Li Shen

Compactness and Consistency: A Conjoint Framework for Deep Graph Clustering

Diffusion Language Model Knows the Answer Before It Decodes

GPTailor: Large Language Model Pruning Through Layer Cutting and Stitching

Convergent Differential Privacy Analysis for General Federated Learning

OptMerge: Unifying Multimodal LLM Capabilities and Modalities via Model Merging

EchoBench: Benchmarking Sycophancy in Medical Large Vision-Language Models

MaskPro: Linear-Space Probabilistic Learning for Strict (N:M)-Sparsity on LLMs

Understanding the Dynamics of Forgetting and Generalization in Continual Learning via the Neural Tangent Kernel

Joint Selection for Large-Scale Pre-Training Data via Policy Gradient-based Mask Learning

TrojanTO: Action-Level Backdoor Attacks Against Trajectory Optimization Models

Memory Efficient Fine-Tuning of LLMs via Forward-Only Hessian-Free Coordinate Descent

Reliable Poisoned Sample Detection against Backdoor Attacks Enhanced by Sharpness Aware Minimization

NBSP: A Neuron-Level Framework for Balancing Stability and Plasticity in Deep Reinforcement Learning

MergOPT: A Merge-Aware Optimizer for Robust Model Merging

Towards Optimism-Pessimism Trade-off in Model-based Offline-to-Online Reinforcement Learning

Qualitative and Quantitative Quality Assessment of Low-Light Enhanced Images: A Dataset and Benchmark Metric

The State of Reinforcement Finetuning for Transformer-based Agents

AdaGC: Improving Training Stability for Large Language Model Pretraining

UltraHorizon: Benchmarking LLM-Agent Capabilities in Ultra Long-Horizon Scenarios

Rewiring Experts on the Fly: Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert models

Merge to Remember: Sharpness-Aware Isotropic Merging for Continual Learning

LOST: Low-rank and Sparse Pre-training for Large Language Models

TeZO: Empowering the Low-Rankness on the Temporal Dimension in the Zeroth-Order Optimization for Fine-tuning LLMs

Model Unmerging: Making Your Models Unmergeable for Secure Model Sharing

SimReg: Achieving Higher Convergence and Generalization in the LLM Pretraining via Embedding Similarity Regularization

MAPLE: Masked Adapter Prototype Learning for OOD generalization

Mediater: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing

SeWA: Selective Weight Average via Probabilistic Masking

Stability and Generalization of Split Learning : Sequential or Federated

Q-learning Penalized Transformer for Safe Offline Reinforcement Learning

Stable-SPAM: How to Stably Train Large Language Models in 4-Bit

ELAS: Efficient Pre-Training of Low-Rank Large Language Models via 2:4 Activation Sparsity

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning

Branching Memory: Task-Specific Expansion for Continual Learning in Large Language Models

Stealthy Fine-Grained Editing Attack on MLLMs

Network Sparsity Unlocks the Scaling Potential of Deep Reinforcement Learning

Retrieval-Augmented Perception: High-resolution Image Perception Meets Visual RAG

Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Scheduler

Open-Vocabulary Customization from CLIP via Data-Free Knowledge Distillation

Combatting Dimensional Collapse in LLM Pre-Training Data via Submodular File Selection

RobustMerge: Parameter-Efficient Model Merging for MLLMs with Direction Robustness

NBSP: A Neuron-Level Framework for Balancing Stability and Plasticity in Deep Reinforcement Learning

Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More

Mulberry: Empowering MLLM with o1-like Reasoning and Reflection via Collective Monte Carlo Tree Search

Unveiling the Power of Multiple Gossip Steps: A Stability-Based Generalization Analysis in Decentralized Training

Enhancing Learning with Label Differential Privacy by Vector Approximation

Value-Guided Decision Transformer: A Unified Reinforcement Learning Framework for Online and Offline Settings

AlphaDecay: Module-wise Weight Decay for Heavy-Tailed Balancing in LLMs

Analytic Energy-Guided Policy Optimization for Offline Reinforcement Learning

Effective Policy Learning for Multi-Agent Online Coordination Beyond Submodular Objectives

Decision Mixer: Integrating Long-term and Local Dependencies via Dynamic Token Selection for Decision-Making

Self-Verification Provably Prevents Model Collapse in Recursive Synthetic Training

Mitigating the Backdoor Effect for Multi-Task Model Merging via Safety-Aware Subspace

Panacea: Mitigating Harmful Fine-tuning for Large Language Models via Post-fine-tuning Perturbation

Convergent Differential Privacy Analysis for General Federated Learning

Tackling Continual Offline RL through Selective Weights Activation on Aligned Spaces

CHPO: Constrained Hybrid-action Policy Optimization for Reinforcement Learning

Safety Reasoning with Guidelines

MixPrompt: Efficient Mixed Prompting for Multimodal Semantic Segmentation

Ada-R1: Hybrid-CoT via Bi-Level Adaptive Reasoning Optimization

Dynamic Neural Fortresses: An Adaptive Shield for Model Extraction Defense

Near-Optimal Online Learning for Multi-Agent Submodular Coordination: Tight Approximation and Communication Efficiency

Beyond Two-Stage Training: Integrating SFT and RL for Improved Reasoning in LLMs

Layer as Puzzle Pieces: Compressing Large Language Models through Layer Concatenation

Merging on the Fly Without Retraining: A Sequential Approach to Scalable Continual Model Merging

Mix Data or Merge Models? Balancing the Helpfulness, Honesty, and Harmlessness of Large Language Model via Model Merging

Robust Policy Expansion for Offline-to-Online RL under Diverse Data Corruption

RoMa: A Robust Model Watermarking Scheme for Protecting IP in Diffusion Models

R1-ShareVL: Incentivizing Reasoning Capabilities of Multimodal Large Language Models via Share-GRPO

PEARL: Towards Permutation-Resilient LLMs

Multinoulli Extension: A Lossless Yet Effective Probabilistic Framework for Subset Selection over Partition Constraints

Targeted Low-rank Refinement: Enhancing Sparse Language Models with Precision

Contextual Bandits for Unbounded Context Distributions

Continual Model Merging without Data: Dual Projections for Balancing Stability and Plasticity

Vad-R1: Towards Video Anomaly Reasoning via Perception-to-Cognition Chain-of-Thought

Efficient Federated Learning against Byzantine Attacks and Data Heterogeneity via Aggregating Normalized Gradients

Understanding the Stability-based Generalization of Personalized Federated Learning

Modeling Multi-Task Model Merging as Adaptive Projective Gradient Descent

GraphCL: Graph-based Clustering for Semi-Supervised Medical Image Segmentation

AlignIQL: Policy Alignment in Implicit Q-Learning through Constrained Optimization

LightSAM: Parameter-Agnostic Sharpness-Aware Minimization

FusionBench: A Comprehensive Benchmark of Deep Model Fusion

Towards Constraint-aware Learning for Resource Allocation in NFV-enabled Networks

Vulnerability-Aware Alignment: Mitigating Uneven Forgetting in Harmful Fine-Tuning

Mastering Massive Multi-Task Reinforcement Learning via Mixture-of-Expert Decision Transformer

Solving Continual Offline RL through Selective Weights Activation on Aligned Spaces

Continual Task Learning through Adaptive Policy Self-Composition

Exploring One-Shot Federated Learning by Model Inversion and Token Relabel with Vision Transformers

Learning with User-Level Local Differential Privacy

Targeted Low-rank Refinement: Enhancing Sparse Neural Networks with Precision

Continuous Spiking Graph ODE Networks

LoRA Recycle: Towards Fine-Tuning-Free Visual Foundation Model via Double-Efficient Data-Free Meta-Learning

Open-World Test-Time Training: Self-Training with Contrastive Learning

DynamicKV: Task-Aware Adaptive KV Cache Compression for Long Context LLMs

Memory Efficient Block Coordinate Descent Method for Forward-Only Second-Order Finetuning of LLM Models

Simultaneous Computation and Memory Efficient Zeroth-Order Optimizer for Fine-Tuning Large Language Models

Memory-Efficient Block Coordinate Descent for Hessian-Informed Zeroth-Order Optimizer

Towards Understanding Memory buffer based Continual Learning

AGLP: A Graph Learning Perspective for Semi-supervised Domain Adaptation

Language Model for Large-Text Transmission in Noisy Quantum Communications

SgCG: Semantic-guided Contrastive Generalization for Medical Image Segmentation