Fast Global Convergence of Policy Optimization for Constrained MDPs.
Tao LiuRuida ZhouDileep KalathilP. R. KumarChao TianPublished in: CoRR (2021)
Keyphrases
- global convergence
- optimization methods
- global optimum
- optimal policy
- convergence analysis
- convergence speed
- convergence rate
- markov decision processes
- constrained optimization problems
- convex minimization
- markov decision process
- line search
- particle swarm
- finite horizon
- optimization method
- optimization problems
- state space
- policy search
- markov decision problems
- reinforcement learning
- state and action spaces
- policy iteration
- reward function
- average cost
- optimal solution
- partially observable markov decision processes
- action space
- partially observable
- infinite horizon
- particle swarm optimization algorithm
- step size
- newton method
- particle swarm optimization
- globally convergent
- dynamic programming