Exponential Convergence and Stability of Howard's Policy Improvement Algorithm for Controlled Diffusions.
Bekzhan KerimkulovDavid SiskaLukasz SzpruchPublished in: SIAM J. Control. Optim. (2020)
Keyphrases
- computational cost
- experimental evaluation
- convergence rate
- computational complexity
- times faster
- objective function
- dynamic programming
- worst case
- optimization algorithm
- probabilistic model
- k means
- cost function
- stochastic approximation
- search space
- preprocessing
- iterative algorithms
- optimal solution
- rapid convergence
- convergence analysis
- linear complexity
- global convergence
- model free
- recognition algorithm
- expectation maximization
- high accuracy
- reinforcement learning
- state space
- significant improvement