Distance-Penalized Active Learning via Markov Decision Processes.
Dingyu WangJohn LiporGautam DasarathyPublished in: DSW (2019)
Keyphrases
- markov decision processes
- active learning
- state space
- finite state
- optimal policy
- policy iteration
- dynamic programming
- reinforcement learning
- decision theoretic planning
- transition matrices
- reachability analysis
- average reward
- partially observable
- least squares
- risk sensitive
- infinite horizon
- factored mdps
- reinforcement learning algorithms
- supervised learning
- finite horizon
- decision processes
- model based reinforcement learning
- average cost
- markov decision process
- semi supervised
- machine learning
- planning under uncertainty
- learning process
- training set
- action space
- action sets
- state abstraction
- reward function
- state and action spaces
- loss function
- model checking