Reinforcement Learning for Cost-Aware Markov Decision Processes.
Wesley SuttleKaiqing ZhangZhuoran YangJi LiuDavid KraemerPublished in: ICML (2021)
Keyphrases
- markov decision processes
- reinforcement learning
- average cost
- optimal policy
- reinforcement learning algorithms
- state space
- policy iteration
- finite state
- partially observable
- transition matrices
- decision theoretic planning
- planning under uncertainty
- dynamic programming
- finite horizon
- state and action spaces
- action space
- model based reinforcement learning
- infinite horizon
- reachability analysis
- markov decision process
- state abstraction
- reward function
- factored mdps
- stochastic games
- total cost
- action sets
- continuous state spaces
- learning algorithm
- average reward
- optimal control
- function approximation
- approximate dynamic programming
- expected cost
- model free
- semi markov decision processes
- linear programming
- machine learning
- real time dynamic programming