Explore advanced concepts in multi-agent reinforcement learning in this one-hour lecture from the Learning and Games Boot Camp. Delve into topics such as crosscourt equilibrium, learning rewards, Q-value estimation, no-regret learning, and Nash equilibrium. Examine the differences between reinforcement learning and supervised learning, and investigate challenges in multi-agent settings. Learn about adversarial bandits, linear Markov games, and partial operability. Gain insights from Princeton University's Chi Jin on cutting-edge techniques and advanced topics in this complex field of artificial intelligence and game theory.
Multi-Agent Reinforcement Learning - Part II
Overview
Syllabus
Intro
Similar Setting
Interaction Protocol
Reinforcement Learning vs Supervised Learning
Crosscourt Equilibrium
Learning Rewards
Main Techniques
Challenges
Qvalue
No Regret Learning
Vlearn
Adversarial Bandit
General Examples
Nash Equilibrium
Product of AI
No Swap Regret
Advanced Topics
Linear Markov Game
Partial Operability
Recap
Taught by
Simons Institute
Related Courses
-
Multi-Agent Reinforcement Learning - Part I
-
Decision Making and Reinforcement Learning
-
V-Learning - Simple, Efficient, Decentralized Algorithm for Multiagent RL
-
Strategic Deviations in Multi-Agent Imitation Learning: From Value Optimization to Regret Analysis
-
Independent Learning Dynamics for Stochastic Games - Where Game Theory Meets
-
Multi-Agent AI Systems: From Imitation Learning to Markov Game Theory
