Model-based multi-agent RL in zero-sum Markov games with near-optimal sample complexity

Keyphrases

• Action Space 25%
• Discount Factor 25%
• Equilibrium Policies 25%
• Equilibrium Value 25%
• Generative Models 25%
• Learning Settings 25%
• Minimax Optimal 25%
• Model Complexity 50%
• Model-based Approach 25%
• Model-based Reinforcement Learning 25%
• Multi-agent Reinforcement Learning 100%
• Multi-agent RL 100%
• Nash Equilibrium 50%
• Near-optimal 100%
• Non-stationarity 25%
• Optimal Policy 25%
• Optimal Sample Complexity 100%
• Planning Phase 25%
• Reinforcement Learning 25%
• RL Algorithm 25%
• Sample Complexity 75%
• State Space 25%
• State Transition 25%
• Tight 25%
• Two-agent 25%
• Zero-sum Markov Game 100%

Mathematics

• Action Space 50%
• Discount Factor 50%
• Equilibrium Value 50%
• Minimax 50%
• Nash Equilibrium 100%
• Open Question 50%
• Optimal Policy 50%
• Simple Model 50%
• State Transition 50%
• Stationarity 50%

Computer Science

• Generative Model 25%
• Model-Based Reinforcement Learning 25%
• multi agent 100%
• Multi-Agent Reinforcement Learning 100%
• Nash Equilibrium 50%
• Planning Phase 25%
• Reinforcement Learning 25%
• State Space 25%
• State Transition 25%