Censored Markov Decision Processes: A Framework for Safe Reinforcement Learning in Collaboration with External Systems.
Masahiro KohjimaMasami TakahashiHiroyuki TodaPublished in: CDC (2020)
Keyphrases
- markov decision processes
- reinforcement learning
- optimal policy
- decision theoretic planning
- state space
- finite horizon
- finite state
- reinforcement learning algorithms
- policy iteration
- decentralized control
- action space
- dynamic programming
- state and action spaces
- infinite horizon
- decision processes
- transition matrices
- markov games
- reachability analysis
- action sets
- state abstraction
- average cost
- partially observable
- stochastic games
- average reward
- multi agent
- markov decision process
- model based reinforcement learning
- continuous state spaces
- markov decision problems
- data mining
- multistage