Shao-Hua Sun

Implicit State Estimation via Video Replanning

SAD-Flower: Flow Matching for Safe, Admissible, and Dynamically Consistent Planning

CooT: Learning to Coordinate In-Context with Coordination Transformers

Learning Skills from Action-Free Videos

Split Decisions: VLM-Guided Action Sampling for Efficient RL Exploration

MedFuse: Multiplicative Embedding Fusion for Irregular Clinical Time Series

Synthesizing Programmatic Reinforcement Learning Policies with Large Language Model Guided Search

Efficient Action-Constrained Reinforcement Learning via Acceptance-Rejection Method and Augmented MDPs

QMP: Q-switch Mixture of Policies for Multi-Task Behavior Sharing

HERO: Human-Feedback Efficient Reinforcement Learning for Online Diffusion Model Finetuning

Mitigating Forgetting in LLM Fine-Tuning via Low-Perplexity Token Learning

Action-Constrained Imitation Learning

Action-Constrained Imitation Learning

Cross-Domain Reinforcement Learning via Preference Consistency