Regret-based Reward Elicitation for Markov Decision Processes.
Kevin ReganCraig BoutilierPublished in: UAI (2009)
Keyphrases
- markov decision processes
- reward function
- minimax regret
- total reward
- reinforcement learning algorithms
- expected reward
- reinforcement learning
- optimal policy
- state space
- partially observable
- average reward
- finite state
- decision theoretic planning
- transition matrices
- preference elicitation
- reachability analysis
- utility elicitation
- markov decision process
- policy iteration
- planning under uncertainty
- finite horizon
- dynamic programming
- factored mdps
- model based reinforcement learning
- state and action spaces
- stationary policies
- risk sensitive
- average cost
- machine learning
- discounted reward
- learning agent
- action space
- utility function
- infinite horizon
- markov chain
- markov decision problems