Papers

New on arXiv

Tue 11 Feb 2025

385 papers

FlexControl: Computation-Aware ControlNet with Differentiable Router for Text-to-Image GenerationDiffusion models → Image generation, text-to-image and editing
Trustworthy AI: Safety, Bias, and Privacy -- A SurveyLLM safety and security → Safety alignment, refusal and guardrails
MoHAVE: Mixture of Hierarchical Audio-Visual Experts for Robust Speech RecognitionMixture-of-experts → Multimodal MoE models
Principled Data Selection for Alignment: The Hidden Risks of Difficult ExamplesLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
ImitDiff: Transferring Foundation-Model Priors for Distraction Robust Visuomotor PolicyOffline RL and imitation learning from data → Robot manipulation imitation learning
Model Selection for Off-policy Evaluation: New Algorithms and Experimental ProtocolOffline RL and imitation learning from data → Offline RL algorithms +1
Training-Free Safe Denoisers for Safe Use of Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects
Greed is Good: A Unifying Perspective on Guided GenerationDiffusion models → Diffusion foundations and sampling
DiffRatio: Training One-Step Diffusion Models Without Teacher SupervisionDiffusion models → Diffusion foundations and sampling
Unveiling Client Privacy Leakage from Public Dataset Usage in Federated DistillationLLM safety and security → Federated unlearning and poisoning
Universal Adversarial Attack on Aligned Multimodal LLMsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
MetaSC: Test-Time Safety Specification Optimization for Language ModelsLLM reasoning → Reasoning for agents, planning and code
CIRCUIT: A Benchmark for Circuit Interpretation and Reasoning Capabilities of LLMsLLM reasoning → Reasoning for agents, planning and code
A Survey of In-Context Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Training Sparse Mixture Of Experts Text Embedding ModelsMixture-of-experts → MoE large language models
Intrinsic Bias is Predicted by Pretraining Data and Correlates with Downstream Performance in Vision-Language EncodersMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
Advancing Autonomous VLM Agents via Variational Subgoal-Conditioned Reinforcement LearningMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
SurGrID: Controllable Surgical Simulation via Scene Graph to Image DiffusionDiffusion models → Medical imaging and segmentation
Symbiotic Cooperation for Web Agents: Harnessing Complementary Strengths of Large and Small LLMsLLM agents and tool use → Web, search, and research agents
Bit-Level Discrete Diffusion with Markov Probabilistic Models: An Improved Framework with Sharp Convergence Bounds under Minimal AssumptionsDiffusion models → Diffusion foundations and sampling
Active Advantage-Aligned Online Reinforcement Learning with Offline DataOffline RL and imitation learning from data → Offline RL algorithms +1
Distributed Approach to Haskell Based Applications Refactoring with LLMs Based Multi-Agent SystemsLLM agents and tool use → Software engineering coding agents
Elevating Legal LLM Responses: Harnessing Trainable Logical Structures and Semantic Knowledge with Legal ReasoningLLM reasoning → Reasoning for agents, planning and code
DeepSeek on a Trip: Inducing Targeted Visual Hallucinations via Representation VulnerabilitiesMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
TransMLA: Multi-Head Latent Attention Is All You NeedEfficient LLM inference and serving → KV cache optimization
Mamba Adaptive Anomaly Transformer with association discrepancy for time seriesState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
MaRS: A Fast Sampler for Mean Reverting Diffusion based on ODE and SDE SolversDiffusion models → Diffusion foundations and sampling
Vision-Language Models for Edge Networks: A Comprehensive SurveyMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Confidence-calibrated covariate shift correction for few-shot classification in Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Spread them Apart: Towards Robust Watermarking of Generated ContentLLM safety and security → Watermarking generated content and models
Optimal Actuator Attacks on Autonomous Vehicles Using Reinforcement LearningEnd-to-end autonomous driving → Adversarial scenarios and safety testing +1
NanoVLMs: How small can we go and still make coherent Vision Language Models?Multimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
RoboBERT: An End-to-end Multimodal Robotic Manipulation ModelDiffusion models → Diffusion and flow policies for robotics
Polynomial-Time Approximability of Constrained Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
WHODUNIT: Evaluation benchmark for culprit detection in mystery storiesLLM reasoning → Reasoning benchmarks and evaluation
Near-Optimal Sample Complexity in Reward-Free Kernel-Based Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
MAGELLAN: Metacognitive predictions of learning progress guide autotelic LLM agents in large goal spacesLLM agents and tool use → General LLM agent frameworks and benchmarks
Causal Imitation Learning under Expert-Observable and Expert-Unobservable ConfoundingOffline RL and imitation learning from data → Imitation learning methods
From Action Labels to Sets: Rethinking Action Supervision for Imitation Learning from Corrective FeedbackOffline RL and imitation learning from data → Imitation learning methods
Consistency Training with Physical ConstraintsDiffusion models → Flow matching, time series and science
Distributed Value Decomposition Networks with Networked AgentsDeep reinforcement learning methods → RL for games and combinatorial optimization
Divide and Merge: Motion and Semantic Learning in End-to-End Autonomous DrivingEnd-to-end autonomous driving → Reasoning VLA and VLM driving policies
Scaling Pre-training to One Hundred Billion Data for Vision Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language ModelsLLM reasoning → Chain-of-thought reasoning for driving +1
DPO-Shift: Shifting the Distribution of Direct Preference OptimizationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
DMWM: Dual-Mind World Model with Long-Term ImaginationWorld models → World models for language agents
DSV: Exploiting Dynamic Sparsity to Accelerate Large-Scale Video DiT TrainingDiffusion models → Video generation
SEMU: Singular Value Decomposition for Efficient Machine UnlearningLLM safety and security → Machine unlearning
Single-Step Consistent Diffusion SamplersDiffusion models → Diffusion foundations and sampling
PIM Is All You Need: A CXL-Enabled GPU-Free System for Large Language Model InferenceEfficient LLM inference and serving → Serving systems and request scheduling
JBShield: Defending Large Language Models from Jailbreak Attacks through Activated Concept Analysis and ManipulationLLM safety and security → Jailbreaking
O1 Embedder: Let Retrievers Think Before ActionLLM reasoning → Knowledge- and tool-augmented reasoning
Scaling Off-Policy Reinforcement Learning with Batch and Weight NormalizationDeep reinforcement learning methods → Deep RL algorithms and representations
The Devil is in the Prompts: De-Identification Traces Enhance Memorization Risks in Synthetic Chest X-Ray GenerationDiffusion models → Medical imaging and segmentation
Multi-Agent Collaboration for Multilingual Code Instruction TuningLLM agents and tool use → Software engineering coding agents
Less is More: Masking Elements in Image Condition Features Avoids Content Leakages in Style Transfer Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Logarithmic Regret for Online KL-Regularized Reinforcement LearningLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
Approximating Human Strategic Reasoning with LLM-Enhanced Recursive Reasoners Leveraging Multi-agent HypergamesLLM reasoning → Reasoning for agents, planning and code +1
Optimizing Knowledge Distillation in Transformers: Enabling Multi-Head Attention without Alignment BarriersKnowledge distillation → General knowledge distillation methods
RomanLens: The Role Of Latent Romanization In Multilinguality In LLMsMechanistic interpretability → Knowledge and representations inside LLMs
Towards a Formal Theory of the Need for Competence via Computational Intrinsic MotivationDeep reinforcement learning methods → Deep RL algorithms and representations
MoENAS: Mixture-of-Expert based Neural Architecture Search for jointly Accurate, Fair, and Robust Edge Deep Neural NetworksMixture-of-experts → Vision mixture-of-experts
MGPATH: Vision-Language Model with Multi-Granular Prompt Learning for Few-Shot WSI ClassificationMultimodal LLMs / vision-language models → Medical vision-language models
UAV-assisted Joint Mobile Edge Computing and Data Collection via Matching-enabled Deep Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
Spatial Degradation-Aware and Temporal Consistent Diffusion Model for Compressed Video Super-ResolutionDiffusion models → Inverse problems and image restoration
Wheeled Lab: Modern Sim2Real for Low-cost, Open-source Wheeled RoboticsDeep reinforcement learning methods → RL for robot locomotion and manipulation
LLMs Can Easily Learn to Reason from Demonstrations Structure, not content, is what matters!LLM reasoning → Prompting and chain-of-thought reasoning
EvoFlow: Evolving Diverse Agentic Workflows On The FlyLLM agents and tool use → Software engineering coding agents
KABB: Knowledge-Aware Bayesian Bandits for Dynamic Expert Coordination in Multi-Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
Aligning Large Language Models to Follow Instructions and Hallucinate Less via Effective Data FilteringLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
PICTS: A Novel Deep Reinforcement Learning Approach for Dynamic P-I Control in Scanning Probe MicroscopyDeep reinforcement learning methods → RL for energy, networks and operations
Semantic to Structure: Learning Structural Representations for Infringement DetectionDiffusion models → Robustness, privacy and adversarial aspects
CodeI/O: Condensing Reasoning Patterns via Code Input-Output PredictionLLM reasoning → Reasoning for agents, planning and code
Semi-Supervised Vision-Centric 3D Occupancy World Model for Autonomous DrivingWorld models → World models for autonomous driving +1
TRAVEL: Training-Free Retrieval and Alignment for Vision-and-Language NavigationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Flow Matching for Collaborative FilteringDiffusion models → Diffusion recommenders
MM-DADM: Multimodal Drug-Aware Diffusion Model for Virtual Clinical TrialsDiffusion models → Medical imaging and segmentation
MIGT: Memory Instance Gated Transformer Framework for Financial Portfolio ManagementDeep reinforcement learning methods → Adversarial robustness and RL in finance
Exploratory Diffusion Model for Unsupervised Reinforcement LearningDiffusion models → Diffusion and flow policies for robotics +1
Articulate That Object Part (ATOP): 3D Part Articulation via Text and Motion PersonalizationDiffusion models → Video generation
When More is Less: Understanding Chain-of-Thought Length in LLMsLLM reasoning → Prompting and chain-of-thought reasoning
Fairness in Agentic AI: A Unified Framework for Ethical and Equitable Multi-Agent SystemLLM agents and tool use → Social simulation and multi-agent debate
Scaling Online Complex Event Detection with Synthetic Supervision and Mamba-Based Neural Algorithmic ReasoningState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
CAT: Contrastive Adversarial Training for Evaluating the Robustness of Protective Perturbations in Latent Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects
Graph RAG-Tool FusionRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG +1
LLM Unlearning via Neural Activation RedirectionLLM safety and security → Machine unlearning
Improve the Training Efficiency of DRL for Wireless Communication Resource Allocation: The Role of Generative Diffusion ModelsDiffusion models → Diffusion for compression and wireless +1
Playmate: Flexible Control of Portrait Animation via 3D-Implicit Space Guided DiffusionDiffusion models → Audio, speech and music
Monte Carlo Tree Diffusion for System 2 PlanningLLM reasoning → Diffusion reasoning and test-time scaling +1
Provably Efficient Online RLHF with One-Pass Reward ModelingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Bag of Tricks for Inference-time Computation of LLM ReasoningLLM reasoning → Test-time compute and efficient thinking
Don't Just Demo, Teach Me the Principles: A Principle-Based Multi-Agent Prompting Strategy for Text ClassificationLLM agents and tool use → General LLM agent frameworks and benchmarks
Does Training on Synthetic Data Make Models Less Robust?Synthetic data and self-improvement for LLMs → LLM synthetic data generation
A Survey on Mamba Architecture for Vision ApplicationsState-space sequence models (S4, Mamba) → Vision Mamba backbones
HDCompression: Hybrid-Diffusion Image Compression for Ultra-Low BitratesDiffusion models → Diffusion for compression and wireless
Rethinking Fine-Tuning when Scaling Test-Time Compute: Limiting Confidence Improves Mathematical ReasoningLLM reasoning → Test-time compute and efficient thinking
Ask Patients with Patience: Enabling LLMs for Human-Centric Medical Dialogue with Grounded ReasoningLLM reasoning → Clinical and biomedical applications
MiniF2F in Rocq: Automatic Translation Between Proof Assistants -- A Case Study
Which Economic Tasks are Performed with AI? Evidence from Millions of Claude Conversations
DISCOVER: Identifying Patterns of Daily Living in Human Activities from Smart Home Data

All 385