Advantage based value iteration for Markov decision processes with unknown rewards.
Pegah AlizadehYann ChevaleyreFrançois LévyPublished in: IJCNN (2016)
Keyphrases
- markov decision processes
- finite state
- reinforcement learning
- state space
- optimal policy
- dynamic programming
- planning under uncertainty
- policy iteration
- reinforcement learning algorithms
- transition matrices
- average reward
- infinite horizon
- decision theoretic planning
- finite horizon
- reachability analysis
- discounted reward
- action space
- average cost
- total reward
- reward function
- sequential decision making under uncertainty
- markov decision process
- factored mdps
- partially observable
- model based reinforcement learning
- risk sensitive