A Duality Approach for Regret Minimization in Average-Reward Ergodic Markov Decision Processes

Mathematics

• Saddle Point 100%
• Time Step 100%
• Markov Decision Process 100%
• Worst Case 50%
• Action Space 50%
• Ergodicity 50%
• Min-Max 50%
• Duality Gap 50%
• S-Spaces 50%

Computer Science

• Markov Decision Process 100%
• State Space 50%
• Primal-Dual 50%
• Learning Agent 50%

Keyphrases

• Average Reward Markov Decision Processes 33%