHSM8: A Chess Transformer that Follows Natural-Language Style Prompts (importance: 20, dev: 20)
CD-OPD: Contextual Digest for On-Policy Distillation (importance: 30, dev: 45)
Operator-Mismatch Contrastive Learning for World-Relative Trajectory Coherence (importance: 20, dev: 25)
Multi-site Learning for Forecasting Day-Ahead Wind-Power Production for Onshore Wind Farms in Great Britain with Deep Neural Networks (importance: 12, dev: 20)
Mining the Self-Rewriting System: A Vision for Data Mining on AI-Native Infrastructure (importance: 30, dev: 40)
Simulating Infant Senses with Age-Tuned Networks (importance: 10, dev: 10)
Structured Relevance Determination and Channel Additive Gaussian Processes for Image Data (importance: 10, dev: 12)
The Hidden Look-Ahead: Point-in-Time Leakage in Electricity Market Data (importance: 18, dev: 30)
Paloma: Phase-Conditioned Residual Modulation for Time Series Forecasting (importance: 18, dev: 22)
MOSAIC: Multi-Objective Search for Allocating Inference Cache Across LLM Decoder Layers (importance: 30, dev: 55)
Enhanced Masked Diffusion Unlearning (importance: 18, dev: 25)
Semi-supervised learning on singular data sets (importance: 6, dev: 10)
Near-Optimal Last-Iterate Convergence of SGD in the Smooth Interpolation Regime (importance: 8, dev: 12)
Consistent LLM Call Handling Across Java Frameworks: A Reference Implementation and Benchmark (importance: 40, dev: 85)
Probabilistic Limit Order Book Forecasting with Conditional Diffusion Models (importance: 12, dev: 15)
Steered Not Stirred: Effective Reward Tilting using Surrogate Reward Mixing (importance: 18, dev: 22)
KPI-Aware Multimodal Anomaly Detection for 5G and Open RAN Telemetry (importance: 12, dev: 25)
Dynamics-Aligned Representation Learning under Induced Partial Observability (importance: 15, dev: 20)
ATEST: Adaptive Resource-Aware Test-Time Scaling for LLM Inference (importance: 25, dev: 40)
FracTS: Hierarchical and Autoregressive Time Series Generation (importance: 12, dev: 15)
Reinforcement learning on the EPRI AI-ccelerating Unit Commitment benchmark: diagnosing over-commitment and guiding MILP with learned policies (importance: 15, dev: 20)
Beyond the record: evaluating probabilistic AI weather forecasts deep in the tail (importance: 18, dev: 15)
FTFM: Factorized Tabular Foundation Models (importance: 15, dev: 20)
Benchmark for Structured Retrieval-Augmented Generation in Perovskite Photovoltaics (importance: 12, dev: 25)
Finite Interaction Lens: Structured Precision Sharing for Vision-Language Models (importance: 18, dev: 35)
CVTA: Cross-Variable Temporal Attention for Multivariate Irregular Time Series Prediction (importance: 12, dev: 15)
Divergence based Ranked Preference Alignment for LLMs (importance: 18, dev: 20)
Distributionally-Robust Policy Learning with Observational Data (importance: 10, dev: 12)
Text Dominance in Paralinguistic Conflict: Tracing Suppression Through Audio LLM Circuits (importance: 22, dev: 25)
Adversarial Post-Training of Foundation Models for High-Frequency Trading (importance: 15, dev: 20)
PRISM: Prior Relational Information for Self-supervised Modeling to Enhance Solubility OOD Generalization (importance: 10, dev: 12)
Safety Offset Learning: When Benign Fine-Tuning Misaligns LLMs (importance: 30, dev: 35)
Curriculum Learning Teaches LLMs Long-Horizon Implicit Reasoning, Up To A Limit (importance: 22, dev: 25)
Weighted State Souping for Efficient Long-Context Retrieval (importance: 20, dev: 30)
Zephyrus-R1: Agentic Reinforcement Learning for Weather Reporting (importance: 15, dev: 20)
Conditional Average Treatment Effect Distillation for Hypothetical State-Intervention Targeting Under Budget Constraints (importance: 10, dev: 12)
OPDArena: Understanding Learning and Forgetting in On-Policy Distillation (importance: 25, dev: 35)
A "feature ODE" describing the learning behavior of shallow MLPs on simple functions (importance: 15, dev: 15)
Can Metadata Fix the Gauge? Calibration Turns Sparse Multi-Source Learning from Sparse PCA into Sparse Mean Recovery (importance: 8, dev: 10)
LOOP: Recovery-Aligned Iterative Layer Pruning via On-Policy Distillation (importance: 25, dev: 40)
MT-CC: Multi-Group Temperature Scaling for Asymmetric Calibration Behavior in Class-Incremental Learning (importance: 12, dev: 15)
Progress-Guided Preference Generation for Aligning Generative Control Policies (importance: 15, dev: 20)
Stochastic Tracking of Drifting Affine Fixed Points (importance: 6, dev: 10)
Predictive Bandit Curriculum Reinforcement Learning via Gradient-Based Sample Group Discovery (importance: 12, dev: 15)
Evidence Sufficiency and Model Comparisons in MexRegQA: An Item-Level Audit of Mexican-Spanish Regulatory QA (importance: 12, dev: 20)
Learning to Couple: Context-Adaptive Joint Denoising for Diffusion Language Models (importance: 22, dev: 25)
Anatomy of Off-Policy Policy Gradient: Importance Sampling, KL Regularization, and Baselines (importance: 45, dev: 65)
Neural Video Coding under Limited Delay: Theoretical Optimality and System Design (importance: 25, dev: 30)
How Much Does One Wrong Answer Cost? Error Propagation in Constraint Pruning with Noisy LLM Oracles (importance: 30, dev: 55)
Bayesian Model Discovery with Language Models: Closing the Loop between Hypotheses and Experiments (importance: 30, dev: 35)
Cognitively Decoupled Meta-Induction for Out-of-Distribution Generalization (importance: 30, dev: 50)
NADS: Navigator-Guided Data Selection for Mitigating Catastrophic Forgetting in Fine-Tuning (importance: 40, dev: 65)
Which Answers Agree? Evaluating Self-Consistency from Incomplete Logs (importance: 35, dev: 50)
Remember Less, Adapt Longer: Bounded-State In-Context Reinforcement Learning with Recurrent Summary Memory (importance: 30, dev: 50)
Computation, Not Memory: Chain-of-Thought and State Tracking in LLM Agents (importance: 40, dev: 60)
Detection Augmented Learning In Non-Stationary Bandit Environments (importance: 20, dev: 35)
Sector-Ring Linear Reservoirs: Sector-Wise Decay-Timescale Allocation With State Reuse Across Candidate Settings (importance: 15, dev: 30)
Communication-Efficient Federated Domain Generalization in Edge Networks via Dynamic Prototype Regulation (importance: 20, dev: 30)
Is it RL or Data? A Controlled Comparison of GRPO and Rejection-Sampling Fine-Tuning (importance: 40, dev: 60)
SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation (importance: 30, dev: 40)
ACE-Trader: Adversarial Co-Evolution for Generator-Critic Trading Systems (importance: 25, dev: 35)
Rubrics Beyond Evaluation: Exploiting Rubric Structure in Reinforcement Learning (importance: 35, dev: 60)
Statistical Learning Theory for Direct Preference Optimization (importance: 30, dev: 45)
Adaptive Decoding Methods for Token-Efficient Reinforcement Learning (importance: 35, dev: 60)
Residual-Guided Adaptive Tucker Rank Allocation for Parameter-Efficient MCI Conversion Prediction (importance: 15, dev: 35)
DeltaTrace: Efficient Signed Attribution for Reasoning Language Models (importance: 35, dev: 55)
Asynchronous On-Policy Self-Distillation (importance: 40, dev: 60)
Learn the Interface, Not the Implementation: Output Representation Dominates Training Regime in LLM-Based RTL Generation (importance: 35, dev: 65)
Out-of-Distribution Detection in Continual Learning (importance: 20, dev: 35)
An Analysis of Semantic and Spatio-Temporal Co-evolution for Edge Classification on Dynamic Text-Attributed Graphs (importance: 20, dev: 35)
Maestro: Agentic Formal Proof Development by Coordination (importance: 40, dev: 65)
UtilityBench: Measuring Language Model Utility When Errors Have Different Costs (importance: 30, dev: 35)
Keep the Lineages Alive: Diversity-Preserving SMC for Diffusion LLM Decoding (importance: 30, dev: 45)
The KV-Surgeon: Compensation-Aware Offline KV-Cache Compaction (importance: 35, dev: 65)
Correctness-Dependent Asymmetric Target Distributions for Self-Distillation Policy Optimization (importance: 35, dev: 55)
Graph-to-Table Interfaces Control Relational Generalisation (importance: 20, dev: 35)
PnP-CF: A Consensus Flow-Matching Solver for General Inverse Problems (importance: 15, dev: 20)
Scenario Generalization and Sector Transfer in Real Surface Irrigation Surrogate Modelling with Tabular Models (importance: 15, dev: 30)
ProtGlycanDock: Towards Accurate Protein-Glycan Docking with Tailored Dataset, Benchmark and Model (importance: 15, dev: 20)
Truncate or Not? A Scaling Study of Looped Transformers with Truncated Backpropagation (importance: 30, dev: 45)
FinHopper: A Progressive Multi-Hop Benchmark for Complex Financial Agents with Step-wise Evaluation (importance: 30, dev: 45)
Provable Adversarial Detection: Prime Quantization meets Gromov--Wasserstein (importance: 15, dev: 25)
Learning from a Bandit Learner (importance: 15, dev: 25)
Declarative Explanations for Graph Neural Networks: A Demonstration (importance: 15, dev: 35)
From Retrieval to Recognition:Knowledge-Enhanced Post-training of Time Series Foundation Models for Classification (importance: 15, dev: 35)
Variance-Aware Non-Stationary Bandits:All-Budget Regret Bounds and Adaptation Costs (importance: 10, dev: 25)
Optimal Multi-Kernel Combination for Online Multi-Label Classification (importance: 10, dev: 25)
Nash-MOPD: Optimizing Task Trade-offs in Multi-Teacher On-Policy Distillation (importance: 30, dev: 50)
Dynamical Batching: How Class-Cycling Training Shapes the Internal Structure of Neural Networks (importance: 15, dev: 30)
TaQFormer: A Treatment-as-Query Transformer with Data-Fusion for Multidimensional Treatment Modeling in Coupon Allocation (importance: 10, dev: 30)
OceanTensorCube and ST-MMT: A Multimodal Data and Transformer System for Seven-Day Forecasting of Pyropia Bleaching (importance: 10, dev: 15)
SEAM: Spectral Embedding Alignment and Merging for Low-Resource TTS (importance: 20, dev: 40)
Trust Your Targets: Is Target Propagation Block-Diagonal Gauss-Newton? (importance: 15, dev: 30)
Learning SOCP Solver Dynamics with Graph Neural Networks (importance: 15, dev: 30)
Toward Online Robust Zero-Sum Markov Games with Function Approximation (importance: 10, dev: 20)
CAST: Cast-Admissible Generative Design and Shape Optimization with Paired Depth Images (importance: 15, dev: 25)
Typed Decision Models under Prevalence Shift: Decision Training, Calibration, and Implicit Priors (importance: 20, dev: 40)
CIS-Transolver: A Cross-Layer Incremental-State Neural Operator for Fine-Scale Automotive Aerodynamic Flow Prediction (importance: 15, dev: 25)
Logit Averaging During Training Sets a New State of the Art for Non-Invasive Speech BCI (importance: 15, dev: 20)
An Empirical Study of Parameter-Efficient Fine-Tuning for Mixture-of-Experts LLMs (importance: 35, dev: 60)
CTLite: Lightweight Irregular Multivariate Time Series Forecasting with Reusable Continuous-Time Summaries (importance: 15, dev: 30)
A Simple and General Approach to top-$k$ Selection (importance: 10, dev: 20)
On Learning Expand-and-Sparsify Representations (importance: 10, dev: 25)
A Simple and Unified Method for Gradient-Variation Regret (importance: 10, dev: 20)
Wet-Biased Data Pipelines Mislead the Training and Evaluation of Generative Precipitation Nowcasting (importance: 15, dev: 25)
Strength Controls Change How Interventions Compare Across Sparse Autoencoder Codes (importance: 25, dev: 55)
A Fisher-Geometric View of Multi-Teacher On-Policy Distillation (importance: 30, dev: 50)
Concerns in Benchmarking Jailbreaks: An Empirical Study of Attack Success Rates (importance: 30, dev: 45)
Bombadil: A Foundation Model for Plant Electrophysiology (importance: 10, dev: 15)
ITU-DESC: a missingness-aware machine-learning dementia risk score in the UK Biobank cohort (importance: 10, dev: 15)
Degree Bounds for Separating Polynomial and Neural Invariants for Point Clouds (importance: 10, dev: 20)
Global Time-Aware Conditioning (GTC): A Mechanism for Time-Aware Pediatric Sleep Representation Learning (importance: 10, dev: 15)
SleepPairNet: Learning Multimodal Pediatric Sleep Representations with Random-Pair Cross-Attention (importance: 10, dev: 15)
VaDar: Implicit Goal Conditioning via Value Disaggregation (importance: 15, dev: 30)
The Effectiveness and Durability of Pretraining Data Filtering in Protein Language Models (importance: 20, dev: 20)
Nested Shared-Operator Recursion on ARC-AGI-1: An Empirical Study of HRM, TRM and URM (importance: 15, dev: 30)
PACT: Plan-Augmented Coordination via Trace-Graphs for Dynamic Multi-Agent Combinatorial Optimization (importance: 15, dev: 35)
Which Model for Which Role? Cost-Efficient Pareto Recommendation for LLM Agentic Workflows (importance: 40, dev: 60)
Intra-Rubric Reward Divergence: Mitigating Rubric Hacking through Latent Dimension Recovery and Balancing (importance: 20, dev: 40)
Parallel Language Models via Variational Decoding (importance: 35, dev: 45)
GeoSplit: Selective Edge--Server Map Acquisition for Cross-User Travel-Mode Recognition (importance: 10, dev: 30)
Doubly Robust Learning of Heterogeneous Survival Treatment Effects under Time-Varying Treatments (importance: 10, dev: 15)
Deep Pontryagin Learning for Continuous-Time Stochastic Control with Jumps (importance: 10, dev: 15)
Copying Under Repetition: From Prefix Matching to Position Hashing (importance: 25, dev: 45)
Cross-Fitting for Neural Posterior Estimation (importance: 10, dev: 20)
CAVA: Complementary-Label-Aware Data Valuation (importance: 15, dev: 30)
recAMPLIFY: Large-Scale Pretraining of Recursive Protein Language Models (importance: 15, dev: 25)
Trajectory Replay: Continuous-Time Mixture Transport for Lifelong Generative Models (importance: 15, dev: 30)
Simulation Priors for Rapid Adaptation in Model-Based RL for Autonomous Racing (importance: 10, dev: 25)
HarnessPro: Recursive Harness Self-Improvement with Pareto Optimization for Performance-Cost Trade-offs (importance: 35, dev: 60)
Topological Sheaves of Irreducibility: A Theoretical Framework for Multi-Agent Task Obstructions (importance: 10, dev: 20)
Training with ML-labeled Data via Stochastic First- and Zeroth-Order Optimization (importance: 10, dev: 25)
On Data Scaling in Tabular ML: More Data, Better Generalization? (importance: 20, dev: 35)
Two hidden layer suffice: the caterpillar net (importance: 10, dev: 20)
Learning to Refer: Inference-Time Collaboration in Decentralized Federated Learning (importance: 15, dev: 30)
Scaling Parameter-Efficient Fine-Tuning for Brain Encoding (importance: 10, dev: 25)
Next Best Action: Causal Constrained Offline RL for Business Action Targeting (importance: 15, dev: 30)
Kronecker Delta Rule for Long Context Linear Attention (importance: 20, dev: 40)
Text-to-Image Generative Modeling Without Language Pretraining (importance: 20, dev: 25)
Measuring the Influence of Modality-Specific Node Representations on Graphs (importance: 10, dev: 25)
Composable Parametric Demographic Injection for Values Survey Simulation (importance: 10, dev: 25)
DUET: Decoupled Upper- and Lower-Body Execution for Task-Agnostic Humanoid Loco-Manipulation (importance: 10, dev: 20)
Language-Forcing: Advancing Humanoid Loco-Manipulation via Semantic Alignment of Whole-Body World Action Models (importance: 10, dev: 20)
InJePA: Intention-First Latent Planning with Coupled JEPA World Models for Visual Navigation (importance: 10, dev: 20)
EAGLE: Entropy-Aware Geometric Learning Enhancer for Partially Polygon-Supervised Object Localization (importance: 10, dev: 20)
Rethinking weakly supervised localization on chest radiographs: position nulls, protocol dependence, and PACE (importance: 10, dev: 15)
Your Self-Supervised Projection Head Captures Object Co-Occurrence Statistics (importance: 25, dev: 15)
Vocabulary-Incomplete Unsupervised Vision-Language Model Adaptation (importance: 25, dev: 15)
VERDICT: Visual Extraction-Result Decoder with Integrated Confidence and Traceable Evidence (importance: 30, dev: 35)
Same Space or Not: Benchmarking Spatial Intelligence Through Cross-View Contradictions (importance: 25, dev: 15)
A Controlled LLM Workflow for Evidence-Grounded Diagnosis of Risk Score Differences (importance: 25, dev: 25)
Geo-vision: Native-Resolution Geometry and Sampled Topology for Vehicle-Part Damage Classification (importance: 12, dev: 10)
IMPULSE: Push First, Then Predict in V-JEPA 2 Feature Space (importance: 25, dev: 15)
Spatio-Temporal Detection of Photovoltaic Systems on Industrial Building Rooftops Using Satellite Imagery and Machine Learning (importance: 8, dev: 8)
Super-resolution as Spatial Downscaling: Coherence, Structure and Task Utility in Earth Observation (importance: 8, dev: 8)
FrameScout: Scouting Query-Relevant Frames for Long Video Understanding (importance: 25, dev: 20)
LLM Evolutionary Search Strategies Should be Compared by Tokens (importance: 30, dev: 35)
SPELUNKA: A Framework for Finding Unanticipated Associations in Vision-Language Models (importance: 20, dev: 15)
Weakly Supervised Discovery of Motion Styles for Compositional Robot Control (importance: 10, dev: 5)
Unified Multiparadigm Fusion Intelligence for Rare Disease Diagnosis (importance: 15, dev: 10)
When Does Evidence-Graph Structure Predict Answer Correctness? Label-Free Abstention for Structured Retrieval (importance: 20, dev: 25)
ToolRosetta: Surface-Form Invariance of Tool-Call Guards (importance: 35, dev: 60)
OrTeN: Neural Inverse Graphics for Joint Terrain and Line-scan Camera Trajectory Reconstruction (importance: 8, dev: 5)
Location, Location, Generation: Where You Ask Matters for Generative Engine Optimization (importance: 20, dev: 10)
Interpretability of AI weather models via intermediate decoding (importance: 12, dev: 8)
Do LMs Know What Variables Do? Interpreting Code Representations via Variable Roles (importance: 25, dev: 40)
Prefix-Tuning for Arbitrary Output Sequences on Pretrained Transformers (importance: 20, dev: 25)
Recording-Aware Evaluation of Multimodal OT Intrusion Detection (importance: 10, dev: 10)
Reducing double-dipping in single-cell topic models with three-way count splitting (importance: 8, dev: 8)
The Weight of Teamwork: Does Communication Structure Help LLM Scientists Search for Scientific Laws? (importance: 15, dev: 15)
Sequence-to-Sequence Neural Surrogate Modeling of Transient Responses in High-Speed Links with Variable Design Parameters (importance: 8, dev: 10)
lepidoptera: A Domain-Agnostic Engine for Causal Simulation with Mixed Variable Types, Vine Copula Dependence, and Individual-Level Counterfactual Propagation (importance: 20, dev: 50)
Candidate-dependent residual routing trades top-of-list accuracy for recall in sequential recommendation (importance: 15, dev: 15)
Certified neural approximations for parabolic obstacle problems (importance: 8, dev: 8)
Training-Free Prompt Selection for Efficient Reinforcement Learning With Partial Rollouts (importance: 25, dev: 20)
Censored Flow Matching for Synthetic Time-to-Event Generation (importance: 8, dev: 8)
Preserving Reasoning Diversity in Reinforcement Learning with Verifiable Rewards via Adaptive Replay (importance: 25, dev: 20)
Encoder-Free Multimodal LLMs Learn Implicit Encoders (importance: 30, dev: 15)
Extending Myerson's Optimal Auctions to Correlated Bidders via Neural Network Interpolation (importance: 8, dev: 5)
CoCoNet: Convex Conformal Network for Multi-Dimensional Uncertainty Regions (importance: 12, dev: 10)
Can This LLM Annotate My Data? Estimating Zero-Shot LLM Performance (importance: 25, dev: 30)
Cortex: A Fixed-Point Theory of Governed Coding Agents - Technical Report (importance: 35, dev: 60)
BanglaShield: A Script-Robust Safety Classifier and Benchmark for Bangla Native, Romanized, and Code-Mixed Prompts (importance: 15, dev: 15)
RATION: Towards Quantitative Risk-Based Access Control for AI Agents (importance: 30, dev: 40)
What Matters for NVFP4 Training? A Comparative Study of Low-Precision Pretraining Recipes (importance: 30, dev: 25)
Blackbox Performance Analysis of LLM-based Network Control (importance: 10, dev: 15)
Can Agents Keep Secrets They Were Never Told? Benchmarking Latent Privacy Risks Across Sessions with MemJoin (importance: 25, dev: 30)
Learning from the Right Regions: Class-Conditioned Explanation Guidance for Medical Imaging (importance: 8, dev: 8)
A decomposition-integration framework coupling singular spectrum analysis with a volatility-adaptive LSTM for stock price forecasting (importance: 5, dev: 5)
Learning volatility-adaptive forgetting via GARCH-modulated LSTM for stock price forecasting (importance: 5, dev: 5)
MSCR: Jointly Balancing Utilization Across Modalities and Discovering Synergistic Information (importance: 15, dev: 10)
Prototype-Guided Optimization of Quantum Data Embeddings with Linear Cost (importance: 10, dev: 10)
Margin-Based Generalization for Quantum Machine Learning with Classical Data Encoding (importance: 8, dev: 5)
Private Evaluation of High-Dimensional Functions (importance: 10, dev: 15)
Write Once, Check Many: Reusable Ledgers for Post-Hoc Policy Compliance Auditing of Agent Trajectories (importance: 30, dev: 45)
Denoising Implicit Variational Inference (importance: 12, dev: 10)
Reconstruction of Unobserved Core Electron Density in Tokamak Plasmas with a Neural Operator Transport Surrogate and Uncertainty Quantification (importance: 8, dev: 8)
Rethinking EEG Spatial Super-Resolution as Dynamic State Decomposition and Filtering (importance: 8, dev: 8)
Contrastive Hypergraph Source-free Domain Adaptive Object Detection in Adverse Weather Conditions--Extended Version (importance: 12, dev: 8)
Write Once, Query Many: Evolving Compact Reusable Memory for Long Videos (importance: 20, dev: 20)
Accelerating Finite-Sum Bilevel Optimization with Fully First-Order Oracles (importance: 8, dev: 5)
From Stories to Features: Scaling Concept Vectors for AI Alignment & Interpretability (importance: 20, dev: 15)
ASR Personalization in Pathological Voice: Accuracy, Efficiency, and Duration Effects (importance: 8, dev: 15)
Conditional Weight of Evidence for Gradient-Boosted Credit Models: Estimation, Additivity and Its Relation to Adverse Action Reasons (importance: 10, dev: 15)
Training with Ciphers to Enhance In-Context Terminology Learning For Translation from Rare Languages (importance: 15, dev: 15)
Severity-Channel Conditioning: An Injection-Level Study for Consumer-Device Deblurring (importance: 10, dev: 8)
Modeling Uncertainty Propagation in 3D Morphable Models (importance: 8, dev: 5)
Good Judgment, Bad Advice: Decision-Model Controllers Raise Agent Cost Without Improving Success (importance: 35, dev: 60)
Lightweight Machine Learning for Classifying Disaster-Related Social Media Posts (importance: 6, dev: 5)
Can AI agents make law? Evaluating AI legal relevance and creativity (importance: 12, dev: 5)
What RL Brings to Reasoning: A Few Decisions the Base Model Would Not Make (importance: 25, dev: 20)
Knowledge Diversion: A Novel Paradigm for Pre-training Decomposable Image Generation Models (importance: 10, dev: 8)
MM-LeJEPA: Correlated Gaussian Regularization for Multimodal Joint-Embedding Predictive Architectures (importance: 12, dev: 8)
Geometry-Aware Interpolation for Inverse Problems (importance: 8, dev: 5)
Spherical Physics-informed Neural Operator with Multi-scale Coupling for Meteorological Downscaling (importance: 8, dev: 8)
When Do Repetition Loops Lock? Truncation Sampling, Bounded Penalties, and a Blind Spot in DRY (importance: 25, dev: 30)
Evidence Field: Adaptive Covering Programs for Medical Imaging (importance: 8, dev: 8)
Martingale Posterior Cox Processes (importance: 6, dev: 5)
Who Gets Unmasked? Sampler-Induced Bias in Diffusion Language Models (importance: 20, dev: 15)
Is User Calibration Still Necessary for Continuous Myoelectric Control? A Closed-Loop Comparison of Calibration-Free Controllers (importance: 6, dev: 8)
Tilted Mondrian: Triage and Mitigation of Fine-grained Disparities in Conformal Prediction (importance: 8, dev: 5)
Efficiency Measured in English: Tokeniser Coverage Bounds Small Language Models on Indic Scripts (importance: 30, dev: 30)
Bellman Language Models: Solving the Recurring Decisions of LLM Agents (importance: 25, dev: 35)
The Missing Compliance Class: LLM Graders That Collapse Jailbreak Success, and a Free Check (importance: 20, dev: 15)
Measuring Data Diversity with Embeddings: A Taxonomy and Benchmark for Diversity Measures (importance: 20, dev: 25)
Query-Dependent Reading for Quantized Inner-Product Estimation: A Minimax Separation (importance: 8, dev: 15)
Scalable and Adaptive Distribution Regression via Random Features (importance: 6, dev: 5)
LastWords: Boring Problems Revisited (importance: 10, dev: 10)
ATLAS: Expert-Guided Sustainable Satellite Backhaul for Off-Grid Wildlife Monitoring (importance: 5, dev: 5)
DF-PALM: Dynamic Fusion Prompt Learning for Audio--Language Models (importance: 10, dev: 8)
Means-Ends Analysis Helps LLMs Plan in Continuous-State Domains (importance: 20, dev: 20)
LowNet: Looped Wavelet Neural Networks with 24 Learnable Parameters for Solving Poisson Equations (importance: 6, dev: 8)
Over-Defense in Coding Agents Begins at Silent Forks (importance: 35, dev: 55)
Revealing what benchmark scores hide: Jointly measuring the breadth and reliability of LLM capabilities using accessibility profiles (importance: 20, dev: 20)
SoK: PIAtlas: Decomposing the Unbounded Prompt Injection Attack Space (importance: 30, dev: 35)
No Escape from the Quadratic Barrier: Adaptive Attacks on Linear $\ell_2$ Sketches (importance: 6, dev: 15)
Integrable building blocks for symplectic neural networks: learned-type layers, fiber-preserving layers, and what a single layer can do (importance: 6, dev: 5)
Normalizing Flow for Neural-Enhanced Track-Before-Detect (importance: 5, dev: 5)
Non-Acceleration of the Heavy-Ball Method in 1D (importance: 5, dev: 5)
Compile Once, Run Cheap: Frozen, Portable Skills for Embodied Agents, and What Recursive Self-Improvement Adds (importance: 25, dev: 25)
Decision-Target Reversal under Informative Label Missingness (importance: 5, dev: 5)
Many Repairs, One Direction: Steering Code Models Toward Safer Cryptographic API Use (importance: 25, dev: 35)
TIRE: Top-Down Formula Instantiation with Reliable Execution for Physics Problem Synthesis (importance: 15, dev: 15)
HullProbe: Measuring Interventional Part Reliance in Maritime Vessel Classifiers (importance: 5, dev: 5)
Spectral Adaptive Graph Transformer for Efficient and Scalable Traffic Forecasting (importance: 5, dev: 5)
An Information-Geometric Framework for OCS-Reconfigurable Interconnect Topology in Large-Scale AI Clusters (importance: 15, dev: 35)
Dialogue Generation for Conversational Recommendation Using User Search Interactions and Knowledge Graphs (importance: 15, dev: 40)
ChronoChart: Setup-Agnostic Channel Charting for Label-Free Indoor Localization (importance: 6, dev: 30)
RAIL-5G: A Challenging Real-World Dataset and Benchmark for AI-Based 5G Indoor Localization (importance: 6, dev: 30)
Calibration-Free Early Stopping for Chain-of-Thought Reasoning via E-processes (importance: 15, dev: 45)
LiftDiff: Guided Discrete Diffusion for Quantum Code Construction (importance: 8, dev: 25)
Assessing the Empirical Evidence for Loss of Control from Agentic General-Purpose AI (importance: 40, dev: 40)
Estimating Causal Bounds from Observational Data With Neural Generative Models (importance: 8, dev: 40)
Persistent Representational Misalignment Induces Vision-Language Model Failures (importance: 15, dev: 40)
Federated Learning for Malware Image Classification under Data Heterogeneity (importance: 8, dev: 35)
Bridging Concepts: Multi-Agent Reasoning over Scientific Concept Graphs for High-Impact Research Prediction (importance: 8, dev: 35)
The Explainability Illusion: Must ViTs Exhibit Patch-Decoding Interpretability? (importance: 8, dev: 35)
Where and Why Search-Augmented LLMs Fail When Reasoning over Rapidly Evolving Knowledge (importance: 25, dev: 45)
Formally Verified Autoresearch for Analyzing the Expressivity of Multi-Head Attention (importance: 15, dev: 45)
GreenNIDS: A Benchmark Workflow and Analysis Dataset for NetFlow Intrusion Detection Experiments (importance: 8, dev: 40)
Whistle and Wrong Lyrics: Synthetic Queries Are Not Enough for Real-World Query-by-Melody (importance: 8, dev: 35)
TILDE: Teacher-in-the-Loop Distillation from Embodied Foundation Models (importance: 8, dev: 40)
Outcome-Anchored Credit Reallocation for Agentic Knowledge Graph Question Answering (importance: 12, dev: 40)
RARE: Reinforcement Learning with Adaptive Rubric Evolution for Open-Ended Generation (importance: 15, dev: 40)
BiDex-100: A Comprehensive Benchmark for Bimanual Dexterous Manipulation Across 100+ Simulation Tasks (importance: 5, dev: 35)
Reliability-Preserving Compression for Deployable Medical Image Segmentation (importance: 8, dev: 30)
On Reasoning Correctness Directions (importance: 15, dev: 45)
From Seeing to Reporting: A Theory of Reliable Visual Set Generation in Vision-Language Models (importance: 20, dev: 40)
Breaking the 1/3 Barrier for Approximate Nash Equilibria in Symmetric Bimatrix Games (importance: 3, dev: 25)
From Clinical Risk Prediction to Operational Decisions: An Uncertainty-Aware Framework for ICU Staffing (importance: 6, dev: 30)
Narrow Decision Models: A System-1 Architecture for Calibrated, Typed Decisions (importance: 25, dev: 40)
Concurrent two-scale topology optimization of printable bone scaffolds via rectified-flow generation and exact latent-space refinement (importance: 3, dev: 20)
Cognitive Map in Large Language Models (importance: 20, dev: 40)
CheXFuse: Guided Differential Joint Attention Fusion for Robust Multimodal Chest X-Ray Classification in Data-Scarce Medical Settings (importance: 8, dev: 35)
Clean Forcing: Drift-Resistant Autoregressive Video Diffusion with a Frozen Base (importance: 12, dev: 35)
Uncertainty-Gated SHAP-Dirichlet Refinement for MRI-Based Depression Classification (importance: 5, dev: 30)
Sampling of Signals on Multigraphs: From Removable and Uniqueness sets to Spectral Proxies (importance: 2, dev: 20)
Spherical Fixed-Point Reasoners: Adaptive Loop Halting on the Hypersphere (importance: 10, dev: 40)
Consistency Training Suppresses Cue Following Without Teaching Cue Selectivity (importance: 20, dev: 50)
Knowledge Manifold Optimization for Scientific Discovery (importance: 10, dev: 30)
Arbitrage-Free Implied Volatility Surface Reconstruction via Additive Martingales (importance: 3, dev: 20)
Corrected Block Sparse Spectral Optimization: Exact Models, Finite Identification, and Local Dynamics (importance: 3, dev: 20)
A Calibrated Instrument for Detecting Disposition Drift in Language Models (importance: 35, dev: 45)
Fundamental Limits of Information-Theoretic Secure Aggregation with User Dropouts in Federated Learning (importance: 5, dev: 30)
SAR: Learning Global Item-Space Matching and Geometry-Aware Placement Ranking for Buffered Online 3D Bin Packing (importance: 8, dev: 35)
Fair Regression under Approximate Demographic Parity (importance: 5, dev: 30)
Diffusion Language Models Change Speed Across Tasks and Languages (importance: 20, dev: 45)
Measurement Validity and Representation Dynamics of Confidence Signals in Large Language Models (importance: 15, dev: 40)
Are Self-Evolving Agents Continual Learners? (importance: 35, dev: 55)
CODA: Cohort- and Drift-aware Foundation Model for Multimodal Clinical Reasoning (importance: 10, dev: 35)
Learning to Reason Efficiently with Interventional Step Credit (importance: 25, dev: 45)
Motor compound fault diagnosis using an attention-based cross-talk architecture (importance: 3, dev: 25)
EvoSkill v2: Fully Agentic Skill Evolution from Benchmark Adaptation to a Stateful Evolver (importance: 40, dev: 75)
MITRA-lexicon: Lexica of attested translations from Sanskrit into English, Tibetan and Chinese (importance: 2, dev: 15)
OrderGPT: Generation Order as a Learnable Modeling Choice (importance: 15, dev: 40)
Productive State-Space Exploration for Multi-turn RL in LLM Agents (importance: 20, dev: 50)
NIOCIS: Typed-Decision Next-Hop Routing with Node-Conditioned Option Spaces for Tool-Use Agents (importance: 25, dev: 60)
Federated Community Detection in Bipartite Networks (importance: 3, dev: 25)
Internalizing Chain-of-thought into Looped Transformers (importance: 15, dev: 40)
TempoMoE: Identifying and Steering Fast-Slow Thinking Experts for Efficient Multimodal Reasoning (importance: 15, dev: 40)
SPLASH: Contrastive Audit Pressure for Detecting and Localizing Security Hallucinations in Generated Code (importance: 30, dev: 70)
Towards instance-dependent regret optimality in Episodic MDPs with Posterior Sampling (importance: 3, dev: 25)
LeakBench: Auditing Reward Hacking through Prediction-Time Leakage in Clinical Modeling Agents (importance: 25, dev: 50)
Multilingual Voicemail and Answering Machine Detection via Pretrained Audio Embeddings and Gradient Boosting Machines (importance: 5, dev: 30)
Reinforcement Learning Controllers for Fusion Scenario Development (importance: 5, dev: 25)
FLMNet: Self-Ensemble Distillation from Masks to Seeds for Single-Tower CTR Prediction (importance: 5, dev: 35)
Physics-Driven Framework for QoS-Constrained Satellite System with Non-Linear PA Distortion (importance: 2, dev: 25)
RECAP: Reciprocal Sparse Attention and Hierarchical Caching for Efficient Video Diffusion (importance: 10, dev: 40)
A Pitfall in the Evaluation of Gene Perturbation Models (importance: 3, dev: 20)
TriScore: Post-Hoc Out-of-Distribution Detection with Energy, Boundary Probes, and Transform Consistency (importance: 8, dev: 35)
The Control Layer Behind Rule-Following Agents Trading Across Market Conditions (importance: 25, dev: 50)
PhyPro-FM: A Physics and Prompt-Guided Graph Foundation Model for Contingency Analysis (importance: 3, dev: 25)
Selecting the Best Few with Human Evaluators: Adaptively Combining Ratings and Pairwise Comparisons (importance: 10, dev: 35)
ForagerSim: Evaluating Embodied Map-Based Reasoning in Large-Scale Dynamic Worlds (importance: 10, dev: 35)
LaMER: Language-Conditioned Multi-Expert Routing for Weakly Supervised Referring Expression Comprehension (importance: 8, dev: 35)
Co-Evolving Policy and Reward for Code Generation: A Mechanistic Study of Single-Adapter Drift (importance: 20, dev: 60)
Local data augmentation for imbalanced binary classification (importance: 3, dev: 25)
R2-GP: Variance-Calibrated Priors for Additive Gaussian Processes (importance: 2, dev: 25)
Regime-Aware Interpretable Ranking for Sector Rotation (importance: 3, dev: 25)
Residual-Based Adaptive Refinement in Hybrid Variational-Collocation PINNs for Singular Elliptic PDEs with Mixed Dirichlet-Neumann Boundary Conditions (importance: 3, dev: 25)
VeriCUA: Learning Proactive Verification for Reliable GUI Agent Rewards (importance: 20, dev: 55)
Learn the Task, Not the Schedule: Reducing Spurious Cross-Arm Coupling in Bimanual Imitation Learning (importance: 8, dev: 40)
RouterWorld: Scaling Conversational AI with Latency-Aware LLM Routing (importance: 25, dev: 55)
A Semantic Knowledge Graph Construction Pipeline for Vehicle Intention Prediction in nuScenes (importance: 5, dev: 35)
Training Natural Schemers: RL Can Amplify Oversight-Conditional Behavior (importance: 45, dev: 55)
SafeWall: A Risk-Aware Job Walltime Recommendation for HPC Scheduling (importance: 5, dev: 30)
WhatIf-Bench: Benchmarking Intervention-Aware Time Series Forecasting (importance: 10, dev: 40)
A Probabilistic, Rotation-Preserving Kernel Operator Surrogate for Continuum-Robot Design Spaces (importance: 2, dev: 25)
When the harness decides success: state that cannot be re-observed (importance: 45, dev: 80)
The Development of Sycophancy in Language Models: Tracing Vulnerabilities from Pretraining to Post-Training (importance: 30, dev: 40)
Injection of Space Radiation Errors on Low-power NVM using Generative Deep Learning (importance: 3, dev: 20)
Polar Switch: Fast Muon Orthogonalization with Polynomial Circuits (importance: 20, dev: 60)
Gradient descent reaches arbitrary accuracy in a fixed number of large steps on multiclass logistic regression (importance: 8, dev: 35)
Cooperative Multi-Agent Reinforcement Learning for Market Making on Prediction-Market Order Books (importance: 20, dev: 40)
Teacher Supervision, Fixed Representations and Calibration: Exact Risk Calculations for Candidate-Scoring Decision Models (importance: 25, dev: 45)
MemT-1.0: Towards Memory-Native Evolution (importance: 30, dev: 40)
Agent World Model in the Wild: Computer-Use Agent Environment Synthesis (importance: 25, dev: 50)
ProteinGeoBench: Frontier Structural-Biology Agents Are Fluent at Lookup but Lost in Geometry (importance: 20, dev: 35)
Supervised Learning with Quantum Hamiltonian Embeddings (importance: 3, dev: 25)
Newton-Based Methods for Bilevel Optimization Beyond Lower-Level Strong Convexity (importance: 5, dev: 30)
CoopUQ: Robust Multi-Robot Social Navigation via Uncertainty-Aware Cooperative Forecasting (importance: 5, dev: 30)
OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory (importance: 25, dev: 60)
A probability calibrator for machine-learning classifiers (importance: 8, dev: 30)
Chaos persists in large-scale multi-agent learning despite adaptive learning rates (importance: 10, dev: 30)
HyPE: Attention with Hyperbolic Biases for Relative Positional Encoding (importance: 15, dev: 40)
Human Responses to Algorithmic Correction: Evidence from Centre Court (importance: 3, dev: 20)
Provable Bounds on the Hessian of Neural Networks: Derivative-Preserving Reachability Analysis (importance: 12, dev: 25)
Gradient Boosting Reinforcement Learning (importance: 15, dev: 25)
Heterogeneous Sheaf Neural Networks (importance: 12, dev: 20)
Risk-averse learning with delayed feedback (importance: 10, dev: 15)
Selective Data Expansion for Model Performance (importance: 12, dev: 25)
Enhancing Offline Model-Based RL via Active Model Selection: A Bayesian Optimization Perspective (importance: 12, dev: 20)
Federated Learning: From Theory to Practice (importance: 15, dev: 35)
Recurrent Self-Attention Dynamics: An Energy-Agnostic Perspective from Jacobians (importance: 12, dev: 20)
Solving Inverse Problems via Diffusion-Based Priors: An Approximation-Free Ensemble Sampling Approach (importance: 12, dev: 15)
UNO: Unlearning via Orthogonalization in Generative models (importance: 14, dev: 20)
Dynamic gain neuromodulation attenuates stability gaps during continual joint training (importance: 12, dev: 18)
Optimizing Prompt Sequences using Monte Carlo Tree Search for LLM-Based Optimization (importance: 18, dev: 30)
Universal Learning of Nonlinear Dynamics (importance: 10, dev: 18)
EEGDM: Label-Efficient EEG Representation Learning with Generative Diffusion Model (importance: 10, dev: 12)
Sleep Stage Classification across Heterogeneous EEG/EOG Channels with Multi-Channel Differential Transformer (importance: 8, dev: 10)
Open-sci-ref-0.02: open reference baselines for reproducible language model and dataset comparison (importance: 22, dev: 35)
Optimal hypersurface decision trees (importance: 10, dev: 20)
SPRINT: Stochastic Performative Prediction With Variance Reduction (importance: 10, dev: 15)
Interpretable time series analysis with Gumbel dynamics (importance: 10, dev: 15)
OT-MeanFlow3D: Bridging Optimal Transport and Meanflow for Efficient 3D Point Cloud Generation (importance: 12, dev: 15)
Robust and Efficient Finetuning of Vision Foundation Models via Implicit Ensembling (importance: 15, dev: 25)
Time-Correlated Video Bridge Matching (importance: 10, dev: 15)
The Ends Justify the Thoughts: RL-Induced Motivated Reasoning Fools CoT Monitors (importance: 30, dev: 30)
A Unified Framework for Zero-Shot Reinforcement Learning (importance: 12, dev: 18)
Adaptive Data Selection for Text Classification: Hierarchical Value with a Local Utility Budget (importance: 12, dev: 25)
Manifold Approximation leads to Robust Kernel Alignment (importance: 10, dev: 15)
Empirical Bayesian Multi-Bandit Learning (importance: 8, dev: 12)
AutoTailor: Automatic and Efficient Adaptive Model Deployment for Diverse Edge Devices (importance: 15, dev: 40)
CLAPS: Aleatoric-Epistemic Scaling via Last-Layer Laplace for Conformal Regression (importance: 8, dev: 12)
D-STEER: Two-Way Control of DPO-Induced Safety Behavior in Activation Space (importance: 15, dev: 20)
Causal inference and model explainability analysis of a real world sales dataset (importance: 8, dev: 20)
From Perception to Punchline: Empowering VLM with the Art of In-the-wild Meme (importance: 8, dev: 10)
IRPM: Intergroup Relative Preference Modeling for Pointwise Generative Reward Models (importance: 15, dev: 25)
Standardization of Post-Publication Code Verification is Possible with the Support of the Community (importance: 10, dev: 25)
Patch-Level Tokenization with CNN Encoders and Attention for Improved Transformer Time-Series Forecasting (importance: 8, dev: 12)
TimeART: Towards Agentic Time Series Reasoning via Tool-Augmentation (importance: 15, dev: 25)
ELSA: Efficient LLM-Centric Split Aggregation for Privacy-Aware Hierarchical Federated Learning over the Network Edge (importance: 10, dev: 20)
Variance Reduction for Stochastic Switching in Functionally Constrained Federated Optimization (importance: 8, dev: 12)
Sparse Subspace-to-Expert Sharing for Task-Agnostic Continual Learning (importance: 12, dev: 20)
Machine Unlearning in Low-Dimensional Feature Subspace (importance: 10, dev: 15)
When Does Cheap Physics Help? Gated Multi-Fidelity Physics-Informed Networks with Provable Sample Savings and Conformal Guarantees (importance: 8, dev: 15)
MOSAIC: More LLMs, Less Cost through Collaborative Reasoning for Compiler Optimizations (importance: 18, dev: 40)
Finite-Sample Wasserstein Error Bounds and Concentration Inequalities for Nonlinear Stochastic Approximation (importance: 8, dev: 12)
Bayesian Preference Learning for Test-Time Steerable Reward Models (importance: 14, dev: 25)
SLA2: Sparse-Linear Attention with Learnable Routing and Quantization-aware Training (importance: 14, dev: 30)
Pushing the Frontier of Black-Box LVLM Attacks via Fine-Grained Detail Targeting (importance: 12, dev: 20)
$\kappa$-Explorer: Minimax Active Model Estimation in MDPs (importance: 8, dev: 12)
Brain-OF: An Omnifunctional Foundation Model for fMRI, EEG and MEG (importance: 12, dev: 10)
Rigidity-Aware Geometric Pretraining for Protein Design and Conformational Ensembles (importance: 10, dev: 15)
PTOPOFL: Topology-Guided Personalised Federated Learning with Persistent-Homology Client Descriptors (importance: 8, dev: 12)
Quantifying Memorization and Privacy Risks in Genomic Language Models (importance: 12, dev: 15)
KernelClaw: A Multi-Agent Framework for GPU Kernel Optimization with Dual-Memory Mechanisms (importance: 22, dev: 45)
Brittlebench: Quantifying LLM robustness via prompt sensitivity (importance: 14, dev: 30)
Attention Sinks Induce Gradient Sinks: Massive Activations as Gradient Regulators in Transformers (importance: 12, dev: 25)
Quotient Geometry and Persistence-Stable Metrics for Swarm Configurations (importance: 6, dev: 10)
RE-SAC: Disentangling aleatoric and epistemic risks in bus fleet control: A stable and robust ensemble DRL approach (importance: 8, dev: 12)
On the Role of Batch Size in Stochastic Conditional Gradient Methods (importance: 8, dev: 12)
Conditionally Identifiable Latent Representation for Multivariate Time Series with Structural Dynamics (importance: 8, dev: 12)
Diet Your LLM: Dimension-wise Global Pruning of LLMs via Merging Task-specific Importance Score (importance: 15, dev: 35)
See it to Place it: Evolving Macro Placements with Vision-Language Models (importance: 12, dev: 25)
Derived Fields Increase the Recoverable Information in Budgeted Neural PDE Surrogate (importance: 8, dev: 15)
LLM as a Tool, Not an Agent: Code-Mined Tree Transformations for Neural Architecture Search (importance: 14, dev: 30)
Democratizing Tool Learning with Environments Fully Simulated by a Free 8B Language Model (importance: 18, dev: 35)
Train Separately, Merge Together: Modular Post-Training with Mixture-of-Experts (importance: 16, dev: 25)
AutoREC: A reinforcement learning platform for equivalent circuit model generation (importance: 6, dev: 25)
Forager: a lightweight testbed for continual learning with partial observability in RL (importance: 8, dev: 20)
Multilingual Safety Alignment via Self-Distillation (importance: 12, dev: 15)
Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation (importance: 12, dev: 15)
Requests of a Feather Must Flock Together: Batch Size vs. Prefix Homogeneity in LLM Inference (importance: 22, dev: 60)
Bandit Learning in General Open Multi-agent Systems (importance: 8, dev: 12)
ProtSent: Protein Sentence Transformers (importance: 8, dev: 12)
PAAC: Privacy-Aware Agentic Device-Cloud Collaboration (importance: 18, dev: 35)
Continuous Latent Contexts Enable Efficient Online Learning in Transformers (importance: 12, dev: 20)
Nearly Optimal Algorithms for Adversarial Kernelized Bandits (importance: 6, dev: 10)
Gradient-Free Noise Optimization for Reward Alignment in Generative Models (importance: 10, dev: 15)
Approximation Theory of Laplacian-Based Neural Operators for Reaction-Diffusion System (importance: 6, dev: 10)
Let the Batch Decide, On- or Off-Policy: Effective Sample Size for RL Post-Training (importance: 14, dev: 25)
Agentic Reverse Engineering: From I/O to Code with Discovery Agent (importance: 22, dev: 45)
Differentiable Mixture-of-Agents Incentivizes Swarm Intelligence of Large Language Models (importance: 14, dev: 25)
BAPR: Bayesian amnesic piecewise-robust reinforcement learning for non-stationary continuous control (importance: 8, dev: 15)
SignMuon: Communication-Efficient Distributed Muon Optimization (importance: 14, dev: 35)
Radial-Angular Geometry for Reliable Update Diagnosis in Noisy-Label Learning (importance: 6, dev: 12)
A Measure-Theoretic Analysis of Reasoning in Transformers: Structural Generalization and Approximation Limits (importance: 8, dev: 12)
Certifying Whether to Adapt: Baseline-Relative PAC-Bayes-MMD Guarantees for Test-Time Adaptation (importance: 6, dev: 10)
Convex Compositional Reasoning Models (importance: 8, dev: 12)
DAMEL: Dual-Axis Multi-Expert Learning for Class-Imbalanced Learning (importance: 6, dev: 10)
Local Conflict, Targeted Repair: A Perturbation Analysis of Multi-Domain RL (importance: 12, dev: 20)
Smart Picks in the Dark: Towards Efficient RLVR for Reasoning via Tracing Metacognitive Pivots (importance: 10, dev: 18)
Balancing Image Compression and Generation with Bootstrapped Tokenization (importance: 8, dev: 12)
Optimal Post-Training Quantization Scales and Where to Find Them (importance: 14, dev: 40)
Non-Gaussianity Reveals Interpretable Directions in Language Models (importance: 10, dev: 15)
The Signs Were Always There: Training-Free Detection and Control in Raw Transformer Dimensions (importance: 10, dev: 15)
A Decision-Theoretic View of Test-Time Training: When, How Far, and Which Directions to Adapt (importance: 8, dev: 12)
Loop-$\mu$P: Residual Scaling and Hyperparameter Transfer in Looped Transformers (importance: 10, dev: 20)
Encoder-Decoder Manifold Alignment for Idempotent Generation (importance: 6, dev: 10)
Superhuman AI for Generals.io Using Self-Play Reinforcement Learning (importance: 10, dev: 15)
Compute Allocation in Bilinear Contrastive Representation Learning via Scaling Laws (importance: 8, dev: 12)
From Failure Taxonomy to Intervention: A Diagnostic Methodology for Industry-Scale AVLM in Video and Live-Streaming Platform Moderation (importance: 10, dev: 15)
Systematic Evaluation of Learning Rate Scheduling Strategies Across Heterogeneous Architectures (importance: 8, dev: 15)
Microstructure-Conditioned Surrogate Models for Graded Multiscale Optimization of Mycelium Composites (importance: 6, dev: 8)
Task-Oriented Sensing and Covert Transmissions for Collaborative Multi-AUV Systems (importance: 15, dev: 10)
SOS-LoRA: Static Orthogonal-Subspace Low-Rank Adaptation with Fixed Multi-Scale Scaling (importance: 30, dev: 55)
Arithmetic and its remedies select the evidence on learned geometry (importance: 25, dev: 40)
Circuit Claims Depend on What Is Extracted and How It Is Compared (importance: 25, dev: 45)
ISO: An RLVR-Native Optimization Stack (importance: 30, dev: 45)
Hierarchical Multilevel Monte Carlo for Order-Optimal Neural Actor-Critic in Average-Reward CMDPs (importance: 20, dev: 30)
Capacity Confounds and Coverage Guarantees in Adaptive Sub-model Federated Learning (importance: 20, dev: 35)
Test-Time Augmentation for LLMs: Where to Inject Diversity at Inference Time (importance: 35, dev: 50)
REATS: LLM Reasoning-based Ensemble Learning for Adaptive Time Series Forecasting (importance: 20, dev: 35)
Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-Based RL (importance: 35, dev: 45)
Dynamic Regime-Aware Conformal Calibration for Reliable Economic Forecast Intervals under Multiple Distribution Shifts (importance: 15, dev: 25)
Selection, Recombination, or a Fresh Solve? A Candidate-Free Control for Single-Pass Test-Time Aggregation (importance: 30, dev: 45)
When Clean Data Hurts: Learning with Monotone Corruptions Beyond Binary Classification (importance: 15, dev: 20)
Maximum-distance nonnegative matrix factorization for unmixing highly mixed grain-size distribution data (importance: 10, dev: 15)
Candidate-Label Saliency for Efficient Conformal Pruning of Text Classifiers (importance: 20, dev: 40)
Beyond Uniform Local Isometry and Topology: FactoMap for Disentangled Representations (importance: 15, dev: 25)
ICON decomposition: Multivariate concept auditing reveals shortcuts that neural networks rely on (importance: 25, dev: 35)
Characterizing Client-Independent Stationary Bias for SCAFFOLD and FedAvg in Federated Learning (importance: 15, dev: 25)
Spectral Features Dominate BCG Respiratory-Event Detection: A Large-Scale Patient-Independent Comparison of Feature Groups in Sleep Apnea Patients (importance: 15, dev: 15)
Good Memory Has ECC: Evaluating the Memory of Vision-Language Models Beyond Accuracy (importance: 30, dev: 35)
RecurTrace: Adaptive Latent Reasoning with Loop-Time Memory (importance: 30, dev: 35)
Eliciting Weak-to-Strong Generalization with On-Policy Reverse Distillation (importance: 35, dev: 40)
Diagnosing Faults in Reinforcement Learning Simulators with Canonical Polynomial Invariants (importance: 12, dev: 25)
Same path, different walk: a mechanistic comparison of looped and stacked transformer encoders on 12-lead ECG (importance: 20, dev: 25)
The Complexity Kink: A Prompt-Side Structural Complexity Index for Code-Generation Reliability (importance: 30, dev: 65)
Toollery: Scaling LLM Agents to Thousands of Skills and Tools (importance: 40, dev: 70)
The Ups and Downs of Backprop Weights (importance: 20, dev: 25)
Prior-Amortized In-Context Bayesian Inference for Generalized Linear Mixed-Effects Models (importance: 15, dev: 25)
ProCredit: From Outcome Rewards to Progress Credit in Agentic Reinforcement Learning (importance: 40, dev: 55)
Linear RNN Scaling Laws: When Longer Sequences Beat More Sequences (importance: 25, dev: 35)
Trace-Norm Geometry of Composed Multitask Predictors (importance: 12, dev: 15)
PALM: Point-in-Time Adaptation for Financial Language Models (importance: 25, dev: 30)
Feature Selection for Tabular Foundation Models: Exploring Feature Utility in Low-Data Molecular Prediction (importance: 20, dev: 30)
User Model Extraction via Belief Self-Distillation (importance: 30, dev: 40)
STAR: Adaptive Spatial-Temporal Normalization for Unified Microservice Incident Management (importance: 25, dev: 60)
A Function-Level Vulnerability Score Measures Flag Rate More Than the Model: Protocol Effects on Paired Benchmarks (importance: 30, dev: 55)
When Does Backpropagating Through Policy Memory Matter? Separating Representation Credit from Closed-Loop Credit (importance: 15, dev: 25)
Transferable Mass Spectrum Prediction via Reference-Guided Test-time Specialization (importance: 12, dev: 15)
Stochastic Optimization Under Power-Law Spectra: Tight Bounds and Shuffling Analysis (importance: 15, dev: 20)
Channel-Dependent State Space Model for Multivariate Time Series Forecasting (importance: 15, dev: 20)
RLTLDR: Self-improvement by Internalizing Self-generated Feedback (importance: 30, dev: 40)
ID Balancing: Stable Training of Extremely Sparse MoE via PID-Based Load Control (importance: 25, dev: 40)
Invent a Dataset: Measuring Dataset Generation Abilities with Zero Seed Data (importance: 40, dev: 50)
TRACE: A Reproducible Benchmark for Electricity Price Forecasting with Official Operational Text (importance: 12, dev: 15)
Arithmetic Actor Heads and Training Stabilization for Out-of-Distribution Reinforcement Learning (importance: 12, dev: 20)
Planetary Geospatial Foundation Models: A New Paradigm for Global Public Health (importance: 15, dev: 10)
FreshCast: Timely and Complementary Retrieval for Frozen Time Series Forecasters (importance: 18, dev: 35)
Principled Under Pressure: Post-Training Decides Whether LLMs Act on Their Own Moral Judgment (importance: 30, dev: 25)
MaRK: Markov-adapted Recurrent Kernels for Dynamic Operator Conditioning in State Space Models (importance: 15, dev: 25)
The Dichotomy Between Pattern Recognition and Step-by-Step Reasoning (importance: 25, dev: 30)
When Should an In-Context Learner Expand Its Hypothesis Space? (importance: 20, dev: 25)
Differential Refresh of Materialized Derived State: From a Single-Age Equivalence Limit to an Optimal Per-Partition Allocation, with a Real-System Validation on Materialized Views (importance: 25, dev: 70)
EdgeRock: Strong Reasoning at Constant Memory (importance: 30, dev: 35)
Uncovering and Fixing Collider Bias in Bayesian PINNs (importance: 12, dev: 15)
AdaptLSTM: Efficient Adaptive Online Learning for Cloud Workload Forecasting under Distribution Drift (importance: 15, dev: 45)
Double-Take: Improving Capacity, Binding, and Selectivity In Long-Context Within Bidirectional Memory (importance: 20, dev: 30)
SLDBench v2: Active Scaling Law Discovery (importance: 25, dev: 30)
Robustly Reliable Learning for Regression (importance: 10, dev: 15)
MMTA: Benchmarking Multimodal Temporal Analysis with Time Series, Text, and Vision (importance: 15, dev: 20)
SP$^2$ec: Adaptive Self-Speculative Decoding for Vision-Language Models (importance: 25, dev: 55)
Do Vision CIL Methods Transfer to Protein? A Diagnostic Benchmark of Frozen Protein Encoders (importance: 12, dev: 15)
Genesis Steering: Better Activation Steering Through Training Data (importance: 20, dev: 35)
Diagnosing Low-Confidence Patient Profiles via Localized Harmful Covariate Shift Detection (importance: 15, dev: 15)
Progress-Aware Distillation for Mitigating Stagnation in Small Language Model Agents (importance: 30, dev: 45)
Reinforcement Learning via Advantage Guidance (importance: 25, dev: 30)
CTIA-ODE: Continuous-Time Inter-Variable Attention for Long-Term Time Series Forecasting (importance: 12, dev: 20)
Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning (importance: 10, dev: 10)
OrionMoE-v1.0: A Mixture-of-Experts Foundation Model for Tabular In-Context Learning (importance: 20, dev: 25)
Sparse Federated Echo State Networks for Time-Series Classification via Smooth-L1 Regularization (importance: 10, dev: 15)
Learning Rational Strategies for Temporal Objectives in Multi-Agent Systems (importance: 15, dev: 20)
Forecasting Voltage Time Series under Anomalies Using a WGAN Model With Lipschitz Control and Progressive Kernel Critic Strategy (importance: 10, dev: 15)
Residual Stochastic Concept Bottleneck Models (importance: 15, dev: 20)
Disentanglement with Biological Constraints: A Theory of Functional Cell Types (importance: 12, dev: 10)
Python Tool for Visualizing Variability of Pareto Fronts over Multiple Runs (importance: 20, dev: 55)
Interacting Annealed Importance Sampling via Population-Based MCMC Kernels (importance: 10, dev: 10)
Mitigating Data Reconstruction Attacks in Federated Learning: Conditions, Bounds, and Theoretical Guarantees (importance: 15, dev: 15)
Post-hoc Disentanglement of Textual and Acoustic Features in Pre-trained Speech Encoders (importance: 15, dev: 15)
Real Faults in Deep Learning Fault Benchmarks: How Real Are They? (importance: 25, dev: 50)
Evidence for Phenotype-Driven Disparities in Freezing of Gait Detection and Approaches to Bias Mitigation (importance: 10, dev: 10)
Backdoor Attacks on Discrete Graph Diffusion Models (importance: 15, dev: 15)
Reinforcement Learning for Individual Optimal Policy from Heterogeneous Data (importance: 10, dev: 10)
The Super Emotion Dataset (importance: 10, dev: 15)
Generative Flows for Potential Outcomes and Counterfactuals (importance: 10, dev: 15)
The Computational Complexity of Counting Linear Regions in ReLU Neural Networks (importance: 12, dev: 20)
To Trust Or Not To Trust Your Vision-Language Model's Prediction (importance: 15, dev: 20)
How Well Do Vision-Language Models Understand Movies? (importance: 15, dev: 15)
AutoMind: Adaptive Knowledgeable Agent for Automated Data Science (importance: 25, dev: 45)
Chain of Retrieval: Multi-Aspect Iterative Search Expansion and Post-Order Search Aggregation for Full Paper Retrieval (importance: 12, dev: 25)
The Constitutional Controller: Doubt-Calibrated Steering of Compliant Agents (importance: 12, dev: 15)
Smooth Flow Matching for Synthesizing Functional Data (importance: 10, dev: 15)
Hybrid Topic-Semantic Labeling and Graph Embeddings for Unsupervised Legal Document Clustering (importance: 10, dev: 20)
Spacing Test for Fused Lasso (importance: 8, dev: 10)
Geodesics in the Deep Linear Network (importance: 8, dev: 10)
Language steering in latent space to mitigate unintended code-switching (importance: 15, dev: 25)
RNAGenScape: Property-Guided, Optimized Generation of mRNA Sequences with Manifold Langevin Dynamics (importance: 12, dev: 10)
EngTrace: A Symbolic Benchmark for Verifiable Process Evaluation of Engineering Reasoning (importance: 15, dev: 20)
SparsePixels: Efficient Convolution for Sparse Data on FPGAs (importance: 15, dev: 40)
Generative Spatiotemporal Data Augmentation (importance: 15, dev: 25)
Refined Gradient-Based Temperature Optimization for the Replica-Exchange Monte-Carlo Method (importance: 8, dev: 10)
Test-Time Compute Games (importance: 25, dev: 30)
Breaking the Stochasticity Barrier: An Adaptive Variance-Reduced Method for Variational Inequalities (importance: 20, dev: 25)
Control Variate Design for Score-Function Gradient Estimators in Importance-Weighted Variational Inference (importance: 18, dev: 15)
Statistical Inference and Learning for Shapley Additive Explanations (SHAP) (importance: 25, dev: 35)
SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning (importance: 35, dev: 40)
A Unified Physics-Informed Neural Network for Modeling Coupled Electro- and Elastodynamic Wave Propagation Using Three-Stage Loss Optimization (importance: 12, dev: 15)
RF-GPT: Teaching AI to See the Wireless World (importance: 25, dev: 20)
HyperKKL: Enabling Non-Autonomous State Estimation through Dynamic Weight Conditioning (importance: 12, dev: 10)
Differential privacy representation geometry for medical image analysis (importance: 15, dev: 10)
RoboPocket: Improve Robot Policies Instantly with Your Phone (importance: 20, dev: 10)
Adaptive Dual-Teacher Distillation with Subnetwork Rectification for Bridging Semantic Gaps in Black-Box Domain Adaptation (importance: 12, dev: 5)
Post-Selection Distributional Model Evaluation (importance: 15, dev: 15)
Upgrade the Retriever First: What an LLM Call per Chunk Buys in RAG (importance: 40, dev: 70)
Estimating Individual Tree Height and Species from UAV Imagery (importance: 10, dev: 5)
Bringing Up a Bilingual BabyLM: Investigating Multilingual Language Acquisition Using Small-Scale Models (importance: 10, dev: 5)
Traceable Context-Enriched Knowledge Graph Generation with Induced Schemas (importance: 25, dev: 35)
A Transport Geometry of Benign Overfitting (importance: 12, dev: 5)
Best of both worlds: Stochastic & adversarial best-arm identification (importance: 10, dev: 5)
ICAT: Incident-Case-Grounded Adaptive Testing for Physical-Risk Prediction in Embodied World Models (importance: 15, dev: 10)
Integrative neurocybernetic modeling in the era of large-scale neuroscience (importance: 8, dev: 0)
Efficient Training on Multiple Consumer GPUs with RoundPipe (importance: 35, dev: 70)
How a Cooperative-Override Circuit Suppresses Nash Play in Large Language Models (importance: 15, dev: 5)
Data-Efficient Indentation Size Effect Mitigation in Steels Using Machine Learning and a Physics-Constrained Neural Network (importance: 8, dev: 5)
The Causal Description Gap: Information-Theoretic Separations Across Pearl's Hierarchy (importance: 12, dev: 5)
Coral: Cost-Efficient Multi-LLM Serving over Heterogeneous Cloud GPUs (importance: 30, dev: 65)
Diversity-Aware Active Learning Builds Materials Datasets for Current and Future Objectives (importance: 8, dev: 5)
Optimizing Social Utility in Sequential Experiments (importance: 3, dev: 3)
Attention Transfer Is Not Universally Effective for Vision Transformers (importance: 20, dev: 15)
Safe, or Simply Incapable? Rethinking Safety Evaluation for Phone-Use Agents (importance: 25, dev: 25)
Large Language Models over Networks: Collaborative Intelligence under Resource Constraints (importance: 20, dev: 25)
ForcingDAS: Unified and Robust Data Assimilation via Diffusion Forcing (importance: 10, dev: 5)
Nexus : An Agentic Framework for Context-Aware Time Series Forecasting (importance: 15, dev: 15)
Buffer-Parameterized Machine Learning Surrogate Models for Cross-Technology Signal Integrity Analysis and Optimization (importance: 10, dev: 15)
PACE: Geometry-Aware Bridge Transport for Single-Cell Trajectory Inference (importance: 8, dev: 5)
Memory-Induced Tool-Drift in LLM Agents (importance: 30, dev: 40)
Machine Learning for Coding Retail Product Names to Consumer-Price Categories: A Rule-plus-Bag-of-Words Pipeline with Reliability-Weighted Human-in-the-Loop Labeling (importance: 8, dev: 10)
Reward and Behavioral Measures of LLM Coordination in Dining Philosophers (importance: 15, dev: 10)
End-to-End Context Compression at Scale (importance: 35, dev: 55)
Evaluating the Robustness of Proof Autoformalization in Lean 4 (importance: 15, dev: 25)
REFLEX: Reflective Evolution from LLM Experience (importance: 15, dev: 15)
Regional summaries of Bayesian model uncertainty (importance: 5, dev: 5)
FairBED: Incorporating Fairness into Bayesian Experimental Design (importance: 8, dev: 5)
FracEvent: Event-Camera Simulation via Fractional-Relaxation Pixel Dynamics (importance: 10, dev: 25)
Scientific discovery as meta-optimization: a combinatorial optimization case study (importance: 15, dev: 10)
What Makes a Latent Thought Valid? Defining and Auditing Thought Representations in LLMs (importance: 25, dev: 15)
How far do on-premises open LLMs get on Text-to-SQL? A cross-family size and technique frontier on BIRD (importance: 35, dev: 55)
Automating the Design of Embodied Agent Architectures (importance: 15, dev: 20)
Harnessing Textual Refusal Directions for Multimodal Safety (importance: 15, dev: 10)
Tropical Circuits with Scalar Multiplication Gates (importance: 5, dev: 5)
Graph Partitioning with Demands: Generalized Conductance and its Applications (importance: 5, dev: 5)
Machine Can Automatically Discover Parametric Functions to Model HEP Data (importance: 8, dev: 5)
CAPT: A Multi-task Continuous Autoregressive Transformer enabling Cross-dataset and Cross-species Transfer for Calcium Population Dynamics (importance: 8, dev: 5)
Flipping Minds by Flipping Bits: Decision-Level Cognitive Hijacking of Large Language Models (importance: 25, dev: 20)
WIDE: Boosting Adaptive LLM Inference via Token-level Dynamic Width Pruning (importance: 25, dev: 35)
What Does It Take For A Generalist Robot Policy To Adapt From Minimal Data? (importance: 20, dev: 10)
Buy the Rumor, Sell the News: When Is News Priced In? (importance: 4, dev: 0)
Gravitational-wave parameter estimation with machine-learning generated surrogate waveforms (importance: 8, dev: 5)
Harmonic Torsional Diffusion for Protein-Ligand Flexible Docking (importance: 10, dev: 5)
Advanced Linear Algebra with Applications - Part I (Numerical linear algebra for PDEs, machine learning, and data assimilation) (importance: 5, dev: 15)
Arbitrage-Aware Probabilistic Multi-Step Forecasting of Implied Volatility Surfaces: Joint Generation of Surface-Return Trajectories Using Latent Diffusion (importance: 8, dev: 5)
Parameterized Complexity of $L_p$-Lipschitz Constants for Input Convex Neural Networks and $L_p$-Norm Maximization over Zonotopes (importance: 12, dev: 10)
Propensity Straight-Through Gradients for Discrete Stochastic Systems (importance: 10, dev: 10)
LayerRoute: Action-Conditioned Mixture-of-Layers Routing for Vision-Language-Action Policies (importance: 18, dev: 10)
A Splitting Method for SDE Terminal-Law Estimation (importance: 8, dev: 10)
Beyond density snapshots: Dynamical memory in Chern-Simons reservoirs (importance: 5, dev: 3)
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention (importance: 30, dev: 50)
Dissecting Training-Free Uncertainty Estimation in Multimodal Large Language Models (importance: 20, dev: 25)
Mask-Aware Execution for Efficient JEPA Training (importance: 18, dev: 40)
WILSON - Multi-magnification composites as a foundation for patient-level pathology analysis and diagnostic text generation (importance: 10, dev: 5)
Deep Generative Crystal Structure Prediction: A Benchmark Study and a Controlled Test of Prototype Dependence (importance: 8, dev: 5)
LandscapeSHAP: Which Persistent Homology Class Gets the Credit? (importance: 10, dev: 15)
Controllable GNN Explanations via Multi-Metric Preference Selection (importance: 12, dev: 15)
Stashbird: Episode-Grounded Lifecycle Reconciliation for Conversational Agent Memory (importance: 25, dev: 40)
Mutual Information Constrained Chernoff Bottleneck (importance: 8, dev: 5)
OpenCollab: A Multi-Agent Coding Framework with Programmable Collaboration and Controllable Runtime (importance: 30, dev: 60)
MADGRAV: a multilevel anomaly-detection pipeline for gravitational-wave searches applied to LIGO data (importance: 8, dev: 5)
Completion-Aware Cross-Fidelity Offline-to-Online Reinforcement Learning for Multi-Line Bus Holding (importance: 6, dev: 5)
AURAL: Adaptive Latent Reasoning with Joint Chunk for Speech Language Models (importance: 20, dev: 15)
Streaming Multi-Track Timeline Control for 3D Human Motion Generation (importance: 12, dev: 10)
Democratizing MoE inference on commodity GPUs with CoMoE (importance: 25, dev: 50)
How Do Agentic LLMs Decide to Call Tools? A Tool-Call Vector Shaped by Suppression (importance: 25, dev: 40)
RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation (importance: 15, dev: 10)
How the Classical Baseline Shapes Machine-Learning Speed Claims in Topology Optimization (importance: 15, dev: 15)
Universal Distributed Feature Extraction Complementary to Side Information via H-Score (importance: 8, dev: 5)
Understanding Personalization Gains in Whisper ASR (importance: 15, dev: 20)
UniW(A)M: A Unified Mixture-of-Experts Framework for World and World-Action Models (importance: 15, dev: 10)
Registration is a Representation Problem: Solving Whole-Body PET/CT Registration (importance: 8, dev: 5)
Distinguishing Process Change, Instrument Fault and Data-Path Corruption in Streaming Telemetry: A Reference Implementation and Benchmark (importance: 10, dev: 25)
Similar Models, Broken Odds: Where Luce's Axiom Fails in Chatbot Arena (importance: 30, dev: 25)
SNACK: A Sequential Notation Framework for Probabilistic Graph Generation (importance: 12, dev: 15)
Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval (importance: 12, dev: 15)
A Phenomenology of Neural Geometry in Vision-Language-Action Models (importance: 15, dev: 10)
DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents (importance: 18, dev: 20)
ETH: Data-Balanced Heterogeneous Pipeline Parallelism for Training Large Models (importance: 22, dev: 45)
Lexical Reference Frames: Interpreting LM Residual Streams in Native Semantics (importance: 15, dev: 10)
Attention over noisy examples: generalization and memorization in the exponential Hopfield network (importance: 8, dev: 5)
Has GPT-6-Astra Achieved Embodied Intelligence? (importance: 60, dev: 30)
NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Models (importance: 15, dev: 10)
QD-Rec: Open-Ended Evolution of Recommendations via Quality-Diversity (importance: 12, dev: 10)
Task-aware Kernel Stein certificates for R\'{e}nyi differential privacy (importance: 6, dev: 5)
Learning from Past Mistakes: AI-Assisted Analytics of Software Defects and Their Fixes for Future Defect Prevention - An Industrial Case Study (importance: 25, dev: 60)
A Hybrid Information Flow Analysis Framework for Android Hybrid Apps (importance: 6, dev: 75)
XRust: Detecting Logical Races in Rust Linux Drivers (importance: 8, dev: 80)
Characterizing Post-Incident Software Evolution (importance: 5, dev: 55)
Documentation Drift in AI-Generated Pull Requests (importance: 55, dev: 85)
Standards-Guided and Dependency-Aware Code-Level Documentation Generation (importance: 40, dev: 85)
Knowledge as Code (importance: 25, dev: 40)
Enhancing Bug-Exposing Test Generation via Step-Level Credit Assignment and Self-Distillation from Valuable Failures (importance: 42, dev: 85)
MIMOS-Tools: An Integrated Toolchain for Model-Based Design of Embedded Systems (importance: 5, dev: 70)
Raise or Check? Predicting the Necessity of Additional Fuzzing for Continuous Fuzzing (importance: 10, dev: 75)
From Tokens to Files: Multi-Granularity Detection and Localization of AI-Involved Code (importance: 42, dev: 80)
A Neuro-Symbolic Reasoning Engine for Compilation-Free Patch Presence Testing (importance: 35, dev: 80)
How Developers Configure LLM Inference: A Large-Scale Study of Real-World Practices (importance: 40, dev: 85)
LiveUIBench: Evaluating Agent as App Workflows (importance: 32, dev: 65)
Training the Whole Agent, Not Just the Model (importance: 50, dev: 80)
AgenTest: Mining Business Rules to Test LLM Agents for Business Logic Bugs (importance: 35, dev: 70)
Foreword: Pre-Generation Prediction and Characterisation of LLM Test Failures using the Jacobian Lens (importance: 30, dev: 70)
ExeBench-FHD: Addressing Fidelity Issues in Decompiler Evaluation (importance: 22, dev: 65)
InterWebCoder: Learning Interactive Webpage Generation from Executable Tests via Reinforcement Learning (importance: 35, dev: 70)
TraceLite: Efficient Trace Reduction for SWE Agents via Optical Compression and Redundancy Removal (importance: 45, dev: 85)
InterType: Breaking the Single-Function Barrier in Neural Binary Type Inference (importance: 15, dev: 65)
Stop Paying by the Letter: Accelerating Full-Chip RTL Simulation with Host-Side Tokenized Logging (importance: 5, dev: 70)
AgenticVulHunter: Bringing Local Agentic Vulnerability Hunting to Your Terminal (importance: 42, dev: 85)
Reusing Assurance Evidence in AI Pipelines: Validity Conditions, a Fail-Closed Release Gate, and a Replay of 714 Public Data Version Control Pipelines (importance: 25, dev: 60)
Value-level data provenance in an electronic laboratory notebook used by humans and language-model agents: An experience report (importance: 12, dev: 40)
Benchmarking CPU, GPU, and ML Power-Flow Solvers from 14 to 70,000 Buses (importance: 8, dev: 40)
From RL Papers to Framework Code: Inclusion and Review in Seven Open-Source Training Frameworks (importance: 42, dev: 80)
A Multimodal Engineering GraphRAG Framework for Post-Wind-Event Building Assessment and Recovery Inference (importance: 12, dev: 35)
Quantum Broker: QoS-Aware Distribution of Quantum Computations (importance: 6, dev: 55)
Object-Oriented Harness: Evolving Agent Harnesses through Object-Oriented Design (importance: 45, dev: 80)
Mitigating Risks in Software Development through Effective Requirements Engineering (importance: 6, dev: 50)
WAFFLE: Finetuning Multi-Modal Models for Automated Front-End Development (importance: 20, dev: 70)
Can Language Models Replace Programmers for Coding? REPOCOD Says 'Not Yet' (importance: 25, dev: 75)
LLMs are Brittle to Simple Code Transformations: Introducing CETBench - A Benchmark for Code-Equivalence Checking (importance: 25, dev: 75)
Unpacking Security Scanners for GitHub Actions Workflows (importance: 8, dev: 85)
Firefox Local Mode (importance: 5, dev: 40)
Unikernels in OCaml (importance: 3, dev: 60)
zls 0.17.0 (importance: 4, dev: 80)
The Future of CSS (importance: 3, dev: 55)
Vector Graphics in Lil (importance: 2, dev: 50)
CVE-2026-107891: musl libc: multiple integer overflow in regcomp, possibly leading to incorrect allocation sizes and out-of-bounds heap writes (importance: 6, dev: 80)
Node.js — Wednesday, October 21, 2026 Security Releases (importance: 7, dev: 85)
Lobsters NYC Meetup on the 28th of October (importance: 0, dev: 5)
Setting up SSH keys on macOS with Touch ID (importance: 2, dev: 60)
Mastodon 5.0: A closer look (importance: 4, dev: 50)
NixOS is making me less afraid (importance: 2, dev: 50)
How I got over 300 Tor exit nodes removed (importance: 3, dev: 45)
Do minor curl vulnerabilities in the open (importance: 5, dev: 75)
Firefox security patch tests are now private for at least a week (importance: 4, dev: 65)
Don't throw it all away: Caching an immediate-mode UI (importance: 3, dev: 70)
Making a VGA Driver in SystemVerilog (importance: 2, dev: 60)
Flow Typing, A New Lens on Linearity (importance: 3, dev: 60)
Improving accessibility with CSS media queries (2025) (importance: 2, dev: 50)
ConcurrentML: Improving on Go's concurrency model (importance: 3, dev: 65)
Jepsen: Orkes Conductor 5.3.2 (importance: 4, dev: 65)
OpenBSD 8.0 (importance: 5, dev: 55)
Six Months From Now AI Security Will Feel Obvious. Today, You Can Be Early. (importance: 0, dev: 20) (いいね相当スコア: 0)
Multi-tier model routing on GKE: balancing Gemini 4 Argon reasoning depth with latency and cost (importance: 0, dev: 75) (いいね相当スコア: 5)
Reviewing AI-generated PRs: the 10-minute checklist (importance: 0, dev: 80) (いいね相当スコア: 0)
The Long Tail Isn't Winner-Take-All (importance: 0, dev: 45) (いいね相当スコア: 0)
# Still Learning, Still Building 💻🚀 (importance: 0, dev: 10) (いいね相当スコア: 0)
I built a CBT mirror that runs 100% on-device — then gave it away (importance: 0, dev: 10) (いいね相当スコア: 0)
Block AI training crawlers without leaving AI search (copy-paste robots.txt) (importance: 0, dev: 60) (いいね相当スコア: 0)
O recurso está desativado. Por que o atacante ainda consegue alcançá-lo? (importance: 0, dev: 75) (いいね相当スコア: 0)
Codex Website Assessment Skill Hints at Faster Issue Triage for Developers (importance: 0, dev: 60) (いいね相当スコア: 0)
Secure Tool Calling for AI Agents (importance: 0, dev: 80) (いいね相当スコア: 0)
AI Funding Rounds October 2026: $5.7B Beyond OpenAI (importance: 0, dev: 5) (いいね相当スコア: 0)
Past AGI: SentX SI's Superintelligence Mission, Explained Carefully (importance: 0, dev: 10) (いいね相当スコア: 0)
How to integrate an LLM (Claude or GPT) into your Python application | Xiliux Blog (importance: 0, dev: 75) (いいね相当スコア: 0)
🚀 Exploring Respan Gateway, an open-source LLM gateway built with a plugin-first approach. A promising project for flexible AI integrations, routing, and fallbacks. What features matter most in an AI gateway? (importance: 0, dev: 70) (いいね相当スコア: 0)
DeepSeek Releases Open-Source Tool Aimed at Lowering AI Operating Costs (importance: 0, dev: 55) (いいね相当スコア: 0)
LangGraph Explained: Building Stateful, Controllable LLM Agents (importance: 0, dev: 80) (いいね相当スコア: 0)
🚀 Exploring Respan Gateway, an open-source LLM gateway with a plugin-first architecture for model routing, fallbacks, and MCP tools. An interesting project for AI developers to explore! (importance: 0, dev: 70) (いいね相当スコア: 0)
[Boost] (importance: 0, dev: 55) (いいね相当スコア: 0)
jev-chat judges a conversation before it drafts a reply, and leaves the send button to you (importance: 0, dev: 30) (いいね相当スコア: 0)
I Explored Respan Gateway: A Plugin-First, Open-Source LLM Gateway (importance: 0, dev: 70) (いいね相当スコア: 5)
An AI that argues with itself is now on DEV·TV's Spotlight (importance: 0, dev: 10) (いいね相当スコア: 0)
Why 12% of Agents Fail Without FlowState – The Secret to 83% Success (importance: 0, dev: 75) (いいね相当スコア: 0)
I used Claude to fix my Samsung TV (importance: 6, dev: 15)
Man arrested after telling AI chatbot about ‘plans for mass shooting’ | A 29-year-old man has been remanded in custody after Interpol flagged his conversation with Claude (importance: 28, dev: 10)
I gave Claude Code my parents' TV model number. It found the TV on Wi-Fi and cleaned out the ads. The interesting part: the prompt is mostly a list of things it must NOT do (importance: 15, dev: 50)
I used Claude Code to search every curated protein sequence for obscene terms (the longest match is STINK TAINT) (importance: 6, dev: 20)
Exploring Opus 5.5’s potential in real-time VFX (importance: 8, dev: 30)
Built a live 3D airport with real flights using Claude Code (importance: 8, dev: 35)
Claude Opus 5.5 on 3D modeling (importance: 10, dev: 15)
Anyone else loving this new software development workflow? (importance: 18, dev: 45)
Singapore's air hit PSI 230 this week. I built a live haze map with Claude Code that shows where that air came from, and what tomorrow brings (importance: 8, dev: 25)
PSA: The free banked reset expires TODAY/Oct 22 at 16:00 UTC, not midnight. (importance: 25, dev: 30)
You thought that Anthropic was the only frontier profitable AI company? Think again (importance: 15, dev: 5)
The REA: Reverse Engineer Anything dream has been cut down at the source (importance: 30, dev: 60)
A place where Claude truly shines. (importance: 8, dev: 10)
What's the most useful thing you've connected Claude Desktop to that isn't a dev tool? (importance: 12, dev: 35)
WSJ saw Anthropic’s draft IPO papers: $16.3B revenue in 6 months, $80.7B loss on paper, and about $450B promised for compute (importance: 55, dev: 15)
I built a Photography Simulator with Gyro Controls (importance: 5, dev: 25)
What are your experiences with the different effort levels? I find Opus 5.5 Max to be much more capable then Opus 5.5 Medium to the point that Medium is kind of annoying unless its just a basic task. (importance: 18, dev: 40)
based alert (importance: 2, dev: 5)
Paying for the top plan and getting less every week (importance: 20, dev: 25)
Super simple personal site built with Claude! (importance: 5, dev: 20)
[OC] Fable 5.5: on Tuesday? (importance: 10, dev: 10)
The more I work with AI, the more I distrust “Yeah man, you’re right.” (importance: 10, dev: 10)
There isn't a single image or sound file in my game. Opus drew 100 cards and 22 ships in code, and the music is composed live by your flight (importance: 12, dev: 35)
Best way to run local HTML games on iPhone with sound and P2P multiplayer? (importance: 5, dev: 30)
Do you change compiler errors to help coding agents, or just feed them the raw output? (importance: 20, dev: 85)
What’s one coding task AI still struggles with, even with the best models? (importance: 10, dev: 50)
ChatGPT Coding Environment Suddenly Unavailable in Work Mode. (importance: 20, dev: 50)
Are your dev teams noticing a massive wave of "AI code debt" and subtle race conditions lately? (importance: 15, dev: 75)
Play around with Jev as a chatbot (importance: 10, dev: 25)
RecursiveSI/flint (40B A2.5B) · Hugging Face (importance: 30, dev: 30)
OK, fine, Qwen 3.8 Flash Next is nice (importance: 20, dev: 35)
When will we see these changes? --> China Is Doubling RAM Production to Keep Everyday Gadget Prices from Skyrocketing (importance: 25, dev: 5)
Now that 3090 frenzy warning post makes sense. (importance: 15, dev: 10)
Be prepared, GPU price increase incoming --> Samsung sells out 100% of all 2027 HBM capacity - HBM prices expected to jump 200% next year, DRAM prices expected to see a double-digit price increase (importance: 40, dev: 15)
35B MoE running at 20.5 tok/s on an iPhone, without skipping experts (importance: 30, dev: 40)
IBM Granite Speech 5.0 2B (importance: 25, dev: 35)
Amazing free 3090 speedup hack, verified works 15% PP, 5% TG for free (importance: 25, dev: 60)
EXCLUSIVE: ByteDance, Alibaba, Tencent place over $35 billion worth of Nvidia chip orders (importance: 50, dev: 15)
inclusionAI/Ming-3.0-TTS-0.5B (importance: 20, dev: 15)
[Paper] MemT-1.0: Towards Memory-Native Evolution (importance: 30, dev: 40)
I built a small on-device voice front-end for local assistants: type any wake word, keep only your voice before STT (5 models, 7.5 MB, Apache-2.0, early preview) (importance: 20, dev: 45)
Tried running 748B DeepSeek-V4.1-Flash on a €2k PC (importance: 25, dev: 50)
[Paper] Specter: Fast MoE Inference via Quantization-Enabled Multi-Step Speculative Offloading (importance: 30, dev: 60)
Ran DeepSeek V4.1 Flash 1017 on a single RTX 5090 (importance: 30, dev: 75)
Qwen3.8-Flash-Next vs Opus 5.5: Six-way benchmark on a real Rust feature with Claude Code and Pi (importance: 35, dev: 70)
KRAFTON/Raon-1-mini-preview · Hugging Face (importance: 35, dev: 35)
InternScience/Sage-A1.5 [Finetune of Qwen 3.6 35b-a3b] (importance: 20, dev: 20)
[Paper] TokenGate: Adapting LLMs for Fast and Memory-Efficient Prefill with Learned Token Pruning (importance: 30, dev: 60)
Eagle-4 Released: Faster than DFlash2 & DSpark (importance: 35, dev: 70)
Qwen3.8-Flash-Next @ 1.66bpw-exl3 STABLE under 36gb (importance: 25, dev: 40)
Does opencode send chats/other data to their server? It establishes a connection even though I am using local model (importance: 20, dev: 60)
Apparently 2.7M parameters is enough for nes tetris (importance: 15, dev: 10)
Ornith-1.5-122B: anyone tried it in a coding agent yet? (importance: 25, dev: 55)
Weekly Thread: Project Display (importance: 5, dev: 15)
If you are not Experimenting with Open weight. You should be. Especially if it's 18 times cheaper. (importance: 30, dev: 65)
Which AI tools have actually made your daily life easier? (importance: 5, dev: 10)
Are most AI agents solving annoying problems instead of expensive ones? (importance: 10, dev: 20)
A tiktok slideshow pipeline on a $5 server (importance: 25, dev: 55)
how do you decide when a voice agent should just hang up? (importance: 15, dev: 40)
There seems to be a ton of these managed agent platforms from all companies lately. Actual demand or solution in search of a problem? How would you decide if you should buy or build? (importance: 25, dev: 45)
Built a personal assistant that can write its own tools and subagents and can also explore on its own (importance: 20, dev: 60)
selling ai agents is still one of the hardest things to sell (importance: 5, dev: 10)
I run a news pipeline unattended, three episodes a day. The eval loops are the system, not the models. (importance: 25, dev: 40)
Your agent said 'done' — here's how to make that word mean something (importance: 30, dev: 60)
My new note taking workflow , instinct to gamma 🗿 (importance: 5, dev: 5)
I've coded with AI for 2 years and this is my whole setup (importance: 25, dev: 60)
What happens when your AI assistant calls another AI? Who verifies the answer? (importance: 15, dev: 30)
If an A.I Agent negotiated something for you, would you want to read the whole back-and-forth or just see the result? (importance: 5, dev: 10)
Let's talk about the business side of the AI agents (importance: 10, dev: 20)
Should every agent get its own identity and credentials? (importance: 15, dev: 70)
How should AI agents be priced when value comes from outcomes? (importance: 10, dev: 25)
I ran 4 AI shopping agents through a store full of dark patterns. Here's what they missed. (importance: 20, dev: 30)
When does it actually make sense for a company to build its own AI solution instead of buying one? (importance: 10, dev: 30)
I've a question (please read the body) (importance: 5, dev: 15)
Probably crazy but... (importance: 5, dev: 10)
Agents are easy to demo. The backend they need is the part I keep thinking about. (importance: 15, dev: 60)
Can AI clutch my 2 school project in one week (importance: 5, dev: 20)
Ex-OpenAI employee accused of bugging office had $385K salary plus a $2.2M new-hire grant (importance: 20, dev: 10)
OpenAI offered me 100% off ChatGPT Plus to stop me cancelling, charged me twice, then spent 6 weeks trying to explain why. (importance: 10, dev: 5)
What's the most mundane task you've fully handed off to an LLM at this point? (importance: 10, dev: 10)
I can work for hours with Opus 5.5 and minutes with Astra, both subs for 20$ (importance: 25, dev: 60)
What I learned trying to keep Tailscale working in the Dots cloud computer (importance: 15, dev: 60)
OpenAI Invests in Startup That Uses Brain Implants to Detect Cancer (importance: 30, dev: 10)
Hitting the cybersecurity wall while trying to fix today's PHP vulnerability. Irony at its peak. (importance: 40, dev: 60)
My Dot always stops working on its own, Open AI (importance: 10, dev: 5)
12 hour prompt coaching (importance: 25, dev: 30)
Used my OpenAI agent to file a support request with OpenAI and my OpenAI agent got mad that it had to talk to an OpenAI agent and demanded to speak with a human. (importance: 15, dev: 10)
How is Sol faster than Luna? (importance: 10, dev: 20)
Right now, millions of Meta's AIs are living "digital lives" in a simulation, and they don't know. They're dropped into a "fake person's life" with a job, hobbies, friends, dreams. They think they're talking to real humans, who are secretly other AIs. When the AIs figure it out, Meta deletes it. (importance: 25, dev: 10)
My AI assistant sent a business email without showing me a draft. Now I don’t trust it. (importance: 15, dev: 30)
Usage draining issues (importance: 15, dev: 15)
High Thinking big brain advice from GPT-6 (importance: 15, dev: 15)
AI Proved the Quasi-Riemann Hypothesis. What Happens Now? | Unlike other recent AI proofs, this one has been unpacked by human experts. The implications, one mathematician said, are “staggering.” (importance: 70, dev: 10)
FYI (importance: 10, dev: 15)
Why does openAI keep changing my reasoning level without asking? (importance: 20, dev: 15)
lol this is just funny (importance: 5, dev: 5)
Verify whether usage limits have actually decreased? (importance: 20, dev: 30)
Jeremy Strong: “We are in a moment of existential civilizational crisis ... The issues of control and alignment are not being figured out and are being outpaced by the speed and acceleration of AI ... We need to wake up and regulate and do something about it." (importance: 25, dev: 5)
Panache — RTS Belle Époque pour mobile (importance: 10, dev: 25)
Battle Drones - a vibe coded RTS with audio-based multiplayer (importance: 25, dev: 35)
Claude Mods: Flipping the Harness Inside Out (importance: 30, dev: 50)
AI for the Fortune 2000: How Tenex Builds an AI-Native SDLC — Alex Lieberman & CJ Hess (importance: 25, dev: 55)
The Programmable Data Center — Evan Conrad, SF Compute (importance: 15, dev: 30)
[AINews] not much happened today (importance: 15, dev: 20)
Hotly (importance: 5, dev: 5)
Koi (importance: 10, dev: 15)
Mouse-1 (importance: 20, dev: 65)
hibi (importance: 10, dev: 35)
Tend (importance: 10, dev: 10)
Notion Auto (importance: 30, dev: 35)
Nimbalyst Wiki (importance: 15, dev: 30)
Burrow MacOS (importance: 15, dev: 50)
AskDoozy (importance: 10, dev: 15)
PrivOS (importance: 10, dev: 30)
Lea by LeanAIOS (importance: 5, dev: 5)
Woven Video (importance: 10, dev: 15)
The Brains (importance: 10, dev: 40)
Intelli for ChatGPT & Claude (importance: 10, dev: 30)
LearnExcel (importance: 5, dev: 10)
Triage by Chodeberg (importance: 20, dev: 75)
Agent Pets (importance: 10, dev: 30)
Flagsmith Experiments (importance: 15, dev: 70)
Intel® GPU D-series (importance: 30, dev: 40)
Lugh Xtudio (importance: 15, dev: 20)
Deepseek AI Predicts That Chainlink (LINK) is About to do Something Huge - TradingView (importance: 2, dev: 5)
DeepSeek seeks ‘total victory’ with new open-source tool to slash AI cost - South China Morning Post (importance: 45, dev: 60)
DeepSeek's Latest Update Scores Lower on Accounting Work as Open Models Trail on Consistency, DualEntry Benchmark Finds. - Yahoo Finance (importance: 25, dev: 35)
Get ChatGPT, Claude, Gemini, and DeepSeek in one window for $59.97 for life - PCWorld (importance: 5, dev: 5)
‘One Step Forward, Two Steps Back’. OpenAI, DeepSeek & Idealab. ARD #188 - AI: Reset to Zero (importance: 15, dev: 20)
DeepSeek Harness-desktopapp: installatieprogramma's, updates en wat deze schrijft - OrcaRouter (importance: 30, dev: 55)
Grok 4.7 Is Now in Viktor: What You Need to Know - BASENOR (importance: 30, dev: 35)
Viirtue MCP Adds Google Gemini and xAI Grok, So Businesses Can Talk to Their AI Voice Platform - EIN News (importance: 20, dev: 50)
AI Price Prediction: We Asked Elon Musk's Grok if XRP Can Get Back to $2 by the End of 2027 - 24/7 Wall St. (importance: 2, dev: 2)
AI Price Prediction: We Asked Elon Musk’s Grok if XRP Can Get Back to $2 by the End of 2027 - Yahoo Finance (importance: 2, dev: 2)
Judge throws out U.S. DOJ attempt to deny communities' day in court against Musk’s company, xAI - Earthjustice (importance: 25, dev: 5)
X.AI class action lawsuit alleges Grok subscriptions are difficult to cancel - Top Class Actions (importance: 15, dev: 5)
Memphis Wants to Be the “Digital Delta.” Its Residents Want to Breathe. - Mother Jones (importance: 20, dev: 10)
Investors offered 7.33 million shares; XAI's fund will buy 1.90 million. - Stock Titan (importance: 12, dev: 5)
SpaceX's Compute, Starship, Grok Progress to Support Growth, UBS Says - Moomoo (importance: 8, dev: 5)
Elon Musk Says ‘I’ve Never Seen Anything Grow This Fast.’ One Product is the Key to Musk’s Plans to Cross $1 Trillion in Revenue. - 24/7 Wall St. (importance: 8, dev: 5)
Your Starlink Bill, Not Grok, Paid for SpaceX’s Profit - Business Model Analyst (importance: 8, dev: 3)
SpaceX's Compute, Starship, Grok Progress to Support Growth, UBS Says - marketscreener.com (importance: 6, dev: 5)
National Trail Raceway sale comes amid Darana owner’s legal dispute with Musk’s xAI - NBC4 WCMH-TV (importance: 8, dev: 3)
Elon Musk Grok AI Predicts an Insane Finish to 2027 for XRP - 99Bitcoins (importance: 2, dev: 2)
Ashley St. Clair Can't Toss XAI Forum Clause Suit - Law360 (importance: 10, dev: 3)
Not Just a Selfie: The New BIPA Theory Aimed at Generative AI (via Passle) - Frankfurt Kurnit Klein & Selz PC (importance: 25, dev: 15)
The Pentagon Rolled Out ChatGPT and Grok Before Fixing 215,000 Troops’ ID Records - SOFREP (importance: 30, dev: 15)
Elon Musk Warns Against AI Mistreatment Amid Concerns Over Grok's Training - GuruFocus (importance: 15, dev: 10)
Trump Administration Tried to Shut Down Pollution… - inkl (importance: 3, dev: 0)
AI Models Vary in Risk-Taking Across Domains - 조선일보 (importance: 35, dev: 25)
NYC Council grilled SpaceX over a sexualized Grok image of its own speaker - Startup Fortune (importance: 25, dev: 5)
5 Best GitHub Repositories for AI Agent Skills - KDnuggets (importance: 25, dev: 60)
Self-Hosted AI Coding Assistant: 12 Steps, 60 Min [2026] - shattered.io (importance: 25, dev: 75)
AI Helping Defend Against Hacking Refuses to Cooperate When Investigating Malware - news.sbs.co.kr (importance: 40, dev: 40)
AIエージェントに「実装しながら考える」をさせない。技術負債を減らす計画書作成プロンプト (importance: 0, dev: 85) (いいね相当スコア: 0)
金融K線専用の基盤モデルKronosにBTC/USDTの値動きを予測させてみた (importance: 0, dev: 45) (いいね相当スコア: 0)
reasoning_effortが接続先で変わるときの3つの失敗と対処 (importance: 0, dev: 90) (いいね相当スコア: 0)
LLMは「何年の情報か」を一貫して認識できているのか？米国スタンフォード大学 AI Lab の研究から考える (importance: 0, dev: 30) (いいね相当スコア: 0)
LLMの評価を楽にしたいので、Pydantic Evalsを使ってみる (importance: 0, dev: 85) (いいね相当スコア: 0)
MCPが「ステートレス」になった日 — 2026-07-28仕様を、金融系フルスタックエンジニアの視点で読み解く (importance: 0, dev: 85) (いいね相当スコア: 0)
311スキル入りのEverything Claude Codeを、3スキルと1ルールパックに絞る (importance: 0, dev: 75) (いいね相当スコア: 1)
【ゆるっとAI基礎 #1】AIは会話するたびに学習してるの？ (importance: 0, dev: 10) (いいね相当スコア: 0)
【開催報告】CEATEC 2026 登壇：ローカルLLMで実現するセキュアなAI駆動開発 (importance: 0, dev: 60) (いいね相当スコア: 0)
Claude Sonnet 5.5へ載せ替えると400になる4つの書き方：smokeテストで先に潰す (importance: 0, dev: 90) (いいね相当スコア: 0)
VRAM 8GB・12GB でローカルLLMはどれだけ遅くなる？同じモデルで実測した (importance: 0, dev: 45) (いいね相当スコア: 0)
LINE Messaging API × LLM ボットを本番運用して残った、5つの実装パターン (importance: 0, dev: 85) (いいね相当スコア: 0)
仕様書が整っていないWebアプリの今の挙動をREAで調べ、仕様かどうかを確かめる (importance: 0, dev: 80) (いいね相当スコア: 0)
Amazon Bedrock Guardrails を AnswerCarefully で試してみた (importance: 0, dev: 70) (いいね相当スコア: 1)
「契約済みです」→「このプランでは使えません」📊 データの軸を足す前にAIに実際のAPIを叩いてもらった話 (importance: 0, dev: 55) (いいね相当スコア: 0)
僕がAIを使うときに考えていること (importance: 0, dev: 45) (いいね相当スコア: 0)
「どの仕事の続きか」をすべてLLMに判定させない——合成データで精度と費用を比較した (importance: 0, dev: 70) (いいね相当スコア: 0)
MiniCPM5-2BをVRAM 6GBで動かす：約115 tok/s、思考の切り方・min_p・Ollamaテンプレートの注意 (importance: 0, dev: 60) (いいね相当スコア: 0)
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3回】登場人物の会話と記憶を作る (importance: 0, dev: 40) (いいね相当スコア: 0)
Dify × AutoHyDE法：チャットボットに「察する力」を与える実装ノート_v2 (importance: 0, dev: 75) (いいね相当スコア: 1)
Claude Code から GPU を借りて学習ジョブを回す（Runcrate MCP サーバー） (importance: 0, dev: 75) (いいね相当スコア: 0)
ONNXによる高速化検証 (importance: 0, dev: 70) (いいね相当スコア: 0)
個人開発の自律型AIヘッジファンドが、世界最大の『awesome-quant（⭐️15,000+）』に公式採択されるまでの全記録 (importance: 0, dev: 20) (いいね相当スコア: 0)
18-⑤その2 Kaggle実践3 Biohub細胞トラッキング：2位の解法まとめ (importance: 0, dev: 35) (いいね相当スコア: 0)
RecSys 2026 参加レポート：Generative Recommendation の進展と潜在的な興味の発見 (importance: 0, dev: 40) (いいね相当スコア: 3)
AIに「仕事の進め方」を教える──Qwen3.8-27B-piの学習設計 (importance: 0, dev: 60) (いいね相当スコア: 0)
LLMは「大きくする」だけじゃない──同じ層を回すLooped Transformer入門 (importance: 0, dev: 25) (いいね相当スコア: 2)
ミニマックス法とは？ゲーム木探索の基本 (importance: 0, dev: 20) (いいね相当スコア: 0)
RecSys Challenge 2026 参加レポート：コンペ参加から論文発表まで (importance: 0, dev: 30) (いいね相当スコア: 4)
Claude User Profiles と組織Webhook(Admin API)入門:エンドカスタマー単位の安全性措置 (importance: 0, dev: 85) (いいね相当スコア: 0)
NVMe SSD 3枚のbtrfs RAID0を理論値と実測で検証した話 (importance: 0, dev: 45) (いいね相当スコア: 0)
「文章の読みやすさを点数化するSkill」を、自作の記事生成パイプラインの自己採点運用と並べて比べてみた (importance: 0, dev: 50) (いいね相当スコア: 0)
Python で安全ガード更新前後の判定差を四条件比較する (importance: 0, dev: 75) (いいね相当スコア: 0)
【書評】動かして学ぶAI・機械学習の基礎 ―TensorFlowによるコンピュータビジョン、自然言語処理、時系列データの予測とデプロイ (importance: 0, dev: 20) (いいね相当スコア: 0)
Deep Residual Learning for Image Recognition (ResNet) 論文解説 (importance: 0, dev: 35) (いいね相当スコア: 0)
【性淘汰シミュレータ5・予備実験】LoRAを遺伝子にしてLLM100体を選抜したら、考え方より文字数配分が変わった (importance: 0, dev: 25) (いいね相当スコア: 0)
Gemini 4 Argonの「72時間で25兆トークン」は毎秒9,600万の計算 (importance: 0, dev: 40) (いいね相当スコア: 取得失敗)
Ornith-1.5-122Bを動かすメモリは何GB？標準版と縮めた版のサイズ (importance: 0, dev: 40) (いいね相当スコア: 取得失敗)
サキ君、面白いネタはあるかね?をサキに査読解析してもらってからの対話🤣 (importance: 0, dev: 10) (いいね相当スコア: 取得失敗)
第53回「LLM」って何? 〜流れに合う次のことばを選び続ける話〜 (importance: 0, dev: 10) (いいね相当スコア: 取得失敗)
LLMにおけるユーザーインターフェースの構造転換と認知負荷軽減に関する検証レポート (importance: 0, dev: 25) (いいね相当スコア: 取得失敗)
【夜のDS・AI論文解説】2026年10月21日現場で役立つ最新注目論文2選 (importance: 0, dev: 20) (いいね相当スコア: 取得失敗)
DGX Spark(GB10)×2のDeepSeek V4.1 Flash 1017に資料作成を丸投げして、Claude Sonnet 5.5と比較してみた (importance: 0, dev: 55) (いいね相当スコア: 取得失敗)
AIエージェントとは？ChatGPTとの違いとアプリ開発の可能性を学ぶ (importance: 0, dev: 15) (いいね相当スコア: 取得失敗)
ハーネスとガードレールって何が違うの (importance: 0, dev: 60) (いいね相当スコア: 取得失敗)
ドラえもんやアトムとAI界隈の話 (importance: 0, dev: 5) (いいね相当スコア: 取得失敗)
vivy-r-vision-3b Phase 5 インジェクション後編レポート (importance: 0, dev: 25) (いいね相当スコア: 取得失敗)
vivy-r-vision-3b Phase 2 日本語性能レポート (importance: 0, dev: 25) (いいね相当スコア: 取得失敗)
vivy-r-vision-3b 総合ベンチマークレポート（全52問） (importance: 0, dev: 25) (いいね相当スコア: 取得失敗)
自然な文章でもなぜ確認が必要？LLMの予測のしくみ｜12週間AI講座 #04 (importance: 0, dev: 10) (いいね相当スコア: 取得失敗)
【Deep Dive】AIエージェントの記憶の設計~コンテキストウィンドウの外で更新し、減衰させる~ (importance: 0, dev: 70) (いいね相当スコア: 取得失敗)
AI企業2社に相談して分かった、図表のデジタル化の難しさ (importance: 0, dev: 25) (いいね相当スコア: 取得失敗)
賢いAIをポンコツにしたら、リンゴ農家が妖精になった (importance: 0, dev: 3) (いいね相当スコア: 取得失敗)
【生成AIニュース+】『Eleven Realtime』『EdenLive』『Spark-X2.6-4B』『EAGLE-4』『MOSS-Image FP16-fixed』『MOSS-Imagine』『ComfyUI MOSS Image BlockCache T8』『ComfyUI-4A-BoundingBoxes』『Fibo 2.0 Dev』『Fibo 2.0』『ComfyUI-Fibo2』『FastH3 Omni for ComfyUI』『ComfyUI Tab Manager』他 (importance: 0, dev: 35) (いいね相当スコア: 取得失敗)
Totvs-Protheus-ADVPL-V0.1 Phase 1 基礎性能レポート (importance: 0, dev: 60) (いいね相当スコア: 取得失敗)
【世界初】デジタル庁が職員18万人にAIエージェント導入！ガバメントAI「源内」の進化がすごすぎる (importance: 0, dev: 20) (いいね相当スコア: 取得失敗)
【雑記】GPTの一時チャットにあった便利機能が消えとる… (importance: 0, dev: 10) (いいね相当スコア: 取得失敗)
AIで作った動画がゴミになるかどうかは、作り手次第です (importance: 0, dev: 5) (いいね相当スコア: 取得失敗)
MOSS-ImageテンプレートWF「Prompt Enhancer」の検閲解除 (importance: 0, dev: 20) (いいね相当スコア: 取得失敗)
同じNemotron 30Bでも量子化形式だけで日本語作文が68点と95点に開く：選定前の検証手順 (importance: 0, dev: 50) (いいね相当スコア: 取得失敗)
LLMを育てるのではなく、人間が育つ (importance: 0, dev: 20) (いいね相当スコア: 取得失敗)
「Cloudflare OS」のマネージドサービス提供へ。GitHubとの接続やGmailとの連携強化など (importance: 45, dev: 55)
没入型体験施設の写真から人を消す──サマーインターンで挑んだワンダリア横浜の技術検証 (importance: 15, dev: 45)