Optimistic planning for belief-augmented Markov Decision Processes.
Raphaël FonteneauLucian BusoniuRémi MunosPublished in: ADPRL (2013)
Keyphrases
- markov decision processes
- macro actions
- partially observable
- planning under uncertainty
- decision theoretic planning
- state space
- finite state
- optimal policy
- reinforcement learning
- belief state
- policy iteration
- transition matrices
- probabilistic planning
- average cost
- partially observable markov decision processes
- infinite horizon
- dynamic programming
- risk sensitive
- planning problems
- average reward
- heuristic search
- markov decision process
- reachability analysis
- model based reinforcement learning
- decision processes
- reinforcement learning algorithms
- action space
- finite horizon
- decision theoretic
- factored mdps
- state and action spaces
- belief revision
- action sets
- multistage
- decision problems
- markov decision problems
- blocks world
- ai planning
- discounted reward
- semi markov decision processes