Narrowing the Gap between Adversarial and Stochastic MDPs via Policy Optimization.
Daniil TiapkinEvgenii ChzhenGilles StoltzPublished in: CoRR (2024)
Keyphrases
- optimal policy
- markov decision processes
- stochastic optimization
- markov decision process
- control policies
- finite horizon
- infinite horizon
- policy iteration
- stochastic search
- policy iteration algorithm
- stochastic programming
- markov decision problems
- optimization algorithm
- reinforcement learning
- continuous state spaces
- policy search
- state and action spaces
- reward function
- partially observable
- finite state
- optimization method
- action space
- average reward
- decision problems
- linear programming
- optimization problems
- state space
- reinforcement learning problems
- semi markov decision processes
- partially observable markov decision processes
- policy evaluation
- robust optimization
- average cost
- monte carlo
- dynamic programming