Safe Exploration in Reinforcement Learning by Reachability Analysis over Learned Models.
Yuning WangHe ZhuPublished in: CAV (3) (2024)
Keyphrases
- reachability analysis
- learned models
- markov decision processes
- reinforcement learning
- state space
- learning algorithm
- exploration strategy
- active exploration
- model checking
- optimal policy
- action selection
- generative model
- reinforcement learning algorithms
- training data
- function approximation
- incremental algorithms
- dynamic programming
- timed automata
- partially observable
- markov decision process
- classification models
- action space
- policy iteration
- machine learning
- temporal difference
- model free
- infinite horizon
- dynamical systems
- em algorithm
- simulated annealing
- support vector machine