Papers

New on arXiv

Tue 21 Jul 2026

352 papers

Position: Recommender Systems Should Move Beyond Platform-Centric Ranking toward Personal Agent-Mediated RecommendationLLM agents and tool use → Agentic recommender systems
InternReviewer & InternAdvocate: Objective Reward and Evaluation for Agentic Reinforcement Learning in Peer Review and RebuttalLLM agents and tool use → Web, search, and research agents
PA-CoT: Profile-Adaptive Chain-of-Thought for Personalized Nutritional ConsultingLLM reasoning → Clinical and biomedical applications
Don't Want Your LLM to Recommend Nuclear Strike? Try Asking It in JapaneseLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment +1
MaSRead: Content-Addressed Reading of Replicated Latent StoresEfficient LLM inference and serving → KV cache optimization
Speculative Correction: Draft-then-Refine Decoding for Diffusion Language ModelsDiffusion models → Discrete and text diffusion
Linguistic Context Recodes Visual Representations in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Empowering Cross-Domain Sequential Recommendation with Hybrid Tokenization and Serial-Parallel DecodingEfficient LLM inference and serving → Speculative decoding
An Interactive Vision Language Platform for Cognitive Remediation in SchizophreniaMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
DAMamba-UNet3D: A Parameter-Efficient Mamba State Space U-Net with Dynamic Adaptive Scan for 3D Medical Image SegmentationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Visual Token Compression Enhances Robustness of MLLMsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +2
Child-Oriented AIGC Video Risk Reviewing: A Benchmark and Knowledge-Supported Iterative Reasoning FrameworkLLM reasoning → Multimodal reasoning
SEGRA: A Structured Experience Guided Reasoning Agent for Property Graph Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning +1
Multi-Horizon Consistency as Geometry: When Latent Dynamics Contract, and When They Do NotWorld models → Latent predictive world models
CARE: Pre-Execution Command Verification for Shell-Executing LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
Twin Agent: Context Residual Compression for Privilege Separated AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Knowledge-Centric Self-ImprovementSynthetic data and self-improvement for LLMs → Self-evolving LLM agents
ChronoStitch: Training-Free Composition of Visual KV Memories for Long-Horizon Temporal ReasoningLLM reasoning → Multimodal reasoning
Fine-grained Computation-Communication Overlap via Tile-level Signaling and Scheduling for Mixture-of-ExpertsMixture-of-experts → Efficient MoE training and inference systems
D3VL: Understanding Driving Scenes from 3D Time Series Data and Video with Language ModelsMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
When Reasoning Narrows the Move: Diversity Collapse in LLM Game PlayLLM reasoning → Reasoning for agents, planning and code
Strong Gravitational Lensing Posterior Sampling in Pixel-Space Using Diffusion Models and Recurrent Inference MachinesDiffusion models → Inverse problems and image restoration
MoA-Structured Decode Attention DNF Derivation, KV-Cache Accumulation, GQA/MQA, and OpenACC KernelEfficient LLM inference and serving → KV cache optimization
REGEN: Replay-recycling for Expert-to-Generalist distillation with Offline Reinforcement LearningLLM agents and tool use → Reinforcement learning for LLM agents +1
Guardrails as Scapegoats: Auditing Unfaithful Safety Refusals in Tool-Augmented LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Unlearning as Distribution Restoration: A Controlled Counterfactual Study, a Validated Selective Screen, and the Limits of Oracle-Free CertificationLLM safety and security → Machine unlearning
BaseRT: Advancing Best-in-Class LLM Inference with Apple M5 Neural AcceleratorsMixture-of-experts → Efficient MoE training and inference systems
Building Trust in Autonomous Commerce: A Verifiable Global Event Timeline and AI-Ready Fraud Intelligence LayerLLM agents and tool use → Agent threats, guardrails and governance
Copy Less, Ground More: Overcoming Repetitive Copying in Long-Context Reasoning via Evidence-Aware Reinforcement LearningLLM reasoning → Test-time compute and efficient thinking
Appearance Pointers -- Multimodal Region Control of Diffusion TransformersDiffusion models → Image generation, text-to-image and editing
Masked Visual Actions for Unified World ModelingWorld models → Embodied and robot world models
ExpertVerse: A General-Purpose Benchmark for Expert-Level Reasoning in Knowledge-Intensive Visual SynthesisLLM reasoning → Multimodal reasoning
OmniReasoner: Thinking with Long Audio-Video via Native Tool UseLLM reasoning → Audio-language model reasoning
CodeRescue: Budget-Calibrated Recovery Routing for Coding AgentsLLM agents and tool use → Software engineering coding agents
Agents in the Wild: Where Research Meets DeploymentLLM reasoning → Reasoning for agents, planning and code +1
Provable diffusion-based posterior sampling for linear inverse problems via DDIMDiffusion models → Inverse problems and image restoration
ROMS-IMLE: A Minimalist Approach to Competitive Single-Step Generative ModellingDiffusion models → Diffusion foundations and sampling
ISO: An RLVR-Native Optimization StackLLM reasoning → Reinforcement learning for reasoning
Selective State-Space Adaptation and Retrieval for Language Model ReasoningLLM reasoning → Test-time compute and efficient thinking +1
Off-Context GRPO: Learning to Reason on Hard Problems using Privileged InformationLLM reasoning → Reinforcement learning for reasoning
Graph-Based Agentic AI with LangGraph: Workflow Pathways for Long-Running Stateful Business ProcessesLLM agents and tool use → Software engineering coding agents
No Training, Better Flights: Test-Time Scaled VLMs for UAV NavigationLLM reasoning → Spatial and embodied reasoning +1
BioSecBench-Surveillance: A Verifiable Benchmark for AI Agents in Pathogen Genomic SurveillanceLLM agents and tool use → Medical and scientific domain agents
EviPathBench: Benchmarking Evidence Acquisition and Reasoning in Vision-Language Models for Whole-Slide PathologyLLM reasoning → Clinical and biomedical applications +1
Inference-Time Steering for Cross-Lingual Factual Consistency in LLMsLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
MeetingToM: Evaluating Multimodal LLMs on Theory-of-Mind Reasoning in Multi-Party MeetingsLLM reasoning → Reasoning benchmarks and evaluation +1
The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine TranslationLLM reasoning → Reinforcement learning for reasoning
AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion DraftersEfficient LLM inference and serving → Speculative decoding
Conservative Query and Adaptive Regularization for Offline RL Under Uncertainty EstimationOffline RL and imitation learning from data → Offline RL algorithms
Cognitive Dual-Process Planning for Autonomous Driving with Structured Scene Knowledge and Verifiable Reasoning-Action ConsistencyLLM reasoning → Chain-of-thought reasoning for driving +1
ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPUWorld models → Video-generation world models and simulators
Agentic Real2Sim: Physics-based World Modeling with Vision-Language AgentsWorld models → Embodied and robot world models +1
Reasoning Before Translation: Enhancing Legal Machine Translation with Structured ReasoningLLM reasoning → Reasoning benchmarks and evaluation
Text Template Tokens Are Implicit Semantic Registers in Diffusion TransformersDiffusion models → Image generation, text-to-image and editing
One Model, Many Graphs: Learning over Attributed Graphs across Heterogeneous Modalities with Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Parallel Noising in Neural Markov Logic NetworksDiffusion models → Discrete and text diffusion
Latent Riemannian Flow Matching for Geometry-Grounded 3D Foundation ModelsDiffusion models → Flow matching, time series and science
DAIS: Dependency-Aware Intermediate QA Supervision for Complex ReasoningLLM reasoning → Prompting and chain-of-thought reasoning
Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and EditingDiffusion models → Image generation, text-to-image and editing
Where Should Optimizer State Live? Tiered State Allocation for Memory-Efficient Mixture-of-Experts TrainingMixture-of-experts → Efficient MoE training and inference systems
FilmWorld: Agentic Novel-to-Film Generation through Dynamic Cinematic World ModelingWorld models → Video-generation world models and simulators +1
Mitigating Modality and Language-Style Gaps for Zero-Shot Video Moment RetrievalMultimodal LLMs / vision-language models → Video understanding with MLLMs
Computational Humor with Multimodal LLMs: Methods, Datasets, Evaluation, and ChallengesMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy OptimizationMultimodal LLMs / vision-language models → Medical vision-language models +1
Fishing Out Free Riders: Shapley-Based Reward Attribution for Parallel Reasoning via Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning +1
Verifiable Self-Evolution for Open-Ended Dialogue Skills via Future-Feedback PredictionLLM agents and tool use → General LLM agent frameworks and benchmarks
Dual Adversarial Fine-tuning for Enhancing Robustness of Large Vision Language ModelMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
H2SD: Hybrid Hindsight Self-DistillationLLM reasoning → Reinforcement learning for reasoning
TAP-RAG: Task-Aware Policy Control for Long-Document Multimodal Question AnsweringRetrieval-augmented generation → Multimodal RAG
Reasoning Error from Known Fact: Step-Level Self-Consistency Group Relative Policy Optimization for LLMLLM reasoning → Hallucination detection and mitigation
Black-Mamba: Biologically-Inspired Leaky Accumulation for Conceptual Knowledge under Distribution DriftState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
TraceDev: A Traceability-Driven Multi-agent Framework for Requirement-to-Code DevelopmentLLM agents and tool use → Software engineering coding agents
PhoenixRepair: Rethinking Repair Strategy Exploration in Software AgentsLLM reasoning → Reasoning for agents, planning and code +1
OPD-IAD: From Language Judgment to Industrial Anomaly Detection via On-Policy Self-DistillationMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Data Leakage Prevention in Agentic Applications via Preemptive HardeningLLM agents and tool use → Agent threats, guardrails and governance +1
WorldScape Policy 2.0: Empowering Steerable World Action Modeling with Reasoning-Augmented Memory and In-Context LearningWorld models → Embodied and robot world models
HPD-Parsing: Hierarchical Parallel Document ParsingEfficient LLM inference and serving → Speculative decoding
Cross-Agent Campaign Attribution: Linking Asynchronous Attacks Across LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought FaithfulnessLLM reasoning → Prompting and chain-of-thought reasoning
In-Context Learning for Wound Classification with Small Multimodal Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
AgentTrails: Towards Trust and Reuse for Agentic TasksLLM agents and tool use → Software engineering coding agents
AI Tour Meeting: Group Travel Planning by LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
SkillSight: Calibrating Generic Content Bias for Skill RetrievalLLM agents and tool use → Agent skill learning and evolution
PertReason: A Knowledge-Grounded Benchmark and Framework for Cell-State-Conditioned Mechanistic Reasoning of Perturbation EffectsLLM reasoning → Clinical and biomedical applications
RF-Agent: A Practical Framework for Building Language Agents for RFIC DesignLLM agents and tool use → Software engineering coding agents
RAGAL: A Frugal, Fully Local Retrieval-Augmented Assistant for Technical Support at a Government AgencyRetrieval-augmented generation → RAG pipeline design and evaluation
AgentDebugX: An Open-Source Toolkit for Failure Observability, Attribution, and Recovery in LLM AgentsLLM agents and tool use → Software engineering coding agents
Beyond Text Editing: Algebraic Manipulation of Source CodeLLM agents and tool use → Software engineering coding agents
Stale but Stable: Staleness-Adaptive Trust Regions for Stabilizing Asynchronous Reinforcement LearningMixture-of-experts → MoE in reinforcement learning and robotics
Continual Video-MLLM Adaptation over Evolving DomainsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
DWM: Separating World Effects from Actions in Latent World ModelsWorld models → Latent predictive world models
RoboInter1.5: A Holistic Intermediate Representation Suite for Embodied World Modeling and Robotic ManipulationWorld models → Embodied and robot world models
Attributes Should Come from Images, Not Class Names: Distribution-Conditioned Attribute Selection for Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
MissingBench-Verified: Probing Vision-Language Models' Inability to Detect Missing Object PartsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Broken Gates: Re-evaluating Web Bot Defenses in the Age of LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
Mark, Don't Erase: Token Inoculation for Dual-Use Knowledge in LLMsLLM safety and security → Safety alignment, refusal and guardrails
Searching for Plans You Can Actually Build: A Realizability-Aware Full-Space Optimizer for MoE Training and ServingMixture-of-experts → Efficient MoE training and inference systems
Norm or Direction? Decoding Vision Mambas for High-Resolution VisionState-space sequence models (S4, Mamba) → Vision Mamba backbones
CPInj: Uncovering Prompt Injection Risks in Textual Collaborative Prompt OptimizationLLM safety and security → Prompt injection and agent security
LatentMT: Machine Translation with Latent ReasoningLLM reasoning → Reasoning benchmarks and evaluation

All 352