Federated Reinforcement Learning with Constraint Heterogeneity

arXiv:2405.03236v1 Announce Type: new
Abstract: We study a Federated Reinforcement Learning (FedRL) problem with constraint heterogeneity. In our setting, we aim to solve a reinforcement learning problem with multiple constraints while $N$ training agents are located in $N$ different environments with limited access to the constraint signals and they are expected to collaboratively learn a policy satisfying all constraint signals. Such learning problems are prevalent in scenarios of Large Language Model (LLM) fine-tuning and healthcare applications. To solve the problem, …

abstract access agents aim arxiv constraints cs.lg environments learn multiple reinforcement reinforcement learning solve stat.ml study training type while