Policy Search for the Optimal Control of Markov Decision Processes: A Novel Particle-Based Iterative Scheme.
Giorgio ManganiniMatteo PirottaMarcello RestelliLuigi PiroddiMaria PrandiniPublished in: IEEE Trans. Cybern. (2016)
Keyphrases
- optimal control
- markov decision processes
- reinforcement learning
- risk sensitive
- dynamic programming
- infinite horizon
- policy iteration
- average cost
- reinforcement learning algorithms
- control problems
- optimal policy
- state space
- continuous state
- finite state
- policy gradient
- partially observable markov decision processes
- markov decision problems
- average reward
- partially observable
- function approximation
- reward function
- action space
- control strategy
- finite horizon
- model free
- multistage
- monte carlo
- temporal difference
- decision processes
- finite number
- decision problems