Explaining Reward Functions in Markov Decision Processes.
Jacob RussellEugene SantosPublished in: FLAIRS Conference (2019)
Keyphrases
- markov decision processes
- reward function
- state space
- optimal policy
- reinforcement learning algorithms
- partially observable
- reinforcement learning
- finite state
- dynamic programming
- transition matrices
- policy iteration
- planning under uncertainty
- factored mdps
- markov decision process
- finite horizon
- average reward
- average cost
- decision theoretic planning
- hierarchical reinforcement learning
- decision processes
- stochastic games
- infinite horizon
- action space
- control policies
- learning algorithm