Random search for constrained Markov decision processes with multi-policy improvement.
Hyeong Soo ChangPublished in: Autom. (2015)
Keyphrases
- markov decision processes
- random search
- optimal policy
- policy iteration
- markov decision process
- finite horizon
- infinite horizon
- average reward
- partially observable
- average cost
- state space
- reward function
- state and action spaces
- simulated annealing
- decision processes
- action space
- finite state
- total reward
- search space
- genetic algorithm
- transition matrices
- reinforcement learning
- dynamic programming
- policy evaluation
- parameter optimization
- decision problems
- decision theoretic planning
- discounted reward
- partially observable markov decision processes
- expected reward
- markov decision problems
- stationary policies
- long run
- continuous state spaces
- sufficient conditions
- reinforcement learning algorithms
- action selection
- parameter estimation
- temporal difference