Optimality and Approximation with Policy Gradient Methods in Markov Decision Processes

Mathematics

• Action Space 25%
• Approximates 25%
• Approximation Error 25%
• Approximation Function 50%
• Asymptotics 25%
• Convergence Property 25%
• Convergence Rate 50%
• Function Value 50%
• Global Optimum 50%
• Initial State 50%
• Main Result 25%
• Markov Decision Process 100%
• Optimal Policy 100%
• Optimality 100%
• Parametric 25%
• Polynomial 25%
• State Distribution 50%
• Worst Case 50%

Keyphrases

• Compatibility Function 25%
• Inexact Gradient 25%
• Polynomial Convergence Rate 25%