Robot Learning: Multi-Agent Reinforcement Learning and RLHF

This lecture explores the foundations of multi-agent reinforcement learning (MARL) and its connection to reinforcement learning from human feedback (RLHF). Learn how robots can develop action capabilities while coexisting with other agents in their environment. Discover structured learning approaches that reduce MARL complexity, compare centralized versus decentralized learning algorithms, and understand methods for incorporating online human feedback. The presentation also covers techniques for training large language models using human preference data through RLHF. Presented by Montreal Robotics, this comprehensive exploration provides essential knowledge for understanding how multiple agents can learn collaboratively and how human feedback can guide reinforcement learning systems.