Deheng Ye

Multi-objective Large Language Model Alignment with Hierarchical Experts

Powerful Independence Testing on Heterogeneous Federated Clients with Theoretical Guarantees

Controlling Video Generation with Vision Language Models

Uncertainty-Aware Tree Search for Efficient LLM Reasoning

Leading the Follower: Learning Persuasive Agents in Social Deduction Games

From Coefficients to Directions: Rethinking Model Merging with Directional Alignment

Cross-Domain Offline Policy Adaptation via Selective Transition Correction

Pseudo-outcome Imputation with Post-treatment Variables for Individual Treatment Effect Estimation

Mini-Omni-Reasoner: Token-Level Thinking-in-Speaking in Large Speech Models

AdaptVision: Efficient Vision-Language Models via Adaptive Visual Acquisition

PROF: An LLM-based Reward Code Preference Optimization Framework for Offline Imitation Learning

Merging without Forgetting: Continual Fusion of Task-Specific Models via Optimal Transport

Temporal Difference Learning with Constrained Initial Representations

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

OpenMMEgo: Enhancing Egocentric Understanding for LMMs with Open Weights and Data

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

WALL-E: World Alignment by NeuroSymbolic Learning improves World Model-based LLM Agents

Cross-Domain Offline Policy Adaptation with Optimal Transport and Dataset Constraint

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

Ad Hoc Teamwork via Offline Goal-Based Decision Transformers

WALL-E: World Alignment by Rule Learning Improves World Model-based LLM Agents