TILens Daily Edition 2026-09-30 Filters: topic=ai-research, github=hidden Stats: 765 articles, 7 sources, 756 research papers Top topics: AI, AI Research, Information Retrieval ## News - True Positive Weekly #180 Source: true-positive-weekly Topic: AI (+AI Research) URL: https://aiweekly.substack.com/p/true-positive-weekly-180 - 😺 Hume AI: Voice Has a Listening Problem Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/newsletter/hume-ai-voice-has-a-listening-problem/ - My First Day With OpenAI Dots: Easy Setup and a Very Likeable Assistant Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/explainer-articles/openai-dots-hands-on-review-herman/ - OpenAI DevDay 2026: ChatGPT is becoming an AI operating system Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/news/openai-devday-2026-chatgpt-is-becoming-an-ai-operating-system/ - Zuckerberg’s Next AI Bet: Teaching Muse When to Keep Quiet Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/news/zuckerbergs-next-ai-bet-teaching-muse-when-to-keep-quiet/ - DeepSeek Opened the Code. Can Huawei Deliver the Compute? Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/news/deepseek-opened-the-code-can-huawei-deliver-the-compute/ - 😺 OpenAI launched Dots + 20 more tools Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/newsletter/openai-launched-dots-20-more-tools/ - Microsoft’s Quine Is Bigger Than a Biology Model. It’s a Bet on the Whole Research Loop Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/news/microsoft-quine-bigger-than-biology-model-research-loop/ - What Happens When Every Employer Uses the Same Hiring AI? Source: the-neuron-ai Topic: AI (+AI Research) URL: https://www.theneuron.ai/news/what-happens-employers-same-hiring-ai/ ## Research - How Many Stories Can Your Data Tell? Source: towards-data-science Topic: AI Research (+AI, Python) URL: https://towardsdatascience.com/how-many-stories-can-your-data-tell/ - Insight Is Still the Currency of Data Science Source: towards-data-science Topic: AI Research (+AI, Python) URL: https://towardsdatascience.com/insight-is-still-the-currency-of-data-science/ - How to Solve Issues When You Nest Measures While Overwriting the Same Filter Source: towards-data-science Topic: AI Research (+AI, Python) URL: https://towardsdatascience.com/how-to-solve-issues-when-you-nest-measures-while-overwriting-the-same-filter/ - Towards Spec-Driven Test Automation: Part 2 Source: towards-data-science Topic: AI Research (+AI, Python) URL: https://towardsdatascience.com/towards-spec-driven-test-automation-part-2/ - SCOPE: Observation-Conditioned Full-Target Prediction for Sparse PDE Inference Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36527 - Safe-by-Design Learning via Energy-based Neural Networks Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36942 - PHASE: A Physiology-Guided Hierarchical Foundation Model for Intracranial EEG Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36087 - SelfSearch: Reward-Free Search for Self-Improving Agents Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37968 - Visual Branch is What You Need for CLIP-based Class-Incremental Learning Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37888 - Meta-learning accelerates detector design optimization Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35827 - Identifiability Guarantees for Drivers and Dynamics of Delayed Physical Systems Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37944 - Sage: Formalization with Semantic Correction Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35790 - Formalizing the Sampling Design Space of Diffusion-Based Generative Models via Adaptive Solvers and Wasserstein-Bounded Timesteps Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.12624 - Human-inspired, Task-Dimension-Guided Exploration for Efficient Learning in High Dimensions Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36672 - PDE-OBS: Controlled Evaluation Across Observation Patterns Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36521 - Multi-Depth Temporal Fusion for Feedforward, Locally Trained Spiking Neural Networks Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37047 - QuantMLA: Function-Aligned Dual-Path Quantization for Low-Bit MLA KV Caching Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36760 - JudgeCast: Time Series Forecasting with Experience-Informed Covariate Judgements Source: arxiv-cs-lg Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36966 - Agentic Graph Retrieval-Augmented Generation for Auditable Commercial Registry Analysis Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2605.18770 - SIREN (Luring LLMs onto the Rocks): PAIR-Driven Preference Manipulation in Web-RAG Recommenders Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2607.21951 - Relevance Is Not Sufficient Evidence: Detecting Evidence Gaps Before Generation in RAG Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37469 - Structured Interaction, Visual Localization, and Robust Execution for Complex Web Tasks: A Technical Report on the WebRetriever Challenge Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35904 - Towards Semi-Automatically Comparing Keyword-Based and Semantic Search Accuracy Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37749 - Beyond the Timeline: Augmenting Long-Video Memory with Grounded Entity Biographies Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.38155 - Mnemon: Raw Records, Fast Judgments, Slow Thoughts Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36059 - AX is the New AEO Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.34951 - Optimizing VLP-aligned Multimodal Intent Representation with Correct Visual Instantiation for Zero-Shot Composed Image Retrieval Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36946 - Does the Unsafe Gradient Survive a Conversation? On the Fragility of Gradient-Based Jailbreak Detection in Multi-Turn Dialogue Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36849 - ThuRunel: Dynamic Decoupling for Structured Advisory Dialogue Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36340 - Do Evidence-Reading Diagnostics Improve Interface Selection in Small LLM Recommenders? Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37472 - Evidence-Guided Schema Normalization for Temporal Tabular Reasoning Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2512.00329 - Effective Dense Retrieval using Only In-Context Examples Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.38099 - Generated Query Expansion Still Helps Strong Sparse Retrieval: A Controlled Study with SPLADE-v3 Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37911 - Financial Evidence Crowding: Diagnosing and Mitigating Constraint-Induced Displacement in Retrieval-Augmented Generation Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35782 - ReMem: Rethinking Perception and Memory in Long-Context Recommendation Agents Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37311 - UltRAG: a Universal Simple Scalable Recipe for Knowledge Graph RAG Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2603.28773 - Follow the Entities: A Corpus Map for Agentic Search Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37226 - HELIX: Purified and Unified - Rethinking Feature Interaction and Sequence Modeling for Large-Scale Recommendation Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37183 - Better Nearest Neighbor Graph Indices via (Efficient) LLM-Guided Pruning Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36359 - ARCagent: An Adaptive Retrieval Calibration Agent for Clinical Question Answering Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36392 - TSG Suggester: Tree-Structured Knowledge-Graph Retrieval for Troubleshooting Guide Recommendation in Cloud Incident Management Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35780 - Safer Content or Firmer Refusals? A Hybrid Perturbation Defense for Alignment under Harmful Fine-tuning Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36862 - BadRAG: Identifying Vulnerabilities in Retrieval Augmented Generation of Large Language Models Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2406.00083 - GRP v0.1 Technical Report Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36688 - IROH: Insightful Ranking Of Humor using Multi-Stage Hybrid Retrieval with Rationale-Distilled LLM Judges for JOKER 2026 Track Task 1 English Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.15618 - Socrates-RAG: Premise-Directed Inquiry against Coordinated Evidence Poisoning Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35773 - MERGE: Multi-LLM Ensemble for Retrieval via Generative Enrichment Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37574 - BITEM at the NTCIR-19 R2C2 Task: Predicting Confidence from Agentic RAG Pipeline Signals Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37993 - RecKG: Knowledge Graph for Recommender Systems Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2501.03598 - Post-Generation Verification Dominates Retrieval Optimization: A 2^4 Factorial Ablation of RAG Pipeline Features Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35774 - Auditable Long-Term Memory: A Deterministic Retrieval Chain Measured at 479/475 of 500 on LongMemEval-S Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.38021 - RAISE: Diagnosing Acquisition Collapse in Costly LLM Signals Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2608.10441 - Soft Curriculum Learning for Optimizing Fresh and Generalized Recommendations Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.35783 - Unlocking Spatial Grounding in Large Audio-Visual Retrieval models Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2607.24786 - Backdoor in the Loop: Compromising Agentic Search via Malicious Retrievers Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.37468 - GeoOutageBench: Benchmarking Ambiguity-aware, Ontology-grounded Geospatiotemporal KGQA for Multimodal Power Outage and Resilience Analysis Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36082 - Retrieval Sensitivity to Identity Signals in Queries Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2609.36534 - OneLatent: Latent Reasoning for Efficient Foundation Recommendation Models Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2607.26621 - The Hitchhiker's Guide to Agentic AI: From Foundations to Systems Source: arxiv-cs-ir Topic: Information Retrieval (+AI, AI Research) URL: https://arxiv.org/abs/2606.24937 - Can Agents Design Libraries for Agents? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36730 - MGSM-Pro: A Simple Strategy for Robust Multilingual Mathematical Reasoning Evaluation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2601.21225 - VectraYX-Vision-1B: A Sub-2B Spanish/LATAM Cybersecurity Vision-Language Model, and What Limits Its Visual Grounding Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.08477 - Replay the Curvature: Accurate and Scalable NVFP4 Quantization for Large Language Model Inference Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36654 - Traverse: Learning When to Remember, Reset, and Redirect for Long-Horizon Web Search Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37082 - VStress: Correlation-Aware Auditing and Adaptive Budget Allocation for Repeated Verifiers Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36958 - Adapting Context Compression for Long-Horizon Agents with Counterfactual Continuations Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36526 - Resolving the Missing Financial Data Crisis: A Generative AI Pipeline for SEC 10-K Extraction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35864 - Evaluating Alignment of Behavioral Dispositions in LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.11328 - Eternal Sunshine of the Spotless Mind: Systematically Erasing LLM's Memories Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36414 - How Order-Sensitive Are LLMs? OrderProbe for Deterministic Structural Reconstruction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2601.08626 - VAA-CSEC: Vote-guided Advantage Allocation for Chinese Semantic Error Correction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36804 - On-Policy Visual Evidence Distillation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36838 - What Does Post-Training Change in Multilingual Reasoning? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37104 - Knowing Is Not Choosing: What Explicit Verification Adds Beyond Generative Preference Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.33142 - Does Anthropomorphic Language Impact Public Perceptions of AI? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.29121 - KUPAS MASTER: Distilling the Tacit Expertise of Master Practitioners into Agent-Ready Experience Corpora Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37673 - On Calibration of Large Language Models: From Response To Capability Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.13540 - MONOVAB : An Annotated Corpus for Bangla Multi-label Emotion Detection Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2309.15670 - LongHarness Bench: Stress-Testing Language Model Harnesses for Long-Context Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38137 - DeepRewind: Predicting and Repairing Premature Commitments in Deep Research Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36344 - EmoRES-TTS: Residual-Enhanced Vector Steering for Emotional Speech Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38157 - Vision Is Not Overhead: One-Pass Block Drafting for Lossless Speculative Decoding in Vision-Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.00355 - After the Fix: Transfer of Corrected Agent Experience Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.34603 - ORCA-bench: How Ready Are Language Model Agents for Oncall? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2607.28545 - MoRE: Scaling mixture of experts with hardware-aware low-rank routing Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36301 - Corpus-Guided Dual-Path Propagation for Graph Retrieval-Augmented Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37661 - FastGuide: Accelerating Reward Guidance for Diffusion Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36202 - Lost in Conversation or Lost in Translation? Diagnosing Multi-Turn Degradation in RAG Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36700 - Reasoning with Sampling: Cutting at Decision Points Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.30327 - RAWR: Reward Assignment Without Rollouts in Verifiable Domains Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.17815 - Agent as Policy for Robotic Manipulation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.12541 - Pruning for Efficiency, Paying in Fairness: Demographic Disparities in Pruned Speech-LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38106 - Beyond Semantics: How Temporal Biases Shape Retrieval in Transformer and State-Space Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2510.22752 - Uni-LaDiR: Latent Diffusion Unifies Multimodal Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.19878 - CypherTurn: A Multi-Turn Benchmark for Conversational Text-to-Cypher Evaluation and the Autonomy Divergence Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36987 - Video2Skill: From Streaming Experience to Reusable Embodied Skills Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36691 - LatCom: Cross-Agent Latent Compression for Efficient Multi-Agent Collaboration Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37017 - Selecting The Most Informative Tokens in Natural Language Autoencoders Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37040 - Causal and Interpretable Structures in LLM Compositional Tasks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35970 - When Updating Stops Being Learning: Rethinking LLM Self-Evolution via learnable information gain Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36535 - Learning from Teacher Continuations at Student States Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36246 - Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38143 - Dating the Model: Hidden Dates in System Prompts Affect LLM Evaluation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36931 - Bridging Semantic Gaps in RAG through Generated Context Knowledge Fusion Source: arxiv-cs-cl Topic: AI Research (+AI, Information Retrieval) URL: https://arxiv.org/abs/2609.37171 - JPO: Juris Policy Optimization for Structured Legal Reasoning in Criminal Judgment Prediction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.29616 - VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37105 - JEV vs. LLMs as Rubric Judges: Cheaper, Faster, and Wrong in the Same Places Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.29769 - A Dominant Supplier Slows Recursive Drift More Than It Steers It Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.11146 - Chinese-Jev: Bringing System One Model to Chinese-Language Tasks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36965 - How Local Mixing Encodes Relative Position in Global NoPE Attention Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38109 - Rethinking Multimodal Fake News Detection in the Generative AI Era Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36850 - Billiger.de Products: A Bilingual Entity Matching Benchmark Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37713 - Learning to Retrieve Missing Evidence for Long-Term Memory QA Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37443 - Developing an OCR model for Extracting Information from Invoices with Korean Language Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35796 - Tracing the Evolution of Oracle Bone Characters Across Three Millennia Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35674 - MemFold: Learning Compact Soft Memory for Long-Context Personalization via On-Policy Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36435 - TRACE: Deployable Tree-Relational Structure Enhancement for Oncology LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35810 - Does a prosody-trained representation help beyond trainable fusion? A parameter-matched study with frozen HuBERT Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36754 - Reproducing, Analyzing, and Detecting Reward Hacking in Rubric-Based Reinforcement Learning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.04923 - Your Benchmark Is Not Saturated: Reviving Multiple-Choice Evaluation with Answer Pooling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37494 - Constructing Challenging Browser-Use Tasks by Controlled Environment Interventions Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35814 - SEA-CLIP-Tiny: Efficient Multilingual Text-Vision Embedding for Southeast Asian Languages Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.30739 - Toward a Culturally Adapted Chinese Language Agent: A Wizard-of-Oz Study of Nonverbal Behavior in Chinese-German Intercultural Interaction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35150 - Authority Bias in Language Models: Source Deference and User Agreement Are Not Interchangeable Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37616 - When the Wrong Key Wins: Understanding and Detecting Hallucinations in LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.15106 - Repetition, Not Length: Isolating the Counting Failure in Neural Text-to-Speech Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36974 - CultureConverse: A Multilingual Multi-turn Simulation Harness for Culturally Grounded Assistance in East and Southeast Asia Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.28405 - Revisiting the Capacity Gap in Chain-of-Thought Distillation from a Practical Perspective Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2604.08880 - Evaluating Bounded Autonomy in Regulated Agentic AI: A Diagnostic Harness with Constitutional Rewards, Escalation Labels, and Runtime Governance Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37501 - Cool the Sampler, Not the Learner: Sampling Temperature Moves the Staleness Cliff of Importance-Corrected GRPO Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36953 - CredWise: A Controlled Agentic Decision-Intelligence Framework for Explainable and Auditable Credit-Risk Assessment Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37223 - Which papyrus HTR is good enough? Character-error-rate tolerance of four papyrological tasks on Greek texts Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37755 - A Polyphonic Conception of AI Understanding Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36079 - Verifier-Induced Support Reshaping in On-Policy Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.00220 - Beyond the Context Window: An Adaptive Entropy-Based Routing Framework for Hybrid Retrieval and Long-Context Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35831 - LLM unbranding: Erasing Commercial Identity while Preserving Generic Utility Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37127 - When Trees Are Not Enough: Learning Mixed-Topology Feature Graphs with Adaptive Graph Sparse Autoencoders Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36294 - Evaluating Cross-lingual Knowledge Consistency in Code-Mixed vis-a-vis Indian Languages using IndicKLAR Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.29637 - The Surge of Anti-Semitism in German Social Media following the October 7 Attacks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36290 - Calibrated to Whom? Persona and Language Effects on Cultural Values in JEV Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36399 - MAPLE: Medical Aspect-Based Summarization with Phrase-Level Evidence Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2601.03418 - Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.12149 - Overcoming Scaling Limits in On-Policy Self-Distillation for LLM Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37915 - OTROPE: Optimal Transport-based Robust Off-policy Evaluation for Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36264 - Memory Consolidation Flattens the Temporal Shape of User Facts Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36457 - Correct, Don't Delete: Mitigating Emergent Misalignment with Corrective Supervision Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37624 - TagPR: Tag-Guided Process Supervision for Personalization Reasoning in Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2509.23140 - G\"odel Forest: Balancing Search Depth and Breadth for Data-Centric Recursive Self-Improvement Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36675 - Can Vision-Language Models Stay Helpful When Facing Implicit Risks? Intent-Privilege OPSD for Efficient Safety-Helpfulness Alignment Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37837 - Teach Yourself Where to Look: On-Policy Attention Self-Distillation for Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.33200 - FOCUS: Training-Free Decision-Preserving Context Compression for LLM Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37590 - Zero-shot Dependency Parsing with Unsupervised Cross-Lingual Bootstrapping Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37883 - How Many Labels Does a Language Need? Annotation Budgets and Cross-Lingual Pooling for African-Language Text Classification Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37882 - Targeting Pivotal Decisions for Credit Assignment in Agentic Reinforcement Learning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36178 - It's All Training: A Fully Synthetic Single-Stage Recipe for LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37891 - Asymptotic Universal Alignment: A New Alignment Framework via Test-Time Scaling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2601.08777 - Evaluating the Effects of Prompt Perturbation on Bias and Hallucination in Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35804 - On Trajectory-Aware Training for Masked Diffusion Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37974 - Port-Hamiltonian Latent Deliberation: Mitigating the Deliberation Drift Cliff in Test-Time Compute Scaling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37351 - RLTL;DR: Self-improvement by Internalizing Self-generated Feedback Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37633 - The Rashomon Wikipedia: A Data-Perspectivist Analysis of Divergent Historical Narratives Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37498 - Learning Beyond What You Sample: Off-Policy-Aware Cross-Model Trajectory Exchange for RLVR Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37868 - Compiling Learning Problems into Adaptation Programs for Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37371 - E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37533 - CompOrca: Corpus-Scale Compliance Labelling of Instruction-Tuning Data Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37807 - Hidden Reasoning Must Leak, but Need Not Be Readable: Fundamental Opportunities and Limits for Chain-of-Thought Monitoring Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37312 - PRISM: A Geometric Risk Bound for Decomposing Drift into Scale, Shape, and Head Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.11608 - Is Agent Code Less Maintainable Than Human Code? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.21804 - The Canonical Order Problem: When Large Language Models Are Unreliable Knowledge Bases for Multi-Valued Relations Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36209 - CruxBench: A Benchmark of Information Discovery Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35879 - Less Uniform Discrete Diffusion is More Powerful and Scalable Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35817 - Learning What to Remember: Long-horizon Counterfactual Memory Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37930 - MultiTalk: Scaling Full-Duplex Speech Models to Long, Multi-Party, Bilingual Conversation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36903 - Alignment Forecasting: Predicting Misalignment From Training Data Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35805 - CineSubBench: Evaluating LLMs on Long-Form Narrative and Cultural Understanding from Multilingual Movie Subtitles Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36218 - The Illusion of Replacement: Rethinking Specialized Machine Learning Models in the Foundation Model Era Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.28980 - SemOPT: Fixing Semantic Errors in LLM-based Optimization Modeling via Reward-Guided Search Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37361 - MARCO: Multi-Round Agentic Reinforcement for Conditional Molecular Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36683 - Generating Edit-Inducing Questions for AI Research Manuscripts Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36617 - Risk-Controlled Selective LLM Answering by Pricing Label-Free Checks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37493 - Distilling What Matters: Confidence-Aware Selective Distillation for Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36734 - Multimodal Detection of Higher-Order Behavioral Constructs: Self-Compassion in Structured Reflective Interaction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37148 - OpenART: Scaling Agent Red Teaming via Open-Ended Environment Evolution Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.00677 - Lost in Translation: Measuring the Effect of Non-Native English on End User Performance of Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36214 - Large-scale factor analysis shows machine intelligence is only partially interpretable Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36515 - Rational Clarification by Assistive Agents via Value-of-Information Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37588 - $S^3$: Spectral Null-Space Swap Makes Reasoning Models Efficient Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37976 - SEED: Self-Speculative Decoding via Implicit Encoder-Decoder Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36590 - Can a Cacheable Decision Model Follow Rules? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37832 - Cross-Linguistic Effects in Bilingual Phoneme BabyLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37121 - Diversifying RLVR Rollouts via First-Token Exploration Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.28295 - BaLEEN: Biasing with Latent Encoded Entities for Context-Aware ASR Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36913 - Evaluating and Benchmarking the System One Model Jev Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37647 - In-Context Learning Amplifies a Latent Symbolic Circuit Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36265 - CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36820 - RAEGNet: Relation-Aware Evidence Graph Network for Harm-Aware Multimodal Fake News Detection Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36902 - Quit While You're Ahead: Quit for Efficient Candidate Generation in Machine Translation Reranking Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.00588 - TTMark: Pairwise Distortion-Free Watermarking Beyond Single-Token Entropy Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36372 - Reliable Parallel Decoding in Masked Diffusion Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36452 - DraftTrace: A Multi-View Analytics Environment for AI-Integrated Writing Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36544 - Orthogonal Yet Coupled: Decoupling Geometric Components for Model Merging Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37564 - Unlocking the Critic: Reward-Free Policy Optimization for LLM Post-Training Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37119 - Learning from Think-Mode Advantage via On-Policy Distillation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37044 - EngiWorld: What Can Frontier Agents Deliver in Professional Engineering Environments? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37686 - Larry Caused the Car to Stop, But the Model Didn't Notice: Transformer Blindness to the M-Heuristic Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37497 - RAZOR: Pruning Replaceable Experts in LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.30465 - Do Proactive Agents Need an LLM to Decide When to Act? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.30152 - Dynamic Optimizations of LLM Ensembles with Two-Stage Reinforcement Learning Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2502.04492 - Fisher-IRG: Fisher-Induced Local Invariant Representation Geometry across Language and Vision Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36458 - HPRO: Hierarchical Progressive Reward Optimization via Preference Extraction for Emotional Text-to-Speech Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.28249 - ProgressCompass: Embodied Progress Reward Models Are Lost Without the Right Context Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36684 - Bootstrapping Audiovisual Speech Recognition in Zero-AV-Resource Scenarios Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.08249 - SiDiaC-v.2.0: Sinhala Diachronic Corpus Version 2.0 Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.10861 - One Threshold Does Not Fit All Languages: Language-Conditional Deferral for Reliable and Efficient Low-Resource Text Classification Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37861 - From Answers to Policies: Efficient In-Context Learning System through Emulating Expert Investigation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.16831 - CoEM: Empowering Long-Context Reasoning with Commit-on-Evidence Memory Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36935 - Training LLMs to Verbalize Evaluation Awareness Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36316 - Similar Choices, Different Attention: Cross-Modal Associations in Humans and Vision-Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36475 - Cognitive Expert Language Models Better Align with the Corresponding Brain Systems Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36239 - From Lexical Baselines to Agentic Retrieval-Augmented Generation: Structured Skill and Responsibility-Level Extraction with the SFIA Framework Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35806 - VLM Fine-Tuning for End-to-End Combinatorial Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37175 - Reliable but Design-Sensitive: Instrument Uncertainty in LLM Annotation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35824 - The Detectability Gap: Hidden Heterogeneity in Hallucination Detection Across Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35860 - It's Not What the Image Shows: Irrelevant Context Destabilises VLM Judges Without Informing Them Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37863 - One Readout, Many Repairs: Diffusion-Guided Hierarchical Search for Tool-Agent Repair Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.34879 - Look What You Made Us Cluster: Hate Narrative Extraction from Reddit Discourse Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37408 - GRAVITY: Architecture-Agnostic Structured Anchoring for Long-Horizon Conversational Memory Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.01688 - Thinking in Depth, Speaking Directly: Recurrent Latent Reasoning for Paralinguistically Grounded Spoken Dialogue Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37818 - From Neurons to Conversation: Speech Brain-Computer Interfaces Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36736 - Instability Floors: Separating Bias from Noise in Fairness Audits of Clinical LLM Agents with FairMedAgent Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.03221 - AdaTutoRank: Learning to Rerank Document Sets via Adaptive Tutoring Optimization for RAG and Deep Research Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.32472 - SRJudge: Empowering Large Language Models with Selective Reasoning for Fine-Grained Knowledge Concept Tagging Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36982 - The Copy Ceiling: An Input-Exposure Control for Ontology-Grounded Generation over Curated Corpora Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.24885 - FD-VAD: Semantic Endpoint Detection for Streaming Full-Duplex Speech Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35791 - Grounded Revision vs. Prior Injection: Probing Retrieval-Augmented Patent Claim Amendment Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36550 - Almost Human, Except When It Matters: VoxParity and the Decisions a Voice Should Change Source: arxiv-cs-cl Topic: AI Research (+AI, JavaScript, React) URL: https://arxiv.org/abs/2609.35922 - Tracing mechanisms of sycophantic agreement in language models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35822 - Devils in Question Relay: Source-Conditioned Relay Steering to Mitigate Hallucinations in Audio-visual Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37568 - Local Predictability and Collective Fidelity in LLM-Agent Societies Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35813 - Long-Term Memory-Guided Enhancement for Target Perception in Audio-Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36577 - PowerStep: Memory-Efficient Adaptive Optimization via $\ell_p$-Norm Steepest Descent Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.10335 - Fractional State Space Transition for Long Sequence Modeling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36314 - When Does Correction Become Repair? Mechanistic Auditing of Internal Interventions in Tool-Using LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36138 - Lookahead-R: Budget-Aware Tool Retrieval via Execution-Centric Planning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35811 - From Pixels to Pairs: A Comprehensive Benchmark of LLM-Driven Key-Value Extraction in Noisy Document Settings Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.17538 - Screening Is Enough Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2604.01178 - LLMs learn different forms of metacognition when trained to predict their own accuracy Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.33886 - Decision-Sufficient State Representations: Measuring and Reducing Write-Time Regret Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.32805 - Trajectory Soup: Pushing the Compute-Scaling Frontier of LLM Mid-training via Diverse Trajectories Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37169 - Last Translation Benchmark Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.04173 - The Geometry of Inference in Transformer Residual Streams Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37824 - Invariant Atoms: Sparse Coordinates of Local Semantic Geometry in Language Model Representations Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36451 - The Price of Token Boundaries: Compression Certificates and Prediction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35869 - ER-JEPA: Experience Replay Improves Joint-Embedding Predictive Learning in Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36952 - ATTUNER: Recomputation-Free KV Cache Reuse via Query-Side Adaptation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36722 - Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.22005 - A Character-Level Neural Approach to Sinhala Sandhi Splitting Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36131 - Predicting Team Performance from Communications in Simulated Search-and-Rescue Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2503.03791 - Dr. OPD: Learning What to Follow for Optimal On-Policy Distillation of Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38025 - PACT: Pairwise-Anchored Calibrated Tuning for Single-Token Typed Decisions Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35865 - Multimodal LLMs Outperform Pathology Foundation Models in Cross-Domain Histological Similarity Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.32876 - Reader Proficiency Shapes Layer-wise Surprisal Profiles Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37688 - Population Fidelity: Evaluating Population Representativeness in LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36253 - From Dissonance to Orchestration: Teacher Intervention in On-Policy Distillation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37510 - VehicleArena: A Realistic Urban Environment for Multi-Agent Driving Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35916 - Seeing What Should Be Heard: Diagnosing and Repairing Cross-Modal Shortcuts in Omni-Modal LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36798 - Are We Measuring Strategy or Phrasing? The Gap Between Surface- and Approach-Level Diversity in LLM Math Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.29985 - AdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self-Distillation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38142 - Backpropagated Output Momentum: Relocating Optimizer History from Parameters to Task Space Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36738 - FinRT: Distilling Adaptive Red-Teaming Strategies into Reusable Adversarial Generators in Consumer Finance Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36474 - Correct Answers, Invalid Traces: What Verifiable Grade-School Math Reveals About Chain-of-Thought Traces Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38107 - C-Instrument: Automating RL Data Generation and Hillclimbing with a Constitution-Grid Instrument Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.00180 - Large Language Models Exhibit Human-Like Bayesian Hypocrisy Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35779 - LLMs are not stochastic parrots: Evidence for meaning-mediated abstraction from conlang-like tasks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.34187 - Gender bias across LLMs is common and highly heterogenous Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38036 - AMU:Admission and Memory Update for Personalized Conversations---Structured Memory with SLM Guided Control Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36976 - $\tau$-Multilingual: Benchmarking Voice Agents Across Languages Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35820 - Predictive Geometry of Hidden Trajectories in Transformers Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37717 - Remember Your Trace: Memory-Guided Long-Horizon Agentic Framework for Consistent and Hierarchical Repository-Level Code Documentation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.14563 - Can Multimodal Large Language Models Generate and Detect Multimodal Social Media Fake News? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35809 - Environment Steering: Using Data Flow Control to Improve Agent Utility and Safety Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35807 - Co-Linguistics: AI-augmented Theory Construction in Linguistics Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37635 - LAURA: Knowledge Distillation for Interpretable Ambiguous Clause Identification in Legal Contracts Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36707 - OpenTumorBoard: A Real-World Benchmark of Multidisciplinary Tumor Board Discussion Trajectories Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.32810 - Who Warmed the Archives? LLMs Overestimate Historical Warmth Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37499 - Automated Evaluation of Multi-Turn Dialogues in In-Car Conversational Assistants Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35812 - Retrieval Capacity of Self-Attention Under Competition Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37879 - Break Step: Recursive Training Resonates with Replayed Sampling Noise Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.11149 - Sycophancy Is Not One Thing: Causal Separation of Sycophantic Behaviors in LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2509.21305 - Relative Kinetic Utility: Calibrating Cross-Layer Credit for Global Structured LLM Pruning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.09008 - Sieve and Sage: Efficient Distraction Filtering for Reliable RALM Abstention Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35794 - Decodable but Misrouted: Sparse Features Uncover a Readout Gap in Vision-Language Models for Harmful Meme Detection Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.18860 - Evaluating Test-Time Scaling of General LLM Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.18998 - Isolated Sign Language Recognition for Icelandic Sign Language: Experiments in a Low-resource Setting Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.25862 - Quantization Error Is Spectrally Flat: A Single Random Probe Is a Calibrated, Data-Free Sensitivity Estimator, with Application to Budget-Targeted Mixed-Precision Quantization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.33923 - A theoretical model of dynamical grammatical gender shifting based on set-valued set function Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.03510 - PADM\'E: Preference Alignment Data Synthesis for Meta-Evaluation of LM Agent Evaluators Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36086 - Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38177 - Group-Marginalized Self-Rewarding RL Drives Zero-Label Self-Evolving Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36750 - Reconstructing the Vocal Tract with Differentiable Acoustic Simulation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36737 - Concept Direction Reliability Across Languages with Different Tokenizer Fertility Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36194 - EpiKV: Epiphany-Aware KV Cache Eviction Without the Attention Matrix Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.26472 - IESR:Efficient MCTS-Based Modular Reasoning for Text-to-SQL with Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.05385 - Can Language Models Learn to Forecast Stock Prices Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36914 - When Should LLMs Trust Their Own Revisions? A Risk-Aware Study of Intrinsic Self-Correction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35832 - LoLBench: Evaluating Coding Agents with Long-Horizon Proposals on Large Software Systems Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37143 - KSAFE-MM: A Multimodal Safety Benchmark via Localized Contextualization for Korean Cultural Risks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.28013 - Pretraining Latent Information Feedback Transformers with Teacher Supervision Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38149 - Entity tracking emerges in sub-billion parameter language models and exceeds human performance in naturalistic narratives Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.18083 - MoEGen: Mixture-of-Experts for Instance-Adaptive LoRA Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.03275 - Asking for What Was Never Requested: Horizontal and Vertical Proactivity in Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37236 - Principled Thoughts for Latent Recursive LLM Systems Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36159 - AdvancedMathBench: A Benchmark Suite for Advanced Mathematical Proof Generation and Verification Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2607.11849 - Can We Still Trust Disaster Social Sensing? Empirical Evidence on Detecting AI-Generated Social Media Posts Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35821 - RunyaNER: Auxiliary Language Selection for Runyankore NER Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37543 - SalamahBench: Dialect and Category Level Safety Evaluation of Arabic Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.04410 - FORUM: Frozen Outputs Reconciled Using Model Agreement for Visual Grounding Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37488 - When Successful Memories Mislead Embodied Agents:Memory Adaption For Task-Conditioned Execution Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35808 - Question-Specific Knowledge Graphs for Efficient Visual Reasoning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35942 - Trustworthiness Costs of Domain Adaptation in Small Language Models:A Cross-Architecture Empirical Study Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2608.00042 - Benchmarking Automatic Speech Recognition Tools for Iberian Languages Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36920 - Hyperspherical Semantic Trajectory Analysis: Mapping Technological Diffusion across Academic Preprints, Patent Signals, and Compute Scaling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35845 - HeurEvo: Agentic Evolution of Hybrid Solver-Augmented Heuristics for Time-Critical Mathematical Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36303 - How to Run Statistics over LLM Judges and Trust the Results: Calibrated Inference for Small-Sample AI Evaluation with evalstats Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35815 - Block Sparse Flash Attention Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2512.07011 - Context Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37725 - Selecting What Matters: Semantic Compression-Guided Selective Pooling for Long-Context Embeddings Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37782 - PrimeSeeker: Capability-Oriented Supervision for Deep Search Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35816 - Vision Wormhole: Latent-Space Communication in Heterogeneous Multi-Agent Systems Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2602.15382 - Margins, Not Windows: Training-Free Per-Step Lossy Speculative Decoding Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.02897 - Decomposing and Measuring Evaluation Awareness Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2605.23055 - Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36585 - A Proposed Rubric for Evaluating Expressed Clinical Reasoning in Large Language Model Responses Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37788 - Harness Evolution as Learning: Approximation, Generalization, and Optimization Limits of Self-Improving Personal Agents Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36892 - Quantifying Behavioral Tails in Black-Box Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.33638 - Momentum-Coupled Rubric Adaptation for Detailed Image Captioning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36893 - Coherence-Aware Distributional Evaluation of Open-Ended Text Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.34240 - Time-Anchored Diffusion Language Models: Latent-Space Caching for Fast Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37924 - Geometric Representations of African Languages: A Regional Semantic Hub and Cultural Steering Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36205 - What Makes Recurrence Effective in Looped Language Models? Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36636 - TriageRA-CCF: Source-Side Clinical Confidence and Coverage Signals for Adaptive Rank Budgeting in Medical LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.29375 - From Routing Signals to Selective Review: Visual regrounding in MoE VLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38111 - SCOUT: Synergizing Reasoning and Tool-Use for Computer-Use Safety Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36201 - Act First, Reason Later: Accelerating On-Policy Distillation for Multi-Turn Agents via Reference-Conditioned Inverse Dynamics Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36608 - Think Multilingual, Not Harder: A Data-Efficient Framework for Teaching Reasoning Models to Code-Switch Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2604.15490 - Hierarchical Compression of Vision-Language Model Benchmarks Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37515 - CHAIN: Calibrated LLM Forecasting via Causal-Temporal Hypergraph Inference Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36689 - Adaptive Consistency Graph for Long-Horizon Agents Source: arxiv-cs-cl Topic: AI Research (+AI, JavaScript, React) URL: https://arxiv.org/abs/2609.32754 - CORE-BREW: LLR-Based Soft Decoding for Robust Multi-Bit LLM Watermarking Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2606.24163 - Better Behavioral Prediction, More Faithful Model Ablations? Evidence from Sequential Choice Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36097 - Neurosymbolic Routing for Reliable Reasoning on Resource-Constrained Edge Devices Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35833 - The Unequal Influence of Bad Advice: Using Training Data Attribution to Modulate Emergent Misalignment Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37914 - Storage Is Not Strategy: State-Conditioned Support Control for LLM Unlearning Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37858 - SIPO: Unifying Reinforcement Learning with On-Policy Self-Distillation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36742 - Pair Difficulty Matters: Rethinking Pairwise LLM-as-a-Judge Evaluation and Consistency Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37577 - POET: Preference Optimization for Enhanced Text-to-Image Generation Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2510.12041 - Language Models Are "Insecure" Reporters Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36139 - When Models Don't Manipulate Manifolds: The Geometry of a Comparison Task Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37680 - STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38169 - Toward Robust LLM-Based Judges: Taxonomic Bias Evaluation and Debiasing Optimization Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2603.08091 - Triadic Linear Attention: Three-Dimensional Recurrent States for Long-Context Sequence Modeling Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36529 - How to Tame a Multi-Headed Hydra? Adaptive Multi-Category Safety Steering for Large Language Models Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.34514 - Regime Boundary Alignment for Evidence-Gated Question Answering Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37491 - Solving Without Stopping: On-Policy Distillation at Small Scale Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37326 - Layer-Informed Fine-Tuning via Three-Stage Functional Segmentation of LLMs Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38027 - AnthroDial: Benchmarking LLM Anthropomorphism in Autonomous Social Interaction Source: arxiv-cs-cl Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37853 - Visual sensitivity is not claim retractability: persistence-aware credit assignment for multimodal reinforcement learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36572 - Reasoning with Neural Cellular Automata Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36126 - Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38147 - Explore, Execute, Evolve: A Skill Acquisition and Reuse Loop for Embodied Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37810 - Reconstructing Implicit Scientific Knowledge: Evaluating LLM Agents through End-to-End Reproduction of Astronomy Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35900 - Mara Chain: Rethinking Failure as a Stepping Stone for AI System Auto-Evolution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35855 - Engineering Efficient Self-Play Chess: Search, Replay, and Throughput Under Limited Compute Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37447 - TAEC: Trajectory-Aware Evidence Coordination for Multi-Step Visual RAG Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37349 - SKILLLITE: Evidence-Guided Malicious Skill Auditing with Compact LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36879 - Adversarial Debiasing of Machine Learning Models for Enhanced Network Security against DDoS Attacks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36167 - Locating Answer-Correctness Signals in Frozen Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37700 - Beyond Compression: Diagnosing How Post-Training Changes Mathematical Reasoning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37066 - Salt++: Context-Aligned Post-Training for Few-Step Streaming Multimodal Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36995 - Scaling Influence Functions in LLMs through Eigenbasis-Corrected One-Bit Gradient Projection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37842 - AutoLoCo: Communication Efficient Distributed LLM Training via Adaptive Synchronization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36662 - Position: Let's Strengthen Verifiability If We Can't Enforce Reproducibility Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35854 - Credit-Guided Policy Improvement for Test-time Adaptive Vision-Language Navigation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37591 - PrecogUI: Proactive GUI Agents via Pre-cognitive Simulation and Experience Retrieval Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36923 - ChronoSRL: Temporal Geometry for Self-Supervised Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36238 - DualTrack: Synchronized speech-gesture generation via symmetric coupling of pretrained priors Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36624 - AssayRouter: Historical Utility Priors for Frozen Molecular Predictor Routing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37285 - Think Before You Restore: Risk-Aware Manchu Manuscript Restoration with Stroke-Guided Attention Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36243 - Train Ahead, Distill Back: Bootstrapping On-Policy Self-Distillation for Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37132 - V-Engram: Trigger-Indexed External Memory for Modular Text-to-Image Personalization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37198 - Similarity Is Not Validity: Defending LLM Semantic Caches Against Poisoning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35908 - FineART: Fine-grained Annotated Robotic Trajectory Dataset and Vision-Language-Action Model for Bimanual Manipulation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36416 - VLALight: A Vision-Language-Action Model for Traffic Signal Control Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36934 - Beyond Rule-Based Mutation Testing: Test-Aware Mutant Generation Using Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35841 - MMSkillRisk: Can Agents Stay Safe When Multimodal Skills Become Traps? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35912 - CADOC: Cache-Aware Dynamic Object Context for Long-Horizon Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37012 - Estimation of Room Impulse Responses from Handclaps Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35839 - SCA: Spatial Credit Assignment for Reinforcement Learning of GUI Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36939 - From Judgment Quality to Downstream Utility: Rethinking LLM-as-a-Judge for Open-Ended Tasks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37145 - Video-RSI: Recursive Self-Improvement of Video Understanding Agents via Harness Evolution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37950 - ATLAS: Aligned Transport of Latent Structure for Reliable World Model Planning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36333 - MeanFlowAdvantage: Stable Reward Fine-Tuning for Few-Step Average-Velocity Generators Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37670 - SkillCome: Group Contrast Skill Optimization with Dual Memory Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37128 - Evolving Towards Better Codes: LLM-Guided Search for High-Distance Binary Linear Codes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37056 - ToolFence: Fine-Grained Authorization for Secure Tool-Using LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37196 - NeuronEye: Query-Guided Visual Concept Activation for Vision-Language Reasoning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38098 - Explainability from Training with Applications to TCR-Epitope Prediction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36354 - EnterpriseBench: Benchmarking LLM Agents on Enterprise-Level Strategic Reasoning and Decision-Making Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37658 - Privy to the Foil: Recasting Value Estimation with a Self-Privileged Critic for RLVR Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37825 - Encoder-Sharing Hierarchical Federated Multi-Task Learning for VANETs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36157 - CF-LoRA: Decoupled Factor Aggregation and Adaptation-Aware Client Clustering for Federated LoRA Fine-Tuning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36986 - No Scale Left Behind: Multi-Scale Autoencoder with Bi-directional Attention for Time Series Anomaly Detection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38004 - Risk-Averse Online POMDP Planning via CVaR of the Immediate Cost with Performance Guarantees Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35874 - FLOORA: A Human-Aligned Domain-Specific Language Model for Architectural Design Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36064 - Illusory Truth or Mere Exposure? Model-Dependent Repetition Effects in LLM-Based Social Media Simulations Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36278 - Accessible, but Not Adopted: Increasing LLM Adoption among First-generation, Low-income (FGLI) College Students beyond Expanding Access Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36129 - SimpleEvol: An Agent-Loop Framework for LLM-Driven Automated Heuristic Design with Minimal Human Priors Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37172 - VeriWeave Govern: Evidence-Gated Deterministic Runtime Governance for Enterprise AI Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37457 - Controlled Decoding Attacks on Black-Box LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36956 - RankBuffer: Efficient Ranking-Based Rewards for Open-Ended Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36652 - REALHOP: Rethinking Multi-Hop Reasoning Evaluation via Behavioral Auditing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36984 - PolyOCR-Venus: Unified OCR Foundation Models for Text-Centric Visual Intelligence Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37712 - Do Agent Benchmarks Do What They Say? An Executable-Contract Audit of Tool-Using Agent Environments Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37315 - IronLLM: Forging Compact Edge-Native Language Models for Real-Time Embodied Intelligence Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36860 - Character Training for Risk-Averse Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38093 - Video2STL: Grounding VLM-Generated Temporal Specifications for Robot Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37519 - Beyond Interaction Capacity: Estimator Scaling with Recursive Models for CTR Prediction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37905 - PrivacySkills: How Privacy Guidance Shapes Source Selection in LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35937 - GRFBrain: Graph-Structured Rectified Flows for EEG Dynamic Modeling Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37934 - Dual-Mode Low-Rank Learner with Bridge-Prototype Ensemble for Vision-Language Class-Incremental Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36759 - ThinQuant: Scalable Rotation Learning for Weight and Activation Quantization of LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36120 - MotionInsight: Diagnosing Object Motion Deficiencies in Generated Videos Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37030 - WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36887 - Geometry-Conditioned Fixed-Scaffold Encoders for Time-Warp Robust Sequence Retrieval Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36809 - UserProxyBench: Evaluating LLM User Simulators for Agent Benchmarks and Training Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38043 - Bathtubs, Boundaries, and Sandboxes: AI Regulatory Learning under Legal Uncertainty Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2601.04094 - SAKI: Maximal-Coupling-Routed Teacher Supervision for On-Policy Distillation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36601 - SINGED: Correct Outputs Do Not Certify Safe Execution in LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35889 - Lucid Dreaming for World Models: Learning to Doubt Imagination and Decide by Trust Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37156 - V-JEPA Policy: Building Effective World-Action Models on Predictive Visual Latents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37250 - Beyond Semantic Narrowing: Robust and Efficient LLM Watermarking with Hamming Neighborhoods Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37218 - Commitment Hierarchies under Intent Revision: A Belief-Revision Account of Salvage in Tool-Use Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37453 - Semantic Projection for Continual Self-Evolution of Language Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36626 - Normative Loss Landscape Navigation: A Trajectory-Based Approach to Mitigating Forgetting in Incremental Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35926 - Beyond Keywords: Leveraging Generative LLMs and Label Aggregation to Classify Economic Policy Uncertainty in News Articles Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35856 - Multi-Site Real-World Performance of Commercial AI for Pulmonary and Incidental Pulmonary Embolism Detection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37750 - Pixel-Level Transformers in Remote Sensing: A Canopy Height Case Study Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37809 - Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt Injection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35932 - actr: aligning thoughts and responses for multilingual safety in reasoning llms Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37054 - VISTA-Bench: Benchmarking Multilingual Image Translation with Image-Specific Rubrics Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37287 - Cross-attention encoding models reveal dynamic spatiotemporal routing across human higher visual cortex Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36366 - Beyond Prompt Count: How Data Shapes Transfer in On-Policy Distillation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37377 - Human-AI Collaboration: From Paradoxes to Patterns Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36481 - Learning to Prove, Not Just to Answer: Reinforcement Learning from Formal Verification for Natural-Language Logical Reasoning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37203 - Infrared Subtraction with Artificial Intelligence Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36007 - Normalize-Then-Precondition: A Hierarchical Approach to Marginal Scale and Interaction Geometry for LLM Training Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36692 - ProCTI: Prototype-Refined Global Conditioning for Diffusion-Based Time Series Imputation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37632 - DSWM: Decomposed Spatio-Temporal World Model for Demand-Driven UAV Base Station Repositioning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36845 - Complexity-Aware Evaluation of LLM Comprehension Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37405 - State Trace Rationale As Auxiliary Task in Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36867 - Staircase Policy: Streaming Inference for World-Action Models with Large Action Chunks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36471 - Pixels to Keys: Exploring Spatial and Motion Cues in Gameplay Inverse Dynamics Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37907 - An Exact Generate - Transform Decomposition of Small-LLM Team Scaling Across Orchestration Architectures Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36104 - Diffusion Policy Improvement with Proposal-Conditioned Refinement Flows Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36812 - Guide, Then Let Go: Gap-Adaptive Teacher Scheduling for Sparse-Reward Agentic RL Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37898 - Where Does Staleness Accumulate? Pool Aware Effective Staleness Control for Asynchronous RL in LLM Post-Training Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36830 - Reward-rate Policy Gradient for Efficient Machine Learning Engineering Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36393 - NowcastDiT: Diffusion Transformers are Effective Precipitation Nowcasters Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37038 - Code4Scene: Benchmarking Coding Agents for Constructing and Editing 3D Scenes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36777 - Intrinsic Associative Memory on Riemannian Manifolds: Curvature, Capacity, and Emergent Modes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35948 - Support-Set Target Leakage in Relational Foundation Models during In-Context Learning: Impact, Detection, and Mitigation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36384 - UNBIND: UNlearning By INference-time Directional Steering for Code LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35913 - CRJudgeBench: Can AI Detect Plausible but Invalid Code Reviews? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37216 - Interpolated Policy Distillation: A Controllable Continuum Between Off-Policy and On-Policy Distillation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37170 - PreviewDiff: Multimodal Critic-Guided Search over Diffusion Latents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36199 - ROSS: Relearning from Self-Generated Rollouts through Selective Supervision Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35954 - Do-JEPA: From Masking to Intervention in Latent World Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37378 - Mutually Adversarial Self-Training with Evolving Data for Unified Multimodal Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36224 - Accelerated surrogate dynamics for dynamical, stochastic system evolution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37184 - More Programs or More Rolls? Separating Coverage from Specialization in LLM Harnesses Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35873 - MyoCodec: A Streaming Neural Codec for Electromyography Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36687 - Probability is Not Enough: Exploring and Counting Divergent Tokens for Reasoning Uncertainty Quantification in LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38070 - Teaching LLMs to Generate Challenging MILP Instances via Solver Feedback Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37356 - ARC-KV: Amortizing Anchor Search for Reconstruction-Based KV Cache Compaction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36835 - FedLAFP: Low-Rank Aggregation Meets Full-Rank Personalization in Federated Fine-Tuning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37033 - Simultaneous Neural Optimal Transport Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37424 - Task-Relevant Null-Space Residuals for Non-Injective Neural Mappings Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37272 - Where Predictive Supervision Goes Shapes What VLA Policies Learn Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36645 - BRIDGE: Bilevel Retrieval-Credit-Aware Agentic Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36505 - Generative Interactions: Weaving Multiparty Human Motion with Bilevel Latent Dynamics Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37708 - Language as the Interface: Foundation-Model Contrastive Learning Links Transcriptomes and Electrophysiology Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37024 - Neuro-Symbolic Computer Use: Learning Reusable Policies for Reliable and Efficient Execution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36927 - GeoWind2Plan: Mission-Time 3D Urban Wind Prediction for Energy-Efficient UAV Planning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36056 - Towards Breaking the Learning System Wall Using Multimodal Tutoring Transcriptions Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36502 - FigAct: Turning Scientific Figures into Active Canvases for Explanation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36190 - Quantization Enables Private Dense Retrieval against Malicious Service Providers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36376 - Paired Multimodal Scaling Laws Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36263 - Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38024 - JudgeProfile: Understanding and Steering Subjectivity in LLM Judges Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36705 - Abductive World Modeling via Causal Representation Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36985 - Dagger: Decoupling-based Model Stealing Attack against Graph Neural Networks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37972 - HandAnthro: Automated Hand Anthropometry from a Single Image Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37855 - Predictive Self-Supervised Learning Provably Identifies Stochastic Signals under Nuisance Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37789 - Learn from the Gap: Differential-Aware Advantage Pruning with Adaptive Rollout Sampling for GRPO Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36932 - Rethinking Reasoning Paths as Phase-Structured Trajectories Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36461 - From Surfaces to Volumes: Registered Geometry for Protein Representation Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36277 - The Safety Operator: Modulating the Expression of Safety Instructions via Spectral Optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36434 - WeLike2Party! In-Context Motion Transfer for Multi-Human Image Animation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36937 - Do LLM Agents Execute the Plans They Declare? From Planning-Mode Declaration to Pattern-Specific Execution Source: arxiv-cs-ai Topic: AI Research (+AI, JavaScript, React) URL: https://arxiv.org/abs/2609.38108 - Information Bottleneck-Guided Adaptive Hypergraph Transformer for Brain Disease Diagnosis Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37220 - HiTS-CL: A Continual Learning Framework for Long-Horizon Temporal Knowledge Graph Extrapolation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36559 - HiRAE: Hierarchical Representation Autoencoding with Residual Budgets Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37775 - Retrieve, Reproduce, Reveal: Dissecting Retrieval-Augmented Software Vulnerability Detection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37669 - BiFE: Search-Efficient Discovery of CPU-Only Branching Policies via LLM-based Bi-Fidelity Evolution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36735 - Automated Screw Planning for Reduced Pelvic Fractures Based on Statistical Shape Models and Deep Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36847 - Seek Before You Move: Evidence Seeking for Progress Grounding in Vision-Language Navigation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37353 - Agentic Commerce Bench: Measuring Fraud Detection for Agents That Spend Money Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35886 - Emergent Tonal Structure in Learned Chord Embeddings and Its Relation to Tonal Tension Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36460 - Authority Before Utility: Non-Compensatory Control for Persistent LLM Memory Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37474 - A neural network that maintains and retrieves memories based on context Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37791 - DecoyTrace: Toxic Decoys for Active Defense in Decentralized Federated Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36330 - Calibrating One-Round Membership Inference with Neighbors Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36331 - Neural networks for spectral optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36047 - Can AI Scientists Change Their Minds? Prior-Evidence Conflict in Synthetic Universes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36726 - You Cannot Pick a Provider From the Price List: Market-Aware Routing for Open-Weight LLM Inference Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37902 - From Dead Code and Static Requirements to Working Engines: Software Revival with Coding Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36161 - Support-Set Target Leakage in Relational Foundation Models during In-Context Learning: Model Dependence and Evaluation Reliability Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36417 - AdaST: Adaptive Coupling for Spatial-Temporal Forecasting Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36119 - From Unity Simulation to Diffusion-Based Augmentation: Quantifying Dataset Balance for Robust Object Detection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38010 - VISTA: Value-Informed Event Appraisal for Multimodal Emotion Conflict Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37324 - Solver Agent: an Agentic AI Framework for Theoretical Physics Computations Applied to F-theory Uplifts of O3-planes and S-folds Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35958 - Text2Sim: Agentic Physics-Based Simulation Generation with Distilled Expertise Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36593 - PE-EK-PINN: Physics Embedding with Evolving Kernel for Scalable Physics-Informed Neural Networks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38023 - Brain-SAD: A Brain-Inspired Safe Autonomous Driving Control Framework with Dynamic Fear-Oriented Constraint on Dual-Policy Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38016 - SafeCoEvo: Co-Evolving Safety Harnesses and Guards for LLM Agents at Test-Time Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36580 - OmniVCBench: Benchmarking Evidence-Grounded Multimodal Reasoning Towards AI Virtual Cells Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37773 - Mixture of Self-Improving Branches For Agent Harness Optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37834 - HyperZip: Efficient Data Compression through Personalized Diffusion LLMs with Hypernetworks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36357 - Semantic Map Sharing and Capability-Aware Coverage Planning for AI-Native 6G Robotic Coordination Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37666 - Governing the Edge: Automating Commercial Property and Casualty Insurance Underwriting via a Hybrid Local-Cloud Multi-Agent Framework Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37454 - Measuring trainable degrees of freedom in materials graph neural networks: a random-subspace intrinsic dimension analysis Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36084 - Digital Twin Modeling of Quantum Dynamical Systems: Dissipative Quantum Reservoir Computing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36901 - ARGOS: Reinforcement Learning-Driven Multidimensional Elasticity for Service Orchestration in the Computing Continuum Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37085 - BrainNet Studio: A Unified Toolkit for Brain Network Construction, Intelligent Analysis, and Visualization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37956 - Making Duplicate Reimbursement Unrepresentable: A Verified Ethereum E-Invoice System for Humans and AI Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37819 - Second-Moment Stochastic Approximation Methods Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36600 - Breaking the Illusion of Review Reliability under Static Evaluation: SCOPE Fuzzing for LLM-based Scientific Reviewers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37097 - Which Attention Heads are like the Human Head? Not the Ones that Compute Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37991 - One Geometry, Different Outcomes: Readout-Dependent Effects of the Modality Gap in Vision-Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36101 - Calibration-First Cross-Cohort Multimodal Temporal Learning for Transferable Asthma-Risk Forecasting Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35795 - Purlin: Separating Orchestration from the Datapath of Collectives Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36954 - Direct Experience World-Model Optimization: Learning the World Beyond Action Imitation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37398 - Constitutional adapters: Inference-time interventions for misalignment and misuse Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36657 - Self-discovering RL in the Era of Experience: Is Learning History an Asset or a Burden? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35897 - Does Local Video Understanding Transfer Across Encounters? The EgoGears Benchmark Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37938 - LLMs Learn to Evade Latent Monitors from Prior Feedback Alone Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36490 - Render Before Reading: Visual Rendering as a Prompt Injection Defense Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36121 - Right Words, Wrong Moment: A Clinician-Grounded Analysis of Distress in 19,930 Conversations between Young People and ChatGPT Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35953 - Evaluating Name-Only Directory Routing for One-Shot Code Search Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35918 - STAR-GRPO: Canonical Anchoring and Reliability-First Advantages against Representation-Dependent Reward Hacking Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36900 - Towards Mitigating Deceptive Safety Alignment in Large Reasoning Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36254 - Engineering Simplicity: Simple Mechanism Interfaces Steer LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36365 - Draft in Parallel, Condition Through Depth: Adjacent Causal Injection for Speculative Decoding Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36173 - Generalizable Lifelong Model Editing via Preference Optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36748 - OpenAI-HuggingFace: A Reproduction & Lessons for Alignment Testing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35799 - Dual-Channel Robust Group-Relative Policy Optimization via Advantage and Sequence-Weight Estimation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36944 - Mubric: Mutation Testing-Guided Rubric Generation for LLM Evaluation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37322 - MemEvo: Automatic Discovery of Streaming Video Memory Mechanisms Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36581 - More Features Are Not More Evidence: Limits of Training-Free Human Activity Recognition with Jev Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36154 - Cross-Organizational SysML Model Integration: A Survey of Challenges and AI-Supported Tasks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37000 - ContextRender: From Execution Dependencies to Agent Context Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37743 - PowerMarketJax: A JAX Benchmark Suite for Multi-Agent Reinforcement Learning in Power Markets Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37321 - Embedded Bi-Temporal Building Damage Assessment for On-Board Data Reduction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37013 - Compress to Remember: Learning Compact Memory via On-Policy Distillation for Long Video Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36364 - RLX: A Unified Multi-Backend Tensor Compiler and Distributed Runtime in Rust Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37916 - Watch-Think-Interact: Bootstrapping Long-Horizon Multi-Turn Streaming Video Reasoning with Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37035 - SkillGym: Training Skill-Use Agents with Automatic Verifiable Environment Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37539 - A Benchmark & Dataset for Detecting AI-Manipulated Visual Evidence in the Court System Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37783 - How Much Prompt Is Enough? A Blackbox Minimization of Few-Shots in LLMs Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36289 - GARDiff: Graph-Aligned Residual Diffusion for Probabilistic Multivariate Time-Series Forecasting Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37694 - Jaxolotl: A Unified High-Performance Benchmark Suite for LTL-Based Multi-Task RL Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38065 - SAGE: A Statistical Acceptance Gate for Self-Evolving Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36043 - From Retrieval to Reasoning: Agentic Mechanism Prediction from Cell Painting Profiles Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36406 - FineSID: Scalable and Efficient Semantic Identifier Learning for Generative Recommendation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36670 - What if automating AI R&D triggers an intelligence explosion? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36054 - Foundations of Proactive Agents: Principles, Technical Layers, and Proactivity-Gym Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37267 - Embodied Semantic Communication for Collective Autonomous Agents: A Tutorial on Representation, Wireless Delivery, and Closed-Loop Coordination Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35936 - Absorbed in Inertia: Activation Analysis for Computer-Use Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37176 - How Medical VLMs Underutilize Their Vision Encoders: A Dermatology Perspective Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36557 - Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM Routing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37362 - ResComEmb: Effective and Efficient Multimodal Embedding via Residual Homogeneity Compression Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37225 - Neural topology optimization of ship structures under propulsion machinery vibrations Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38089 - GLaS-JEPA: Gaussian-Regularized Speech SSL without Engineered Prediction Targets Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37798 - MetaCtrl: Your Large Language Models Can Reason Better and More Concisely with a Metacognitive Controller Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37304 - Emergent Specialization in Populations of Self-Supervised Collaborative Vision Experts Without a Shared Gate or Cross-Agent Gradients Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36770 - AI as a Compiler: Compiling Triton kernels without the Triton compiler Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36800 - Systematic Multi-Agent Vision-and-Language Navigation: Formulation, Benchmark, and Method Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35965 - Aperture: Merge-Consistent Rotary States for Compressed Tokens Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36781 - Binarization Flattens the Score Space Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35797 - DScale: Scaling Block-Diffusion Speculative Decoding with Adaptive Verification Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37532 - Channel-Dependent State Space Model for Multivariate Time Series Forecasting Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36453 - AnyAct: Universal Action for Self-Evolving Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37025 - Beam Search as Test-Time Self-Distillation via Counterfactual Contexts Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37041 - FM-ReID: Selective Competitive Token Routing for Object Re-Identification Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36560 - MLToolBench: Learning Tool-Augmented Agents for Machine Learning Development Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36679 - Periodic Weak Spots: Phase Sensitivity from Chunked KV-Cache Compression Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36322 - LatentSift: Policy-State Filtering for Token-Efficient Verification of Software Engineering Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36371 - Cross-Entropy Guided Routing in Mixture-of-Experts Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37751 - Interactive-Policy Distillation with Bidirectional Propose-and-Verify Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36546 - Bits Under ZK-LLM: Evaluating Zero-Knowledge-Friendly Quantization for Verifiable Private LLM Inference Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36437 - Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routing Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37402 - Is manual software optimization a thing of the past? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37849 - Where Should Physics Enter a Molecular Crystal Generator? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36398 - A Comprehensive View of Fairness through Distributional Stability Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37061 - Concealing LLM-Based Multi-Agent Topology via Phantom Structure Injection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37567 - SPLASH: Switching Parallel Layouts of Attention with Seamless Handoff for LLM Serving Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37626 - Codebook-Guided Cross-Modal Knowledge Distillation for Structurally Heterogeneous Features Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37243 - Adam under Generalized Smoothness with Second-Moment-Type Stochastic Gradients Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37787 - LoopICL: Looping a single transformer block to solve tabular tasks Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36108 - Learning to Harvest Without Collapse in a Regenerative Commons: A Lagrangian Framework Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36478 - Representation by Design in Generation: Cross-View Class-Token Alignment in Diffusion Transformers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36348 - Neural Structural Reasoner: A Brain-inspired Architecture for Reasoning over Structured Knowledge Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36620 - Boundary-State Control for Tool-Using Language-Model Agents: Commit-Time Consistency under State Drift Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37475 - Calibrate the Decisions That Change the Future: On-Policy Post-Training Quantization for Multimodal Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36828 - InterBias-SV: Compound Conditions in Speaker Verification Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36500 - Topological Coherence for Self-evolving Multi-agent Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37953 - Online Versatile Incremental Learning: Towards Class and Domain-Agnostic Adaptation at Any Time Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36442 - Active Budget Can Kill Sensitivity: Diagnosing and Repairing TopK Sparse Autoencoder Reliability Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37857 - Co-PiLOT: Constrained Physics-Informed Latent Optimization for Target-Driven Inverse Design Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37875 - UpliftMem: Learning Set-Level Uplift for Agent Memory Retrieval Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36805 - Divide and Inject: Can Agents Reconstruct an Indirect Prompt Injection from Fragments? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36576 - Inducing Process Supervision from Outcome-Only Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36641 - Guard Models Are Overconfident Where Base Models Are Uncertain Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36477 - Spotter: Let the Embodied Model Lead, and the VLM Reflect for It Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36808 - OPFL: Optimistic Verification of Federated Learning via Empirical Boundary Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37011 - Quantum Computing for Network Security Classification: Near-Term Classification and Long-Term Memory Efficiency Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36479 - CoRe: Co-Evolving Reward Models for Mitigating Latent Reward Hacking in Video Diffusion Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36245 - MatToolBench: Benchmarking Multimodal Agents in Real-World Materials Science Workflows Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37053 - Audience-Bound Persistent Memory: Authorization Across the Memory Lifecycle Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36373 - doPlan: A Variable-Horizon Dataset for Multi-Stage Language-Conditioned Planning in Autonomous Driving Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38028 - From Learner Behavior to Reusable Skills for Effective and Efficient Learner Simulation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37157 - Proofs Without Nominals: G\"odel's Ontological Argument, its Shallow Embedding, and the Open Questions of the Monatshefte Notes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36279 - Stochastic World Models for Verifying Vision-Based Neural Feedback Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38120 - AerialDojo-200K: A Large-Scale Benchmark Suite for Open-World Aerial Object-Goal Search Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36066 - Identifying ODEs from Unstructured Data with Causal Representation Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37083 - ThinkingGuard: Decoding Implicit Hazards via Step-by-Step Risk Attribution in Multimodal Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36562 - KV-Kaizen: Learning Context-Adaptive Cache Compression Choices Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37988 - Designing a Boundary Negotiating Artifact for Collaborative Socio-Technical Sense-Making in AI Regulatory Sandboxes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37109 - Representational Simplicity and Circuit Size Dissociate in a Threshold-Dependent Way: A Controlled Test via Adversarial Training Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35890 - Boids of a Feather Flock Together - Evolving Prey Behaviours Under Different Predator Attack Strategies Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37885 - Multichannel Audio Quality Assessment: Extending Pretrained Perceptual Models to Spatial Audio Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37116 - CounterSteer: Suppressing Indirect Prompt Injection with Activation Steering Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36570 - FairDiff: Mitigating the Self-Reinforcing Matthew Effect in Diffusion Recommender Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36671 - WISE-ATTA: When to Ask for Labels in Budgeted Active Test-Time Adaptation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37687 - SQUARE: Structured Quantum Representation Adapters as Compact Quadratic Feature Maps for Frozen Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37134 - PyroStack: A Multi-Band Spatio-Temporal Sub-Daily Dataset for Wildfires in the United States Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36315 - The Default Trap: Rethinking Plan Evaluation in Tool-Using LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36829 - FocusVTC: Efficient and High-Performance Visual Text Compression with Adaptive Resolution Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36651 - UniAfford: Token-Routed Multitask Learning for Generalizable 2D-3D Affordance Perception Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37264 - Loss-Guided Pretraining Data Selection for Time-Series Foundation Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37255 - Predictive Safety Curricula for Robust Legged Locomotion Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37070 - The Uneven Decline of Collective Knowledge Production: Evidence from Stack Overflow After Generative AI Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36069 - SERA: Scale-Equalized Rollout Allocation for Maximum Likelihood Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36552 - Beyond Conditional Independence: Root Cause Analysis with Deep Causal Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36771 - XU-RS: Explaining Credal Width in Random-Set Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37594 - Cooperative Multi-Agent Vision-Language-Action Models via Reinforced Fine Tuning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36588 - Understanding Decision-Making Mechanisms in Neural Routing Solvers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36063 - Parameterized Stripe Attention for Efficient Video Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37001 - Simple Agentic Memory for Generalist Robot Policies Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36595 - LIBERO-MAX: Do Robot Policies Adapt When the World Changes? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36518 - CheatBench: Measuring Reward Gaming in AI Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36308 - Encore: Few-Shot Agentic Discovery of Manipulation Strategies Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37359 - Demistifying Data and Simulator Assumptions in Supervised Causal Discovery Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37446 - Persona Dosing: Calibrated Activation Steering for Graded Trait Control Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36388 - DIET: Deletion-response Expert Trimming for Video Diffusion Transformers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37829 - Prompted Identity Degrades Cooperation in Multi-Agent LLM Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35928 - Diagnosing and Improving Probabilistic Reasoning in Large Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38005 - Scaling Video Generation for Reasoning: At What Cost? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36599 - Flattening the Connectome Spectrum: A Spectral Filter for FC Induces a Pretraining Target for fMRI Encoders Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37642 - Mirror-Score: Calibrated, Inference-only Scoring Exposes the Limits of Sequence-compatibility Ranking in D-peptide Design Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36057 - TORQUE: Optimizing What (not) to Quantize Before and After Rotation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36032 - AS$^2$D: Accelerating On-Demand Audio Understanding on Mobile Devices Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37617 - Rollout-Marginal Distillation for Long-Horizon Autoregressive Video Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37925 - AVIO: Learning to Add and Remove Sounding Objects in Audiovisual Scenes Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36503 - A Sharp Transition in Data Reconstruction under Differential Privacy Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37344 - CAD-Native Transformer Operators for AI-Aided Engineering Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36806 - Independent Verification Paths Are Not Independent: A Case Study of Common-Mode Failure in a Satellite Catalogue Pipeline Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37603 - ReLMem: Learning Recurrent Memory for Longitudinal EHR Modeling Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37587 - Learn Now, Use Next, Trust Later: Prequential Test-Time Learning for LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35911 - When Tools Silently Lie: Evaluating and Mitigating Blind Compliance in Tool-Augmented Data Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37153 - Visual Parallel Search: Learning to Search High-Resolution Images with Parallel Tile Inspection and Adaptive Zoom Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37002 - NesTok: Nested Self-Aligned 1D Tokenizer for Autoregressive Image Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36756 - HARISSA: Inference-Time Self-Checks for Efficient and Safe Local Language Model Deployment Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.38006 - MAADBench: The Refreshable Paradigm for Anomaly Detection in Multi-Agent Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36556 - Distilling Agentic Systems: A Roadmap across Models, Artifacts, and Harnesses Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36630 - When Upstream Messages Override Correct Answers: A Controlled Study of Multi-Agent LLM Collaboration Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36855 - ExceptionDrive: A Planning-Oriented Counterfactual Corner-Case Benchmark for Autonomous Driving Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37871 - An Empirical Study and Assessment of EU AI Act Compliance Checkers Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36228 - TReVS: Integrating Textual Relevance and Visual Saliency for Efficient Vision-Language Model Token Pruning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37581 - Learning from Shared-Control Overrides: Context-Driven Acceleration Profile Prediction for Personalized Overtaking Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37684 - Multi-Channel Mitigation of Source-Trust Shortcuts in Fact-Checking RL Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36611 - Frontier Autolab: Organizational Memory, Adversarial Dissent and Temporal Leakage in Multi-Agent LLM Firms Across Fifty Years of Technological Change Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36739 - Physics-Informed Multi-Agent Coordination for Hospital Patient Flow Optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37022 - DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37233 - LazySloth: Bounded LLM-based Lazy Tree Search for Fast Long Video Comprehension Source: arxiv-cs-ai Topic: AI Research (+AI, Information Retrieval) URL: https://arxiv.org/abs/2609.37426 - PILLAR: Private Inverted-Index Lexical Lookup for Augmented Retrieval Source: arxiv-cs-ai Topic: AI Research (+AI, Information Retrieval) URL: https://arxiv.org/abs/2609.36326 - PowerZooJax: A JAX-based Power System Benchmark for Reinforcement Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36052 - Width Expansion as a Method for Class Incremental Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37702 - Challenges and Solutions for Bandits in the Wild: Warm-Started Mixture Bandits for Cross-Cohort Slate Recommendation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37800 - VIF-Bench: Evaluating Visual Instruction Following in Multi-Reference Image Generation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37709 - Improving scalable oversight with co-trained monitors Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36049 - The Layer Mystery of VLA: An Information-Theoretical Analysis of VLA Latent Interface Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36118 - State Transport Routing for Short-horizon Adaptation in Multi-horizon Photovoltaic Forecasting Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36926 - Spatiotemporal Hyperedges for EEG Seizure Detection and Prediction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37730 - Beyond Low-Rank Parameterization: Narrowing the Gap Between LoRA and Full Fine-Tuning via Gradient Decomposition Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37027 - Graph neural networks for sampling-invariant embeddings of organized signal sets Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35934 - Grab a Coffee: Future-Aware Guidance for Discrete Diffusion with Compiled Objectives Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35924 - Beyond a single latent space: a dual-latent world model for long-horizon planning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37644 - Reliability Testing of Medical Model Performance under Distributed Deployment Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36525 - Is Human-Readable Text Necessary for Effective LLM Fine-Tuning? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35868 - Learning from Viable Failure Prefixes: Milestone Viability Potential Policy Optimization for Long-Horizon LLM Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37111 - Representable but Unlearned: Encoding Rank and the Interaction-Prediction Floor Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36208 - Beyond Symmetric Agents: Cognitive Diversity and Multi-Agent Debate in Small Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35875 - EASE: Behavior-Adaptive Skill Curation for Self-Evolving Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36746 - Cheap to Hypothesize, Costly to Verify: The Defense Surface of Agentic Vulnerability Discovery Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35909 - FedSocket: Recipient-Executable Knowledge Exchange for Heterogeneous Multimodal Federated Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37582 - LongSpark: Efficient speculative decoding with a fixed-cost parallel drafter Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37029 - Towards an AI Software Factory for Data Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36323 - Memory Is a Derivation: The Distributed-Evidence Paradox in Long-Term Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36130 - When Should Agents Check External State? Budgeting Observations for Stored Intentions Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37125 - SMat-Attention: Structured Long-Context Sequence Modeling Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36062 - OptiCom : A Unified Framework for State-Conditioned Composition in LLM-Driven Optimization Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37221 - MERID: Multimodal Exploration via Recursive Self-Improvement Agents for Major Depression Analysis Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36235 - LongCat-DeepResearch Technical Report Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36071 - ImbalancE: Inference-Time Latent Search Against Degree Imbalance in Link Prediction Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36996 - HEAR: Real Voices, Real Bias: A Large-Scale Human-Recorded, Demographically Diverse Benchmark for Audio Language Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.35952 - StateTape: Action-Conditioned Evidence Lifecycle Modeling for Long-Horizon Coding Agents Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36319 - CRASM-Gate: Deterministic-First Constraint- and Role-Aware Semantic Mapping with Selective Model Assistance Across Heterogeneous Industrial Standards Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37458 - Distinguish or Homogenize: Last-Chance Policy Identification and Risk-Budgeted Recovery under Irreversible Resource Depletion Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36741 - From Checkpoint Variation to Selection Gains in Supervised Fine-Tuning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36569 - Risk-Aware Semantic Grounding for Trustworthy LLM-Based Robot Planning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37554 - AgentBug-Smith: Automatically Reproducing Real-World Harness Bugs in Agentic Systems Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37864 - AdaKerNet: Neural Kernel Decoding for Task-Adaptive Prediction with Multimodal Large Models Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36368 - Learning as Deepfakes Evolve: RF-Prompt for Continual Audio Deepfake Detection Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37586 - Longer Records, Broader Invariance: The Hidden Scaling Problem in Longitudinal Contrastive Learning Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36409 - Going Beyond State-Reaching: Learning Abstractions for Intrinsically Motivated Option Discovery Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36473 - Beyond Sub-Gaussian Detector Scores: Robust Weighted Profile-Loss Change Point Detection for Human-LLM Text Segmentation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36888 - EgoHumanoid-V2: Human-to-Humanoid Transfer of Coordinated Whole-Body Skills for Loco-Manipulation Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37181 - How Can Recommendation Feedback Evolve Agent Memory? Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37544 - AeroManip-VLA: Scalable Vision-Language-Action Learning for Aerial Manipulation with RL-Generated Demonstrations Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36915 - HorizonFlow: Variable-Length Planning for Offline Goal-Conditioned RL Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36896 - Benchmarking Vision-Language Models on Synapse Detection and Proofreading in Connectomics Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36492 - WitnessGym: Benchmarking Coding Agents on the Construction of Bug Witnesses Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36635 - FACT: Fidelity-Aware Construction of Articulated Twins Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37067 - Transolver-$\sigma$: Joint Spectral-Physical Subspace Modeling for Neural PDE Solving Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.37279 - Factorized Scheduling Principle: Learning Interpretable and Transferable Policies via Structured Additive Functions Source: arxiv-cs-ai Topic: AI Research (+AI) URL: https://arxiv.org/abs/2609.36578