Policy gradient stochastic approximation algorithms for adaptive control of constrained time varying Markov decision processes.
Felisa J. Vázquez-AbadVikram KrishnamurthyPublished in: CDC (2003)
Keyphrases
- approximation algorithms
- adaptive control
- markov decision processes
- policy gradient
- reinforcement learning
- reinforcement learning algorithms
- average reward
- actor critic
- np hard
- optimal policy
- state space
- special case
- policy iteration
- function approximation
- control problems
- finite state
- worst case
- stochastic games
- control method
- optimal control
- dynamic programming
- partially observable markov decision processes
- model free
- monte carlo
- multi agent
- infinite horizon
- action space
- state action
- temporal difference
- learning algorithm
- machine learning
- learning problems
- optimal solution
- function approximators
- markov decision process
- single agent
- reward function
- lower bound
- linear programming
- dynamic environments
- long run
- linear program
- rl algorithms
- neural network
- decision problems