Optimistic Planning in Markov Decision Processes Using a Generative Model.
Balázs SzörényiGunnar KedenburgRémi MunosPublished in: NIPS (2014)
Keyphrases
- generative model
- markov decision processes
- macro actions
- planning under uncertainty
- decision theoretic planning
- partially observable
- finite state
- state space
- probabilistic model
- bayesian framework
- policy iteration
- reinforcement learning
- transition matrices
- prior knowledge
- optimal policy
- heuristic search
- planning problems
- partially observable markov decision processes
- dynamic programming
- posterior probability
- em algorithm
- reinforcement learning algorithms
- reachability analysis
- average cost
- probabilistic planning
- action sets
- ai planning
- reward function
- decision theoretic
- infinite horizon
- action space
- machine learning
- topic models
- average reward
- semi supervised
- action selection
- markov decision problems
- planning graph
- markov decision process
- initial state
- belief state
- expectation maximization
- training data