Variational policy gradient method for reinforcement learning with general utilities

Keyphrases

• General Utility 100%
• Reinforcement Learning 100%
• Variational 100%
• Policy Gradient 100%
• Policy Gradient Methods 100%
• Convergence Rate 66%
• Policy Gradient Theorem 66%
• Cumulative Reward 66%
• Optimization Problem 33%
• Utility Function 33%
• Non-convexity 33%
• Bellman Equation 33%
• Gradient-based 33%
• Estimation Algorithms 33%
• Parametrized 33%
• Dynamic Programming 33%
• State Action 33%
• Hidden Convexity 33%
• State Function 33%
• Optimal Policy 33%
• Variational Monte Carlo 33%
• Strong Convexity 33%
• Cumulative Sum 33%
• Direct Policy Search 33%
• Policy Optimization 33%
• Fenchel Duality 33%
• Reinforcement Learning Problems 33%
• Gradient Schemes 33%
• Constrained Problems 33%
• Markov Decision Problem 33%
• Occupancy Measure 33%
• Problem Exploration 33%
• Monte Carlo Gradient Estimation 33%
• Stochastic Saddle Point Problems 33%
• General Objectives 33%
• Sample Path 33%
• Reinforcement Learning System 33%
• Concave Utility Function 33%

Mathematics

• Utility Function 100%
• Saddle Point 50%
• Stochastics 50%
• Monte Carlo 50%
• Generality 50%
• Bellman Equation 50%
• Convergence Rate 50%
• Sample Path 50%
• Optimal Policy 50%
• Cumulative Sum 50%
• Rate of Convergence 50%
• Dynamic Programming 50%