Conference

We propose a principled exploration method for DRL through Optimistic Bootstrapping and Backward Induction (OB2I).

We propose a Dynamic Bottleneck (DB) model, which attains a dynamics-relevant representation based on the information-bottleneck principle.

We propose Pessimistic Bootstrapping for offline RL (PBRL), a purely uncertainty-driven offline algorithm without explicit policy constraints.

We study how RL can be empowered by contrastive learning in a class of Markov decision processes (MDPs) and Markov games (MGs) with low-rank transitions. For both models, we propose to extract the correct feature representations of the low-rank model by minimizing a contrastive loss.

We propose Robust Offline Reinforcement Learning (RORL) with a novel conservative smoothing technique.

We propose falSe COrrelation REduction (SCORE) for offline RL, a practically effective and theoretically provable algorithm.

We propose a novel unsupervised skill discovery method through contrastive learning among behaviors, which makes the agent produce similar behaviors for the same skill and diverse behaviors for different skills.

We reveal the limitations of these methods and explore the problem from the value difference perspective via a novel insight on the value consistency across domains.

We aim to investigate the effectiveness of a single diffusion model in modeling large-scale multi-task offline data, which can be challenging due to diverse and multimodal data distribution.

We propose Optimistic Value Distribution Explorer (OVD-Explorer) to achieve a noise-aware optimistic exploration for continuous control.