2 papers
cs.LG2025
Uncertainty-Based Smooth Policy Regularisation for Reinforcement Learning with Few Demonstrations
Yujie Zhu, Charles A. Hepburn, Matthew Thorpe +1
In reinforcement learning with sparse rewards, demonstrations can accelerate learning, but determining when to imitate them remains challenging. We propose Smooth Policy Regularisa…
stat.ML2025
State-Constrained Offline Reinforcement Learning
Charles A. Hepburn, Yue Jin, Giovanni Montana
Traditional offline reinforcement learning (RL) methods predominantly operate in a batch-constrained setting. This confines the algorithms to a specific state-action distribution p…