Multi-Agent Reinforcement Learning

• From a Game Theoretic perspective, we can compute various solution concepts in the stochastic games that MARL captures. However, computation of these solution concepts is known to be NP-hard. Any MARL algorithm will be exponentially hard in the worst case.
  • This necessitates exploiting structures or assumptions (inductive biases) about certain game types .

Research §

• Current (2023) trends suggest that mixed play algorithms more or less perform the same.

Topics §

• MARL — Notation Guide
• MARL from a Game Theoretic Perspective
• MARL Problem Statement
• MARL Algorithms and Approaches
• MARL Research

Links §

• Reinforcement Learning
• Reinforcement Learning — Industrial Applications of Intelligent Agents by Winder
• Multi-Agent Reinforcement Learning — Foundations and Modern Approaches by Albrecht, Christianos and Schafer