Towards Painless Policy Optimization for Constrained MDPs.
Arushi JainSharan VaswaniReza BabanezhadCsaba SzepesváriDoina PrecupPublished in: CoRR (2022)
Keyphrases
- optimal policy
- markov decision processes
- concave convex procedure
- markov decision process
- markov decision problems
- finite horizon
- partially observable
- reinforcement learning problems
- policy iteration
- optimization algorithm
- optimization process
- optimization method
- policy search
- state space
- reinforcement learning
- average reward
- dynamic programming
- constrained optimization
- state and action spaces
- optimization problems
- infinite horizon
- global optimization
- average cost
- reward function
- action space
- saddle point
- discounted reward
- evolutionary algorithm