Exploiting policy structure for solving MDPs with large state space.
Libin LiuArpan ChattopadhyayUrbashi MitraPublished in: CISS (2018)
Keyphrases
- state space
- markov decision problems
- optimal policy
- markov decision processes
- markov decision process
- reinforcement learning
- action space
- partially observable
- continuous state spaces
- heuristic search
- reward function
- linear programming
- factored markov decision processes
- factored mdps
- dynamic programming
- infinite horizon
- average cost
- policy iteration
- finite horizon
- markov chain
- state variables
- finite state
- state and action spaces
- domain independent
- particle filter
- reinforcement learning problems
- machine learning
- sequential decision making problems
- belief state
- policy search
- dec pomdps
- stationary policies
- decision processes
- initial state
- semi markov decision processes
- reinforcement learning algorithms