Papers

New on arXiv

Fri 14 Mar 2025

362 papers

Relevance Isn't All You Need: Scaling RAG Systems With Inference-Time Compute Via Multi-Criteria RerankingLLM reasoning → Knowledge- and tool-augmented reasoning +1
Policy Frameworks for Transparent Chain-of-Thought Reasoning in Large Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
Agent-Enhanced Large Language Models for Researching Political InstitutionsLLM agents and tool use → Web, search, and research agents
CURIE: Evaluating LLMs On Multitask Scientific Long Context Understanding and ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
RAG-KG-IL: A Multi-Agent Hybrid Framework for Reducing Hallucinations and Enhancing LLM Reasoning through RAG and Incremental Knowledge Graph Learning IntegrationLLM reasoning → Knowledge- and tool-augmented reasoning +2
Monitoring Reasoning Models for Misbehavior and the Risks of Promoting ObfuscationLLM reasoning → Prompting and chain-of-thought reasoning +1
Sketch-to-Skill: Bootstrapping Robot Learning with Human Drawn Trajectory SketchesOffline RL and imitation learning from data → Robot manipulation imitation learning +1
Upcycling Text-to-Image Diffusion Models for Multi-Task CapabilitiesDiffusion models → Image generation, text-to-image and editing
FedALT: Federated Fine-Tuning through Adaptive Local Training with Rest-of-World LoRAMixture-of-experts → Federated mixture-of-experts
Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-Tuning and Can Be Mitigated by Machine UnlearningMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Key, Value, Compress: A Systematic Exploration of KV Cache Compression TechniquesEfficient LLM inference and serving → KV cache optimization
Diffuse-CLoC: Guided Diffusion for Physics-based Character Look-ahead ControlDiffusion models → Diffusion and flow policies for robotics
Semantic-Clipping: Efficient Vision-Language Modeling with Semantic-Guidedd Visual SelectionMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Visualizing Thought: Conceptual Diagrams Enable Robust Planning in LMMsLLM reasoning → Multimodal reasoning
Controllable Latent Diffusion for Traffic SimulationDiffusion models → Diffusion and flow policies for robotics
reWordBench: Benchmarking and Improving the Robustness of Reward Models with Transformed InputsLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Making Every Step Effective: Jailbreaking Large Vision-Language Models Through Hierarchical KV EqualizationMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Safe Vision-Language Models via Unsafe Weights ManipulationMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
BioMamba: Leveraging Spectro-Temporal Embedding in Bidirectional Mamba for Enhanced Biosignal ClassificationState-space sequence models (S4, Mamba) → Physiological signal modeling
CoLLMLight: Cooperative Large Language Model Agents for Network-Wide Traffic Signal ControlLLM agents and tool use → Multimodal and embodied agents
LLM Agents for Education: Advances and ApplicationsLLM agents and tool use → Educational agents and math problem solving
Centaur: Robust End-to-End Autonomous Driving with Test-Time TrainingEnd-to-end autonomous driving → End-to-end system architecture
Adversarial Data Collection: Human-Collaborative Perturbations for Efficient and Robust Robotic Imitation LearningOffline RL and imitation learning from data → Robot manipulation imitation learning
Tit-for-Tat: Safeguarding Large Vision-Language Models Against Jailbreak Attacks via Adversarial DefenseLLM safety and security → Vision-language and text-to-image safety
From Score Matching to Diffusion: A Fine-Grained Error Analysis in the Gaussian SettingDiffusion models → Diffusion foundations and sampling
Neutralizing Bias in LLM Reasoning using Entailment GraphsLLM reasoning → Reasoning benchmarks and evaluation
Rethinking Few-Shot Adaptation of Vision-Language Models in Two StagesLLM reasoning → Multimodal reasoning +1
Vamba: Understanding Hour-Long Videos with Hybrid Mamba-TransformersMultimodal LLMs / vision-language models → Video understanding with MLLMs +2
SmolDocling: An ultra-compact vision-language model for end-to-end multi-modal document conversionMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Implicit Bias-Like Patterns in Reasoning ModelsLLM reasoning → Reasoning benchmarks and evaluation
VERIFY: A Benchmark of Visual Explanation and Reasoning for Investigating Multimodal Reasoning FidelityLLM reasoning → Multimodal reasoning
Similarity-Aware Token Pruning: Your VLM but FasterMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
AugGen: Synthetic Augmentation using Diffusion Models Can Improve RecognitionDiffusion models → Image generation, text-to-image and editing
Prompt Injection Detection and Mitigation via AI Multi-Agent NLP FrameworksLLM safety and security → Prompt injection and agent security
V-STaR: Benchmarking Video-LLMs on Video Spatio-Temporal ReasoningLLM reasoning → Multimodal reasoning
Unicorn: A Universal and Collaborative Reinforcement Learning Approach Towards Generalizable Network-Wide Traffic Signal ControlDeep reinforcement learning methods → RL for autonomous driving and traffic
A Review of DeepSeek Models' Key Innovative TechniquesLLM reasoning → Reasoning benchmarks and evaluation
Dynamic Obstacle Avoidance with Bounded Rationality Adversarial Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
Cerebrum (AIOS SDK): A Platform for Agent Development, Deployment, Distribution, and DiscoveryLLM agents and tool use → General LLM agent frameworks and benchmarks
Combining Causal Models for More Accurate Abstractions of Neural NetworksMechanistic interpretability → Circuits and mechanistic analysis
TASTE-Rob: Advancing Video Generation of Task-Oriented Hand-Object Interaction for Generalizable Robotic ManipulationOffline RL and imitation learning from data → Robot manipulation imitation learning
Classifying Long-tailed and Label-noise Data via Disentangling and UnlearningLLM safety and security → Machine unlearning
Empowering Time Series Analysis with Synthetic Data: A Survey and Outlook in the Era of Foundation ModelsSynthetic data and self-improvement for LLMs → Model collapse from synthetic data
Towards A Correct Usage of Cryptography in Semantic Watermarks for Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
A Framework for a Capability-driven Evaluation of Scenario Understanding for Multimodal Large Language Models in Autonomous DrivingMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
Reinforcement Learning-Based Controlled Switching Approach for Inrush Current Minimization in Power TransformersDeep reinforcement learning methods → RL for energy, networks and operations
Hierarchical Information-Guided Spatio-Temporal Mamba for Stock Time Series ForecastingState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
Creating a Good Teacher for Knowledge Distillation in Acoustic Scene ClassificationKnowledge distillation → Speech and audio distillation
PARIC: Probabilistic Attention Regularization for Language Guided Image Classification from Pre-trained Vison Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
RESPONSE: Benchmarking the Ability of Language Models to Undertake Commonsense Reasoning in Crisis SituationLLM reasoning → Reasoning benchmarks and evaluation
AIstorian lets AI be a historian: A KG-powered multi-agent system for accurate biography generationLLM agents and tool use → Web, search, and research agents
Road Rage Reasoning with Vision-language Models (VLMs): Task Definition and Evaluation DatasetLLM reasoning → Chain-of-thought reasoning for driving +1
Contextual Similarity Distillation: Ensemble Uncertainties with a Single ModelOffline RL and imitation learning from data → Offline RL algorithms +1
Rule-Guided Feedback: Enhancing Reasoning by Enforcing Rule Adherence in Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Learning to reset in target search problemsDeep reinforcement learning methods → Deep RL algorithms and representations
Safe-VAR: Safe Visual Autoregressive Model for Text-to-Image Generative WatermarkingLLM safety and security → Watermarking generated content and models
Unlocking General Long Chain-of-Thought Reasoning Capabilities of Large Language Models via Representation EngineeringLLM reasoning → Prompting and chain-of-thought reasoning +1
GNNs as Predictors of Agentic Workflow PerformancesLLM agents and tool use → Software engineering coding agents
DynRsl-VLM: Enhancing Autonomous Driving Perception with Dynamic Resolution Vision-Language ModelsMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
Dark Noise Diffusion: Noise Synthesis for Low-Light Image DenoisingDiffusion models → Diffusion foundations and sampling
Reasoning-Grounded Natural Language Explanations for Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Compound Expression Recognition via Large Vision-Language ModelsMultimodal LLMs / vision-language models → Multimodal emotion understanding
Towards Better Alignment: Training Diffusion Models with Reinforcement Learning Against Sparse RewardsDiffusion models → Diffusion and flow policies for robotics
PrivacyScalpel: Enhancing LLM Privacy via Interpretable Feature Intervention with Sparse AutoencodersMechanistic interpretability → SAE features and their applications
Exploring the Potential of Large Multimodal Models as Effective Alternatives for Pronunciation AssessmentMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Can Large Reasoning Models do Analogical Reasoning under Perceptual Uncertainty?LLM reasoning → Reasoning benchmarks and evaluation
Reinforcement Learning Outperforms Supervised Fine-Tuning: A Case Study on Audio Question AnsweringLLM reasoning → Audio-language model reasoning +1
Provenance Detection for AI-Generated Images: Combining Perceptual Hashing, Homomorphic Encryption, and AI Detection ModelsLLM safety and security → Watermarking generated content and models
Bleeding Pathways: Vanishing Discriminability in LLM Hidden States Fuels Jailbreak AttacksLLM safety and security → Jailbreaking
Multi-Stage Generative Upscaler: Reconstructing Football Broadcast Images via Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Don't Take Things Out of Context: Attention Intervention for Enhancing Chain-of-Thought Reasoning in Large Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
Enabling Weak Client Participation via On-device Knowledge Distillation in Heterogeneous Federated LearningKnowledge distillation → Federated, robust and private distillation
MoLEx: Mixture of Layer Experts for Finetuning with Sparse UpcyclingMixture-of-experts → MoE large language models
GaussianIP: Identity-Preserving Realistic 3D Human Generation via Human-Centric Diffusion PriorDiffusion models → 3D, point cloud and scene generation
X-EcoMLA: Upcycling Pre-Trained Attention into MLA for Efficient and Extreme KV CompressionEfficient LLM inference and serving → KV cache optimization
Don't Forget It! Conditional Sparse Autoencoder Clamping Works for UnlearningLLM safety and security → Machine unlearning +1
MUSS: Multilevel Subset Selection for Relevance and DiversityRetrieval-augmented generation → RAG pipeline design and evaluation
DriveGEN: Generalized and Robust 3D Detection in Driving via Controllable Text-to-Image Diffusion GenerationDiffusion models → Diffusion and flow policies for robotics
Time and Memory Trade-off of KV-Cache Compression in Tensor Transformer DecodingEfficient LLM inference and serving → KV cache optimization
Open3D-VQA: A Benchmark for Comprehensive Spatial Reasoning with Multimodal Large Language Model in Open SpaceLLM reasoning → Spatial and embodied reasoning +1
EmbodiedVSR: Dynamic Scene Graph-Guided Chain-of-Thought Reasoning for Visual Spatial TasksLLM reasoning → Spatial and embodied reasoning +1
Semantic and Contextual Modeling for Malicious Comment Detection with BERT-BiLSTMLLM safety and security → Safety alignment, refusal and guardrails
Understanding Flatness in Generative Models: Its Role and BenefitsDiffusion models → Diffusion foundations and sampling
Exploring the Necessity of Reasoning in LLM-based Agent ScenariosLLM reasoning → Reasoning for agents, planning and code +1
Perceive, Understand and Restore: Real-World Image Super-Resolution with Autoregressive Multimodal Generative ModelsDiffusion models → Inverse problems and image restoration
Harnessing Frequency Spectrum Insights for Image Copyright Protection Against Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
Falcon: A Remote Sensing Vision-Language Foundation Model (Technical Report)Multimodal LLMs / vision-language models → Embodied and 3D vision-language models
API Agents vs. GUI Agents: Divergence and ConvergenceLLM agents and tool use → GUI and mobile agents
Low-cost Real-world Implementation of the Swing-up Pendulum for Deep Reinforcement Learning ExperimentsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Active Learning from Scene Embeddings for End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
Training Directional Locomotion for Quadrupedal Low-Cost Robotic Systems via Deep Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
Flow to the Mode: Mode-Seeking Diffusion Autoencoders for State-of-the-Art Image TokenizationDiffusion models → Diffusion foundations and sampling
LUSD: Localized Update Score Distillation for Text-Guided Image EditingDiffusion models → Image generation, text-to-image and editing
InverseBench: Benchmarking Plug-and-Play Diffusion Priors for Inverse Problems in Physical SciencesDiffusion models → Inverse problems and image restoration
FMNet: Frequency-Assisted Mamba-Like Linear Attention Network for Camouflaged Object DetectionState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
EmoDiffusion: Enhancing Emotional 3D Facial Animation with Latent Diffusion ModelsDiffusion models → Audio, speech and music
MAVFlow: Preserving Paralinguistic Elements with Conditional Flow Matching for Zero-Shot AV2AV Multilingual TranslationDiffusion models → Audio, speech and music
Residual Policy Gradient: A Reward View of KL-regularized ObjectiveOffline RL and imitation learning from data → Offline RL algorithms +1
Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object DetectionMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Statistical Impossibility and Possibility of Aligning LLMs with Human Preferences: From Condorcet Paradox to Nash EquilibriumLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF

All 362