Solving Markov Decision Processes with Reachability Characterization from Mean First Passage Times.
Shoubhik DebnathLantao LiuGaurav S. SukhatmePublished in: CoRR (2019)
Keyphrases
- markov decision processes
- state space
- transition matrices
- semi markov decision processes
- finite state
- markov decision problems
- optimal policy
- reinforcement learning
- policy iteration algorithm
- policy iteration
- dynamic programming
- decision processes
- markov decision process
- partially observable
- stochastic shortest path
- reachability analysis
- decision theoretic planning
- average reward
- finite horizon
- reinforcement learning algorithms
- factored mdps
- action space
- state and action spaces
- planning under uncertainty
- risk sensitive
- state abstraction
- reward function
- infinite horizon
- heuristic search
- average cost
- model based reinforcement learning
- solving planning problems
- probabilistic planning
- action sets
- dynamical systems