Constrained Reinforcement Learning Under Model Mismatch

arXiv:2405.01327v1 Announce Type: new
Abstract: Existing studies on constrained reinforcement learning (RL) may obtain a well-performing policy in the training environment. However, when deployed in a real environment, it may easily violate constraints that were originally satisfied during training because there might be model mismatch between the training and real environments. To address the above challenge, we formulate the problem as constrained RL under model uncertainty, where the goal is to learn a good policy that optimizes the reward and …

abstract arxiv constraints cs.lg environment environments however policy reinforcement reinforcement learning studies training type