Optimal Decision Tree Policies for Markov Decision Processes.
Daniël VosSicco VerwerPublished in: IJCAI (2023)
Keyphrases
- markov decision processes
- optimal policy
- average cost
- stationary policies
- total reward
- dynamic programming
- discounted reward
- decision trees
- average reward
- policy iteration algorithm
- finite horizon
- finite state
- markov decision process
- decision processes
- action sets
- state space
- reinforcement learning
- discount factor
- control policies
- policy iteration
- action space
- reward function
- expected reward
- transition matrices
- macro actions
- infinite horizon
- decentralized control
- reinforcement learning algorithms
- decision theoretic planning
- reachability analysis
- risk sensitive
- control policy
- planning under uncertainty
- partially observable
- partially observable markov decision processes
- optimal control
- decision problems
- markov decision problems
- continuous state spaces
- optimal solution
- finite number
- model based reinforcement learning
- linear programming
- multistage
- factored mdps
- state abstraction
- optimality criterion
- long run
- search algorithm
- stochastic shortest path