Peer Learning: Learning Complex Policies in Groups from Scratch via Action Recommendations

arXiv:2312.09950v2 Announce Type: replace
Abstract: Peer learning is a novel high-level reinforcement learning framework for agents learning in groups. While standard reinforcement learning trains an individual agent in trial-and-error fashion, all on its own, peer learning addresses a related setting in which a group of agents, i.e., peers, learns to master a task simultaneously together from scratch. Peers are allowed to communicate only about their own states and actions recommended by others: "What would you do in my situation?". Our …

abstract agent agents arxiv cs.ai cs.lg cs.ma error fashion framework novel peer policies recommendations reinforcement reinforcement learning scratch standard trains type via while