Deriving Explicit Control Policies for Markov Decision Processes Using Symbolic Regression.
A. HristovJoost W. BosmanSandjai BhulaiRobert D. van der MeiPublished in: VALUETOOLS (2020)
Keyphrases
- control policies
- markov decision processes
- symbolic regression
- optimal policy
- action space
- genetic programming
- finite horizon
- reinforcement learning
- state space
- reward function
- continuous state
- finite state
- dynamic programming
- infinite horizon
- evolutionary computation
- markov decision process
- average cost
- decision problems
- regression problems
- long run
- initial state
- evolutionary algorithm
- motion control
- control policy
- multistage
- sufficient conditions
- average reward
- maximum likelihood
- machine learning
- decision making
- learning algorithm