A Theoretical Analysis of Optimistic Proximal Policy Optimization in Linear Markov Decision Processes.
Han ZhongTong ZhangPublished in: NeurIPS (2023)
Keyphrases
- markov decision processes
- theoretical analysis
- optimal policy
- policy iteration
- markov decision process
- finite horizon
- average reward
- infinite horizon
- state space
- average cost
- finite state
- state and action spaces
- reward function
- action space
- decision processes
- transition matrices
- partially observable
- reachability analysis
- dynamic programming
- decision problems
- factored mdps
- reinforcement learning
- policy evaluation
- stationary policies
- decision theoretic planning
- reinforcement learning algorithms
- state dependent
- discounted reward
- control policies
- markov decision problems
- policy iteration algorithm
- semi markov decision processes
- total reward
- multistage
- state abstraction
- expected reward
- sufficient conditions
- model based reinforcement learning
- continuous state spaces
- partially observable markov decision processes
- linear programming
- initial state
- multi agent
- model free
- machine learning
- planning under uncertainty