Reinforcement Learning
Conservative Offline Policy Adaptation in Multi-Agent Games
We prove that CSP learns a near-optimal risk-free offline adaptation policy upon convergence.
Unsupervised Behavior Extraction via Random Intent Priors
Reward-free data is abundant and contains rich prior knowledge of human behaviors, but it is not well exploited by offline reinforcement learning (RL) algorithms. In this paper, we propose UBER, an unsupervised approach to extract useful behaviors from offline reward-free datasets via diversified rewards.
9dc5accb1e4f4a9798eae145f2e4869b-Supplemental-Conference.pdf
Both challenges involve modeling long-term dependencies.
A Definition of Continual Reinforcement Learning
In a standard view of the reinforcement learning problem, an agent's goal is to efficiently identify a policy that maximizes long-term reward.
Generalized Weighted Path Consistency for Mastering Atari Games
Reinforcement learning with the help of neural-guided search consumes huge computational resources to achieve remarkable performance.
Importance Weighted Actor-Critic for Optimal Conservative Offline Reinforcement Learning
We provide both theoretical analysis and experimental results to validate the effectiveness of our proposed algorithm.
