Optimality and Approximation with Policy Gradient Methods in Markov Decision Processes.
Alekh AgarwalSham M. KakadeJason D. LeeGaurav MahajanPublished in: CoRR (2019)
Keyphrases
- markov decision processes
- average cost
- average reward
- policy gradient methods
- policy gradient
- stationary policies
- reinforcement learning
- state space
- optimal policy
- finite state
- policy iteration
- natural actor critic
- actor critic
- infinite horizon
- dynamic programming
- markov decision process
- partially observable
- reinforcement learning algorithms
- decision processes
- approximate dynamic programming
- optimal solution
- real valued
- action space
- stochastic games
- linear program
- machine learning
- initial state
- partially observable markov decision processes
- long run
- function approximation
- decision problems