Rule-based Shielding for Partially Observable Monte-Carlo Planning.
Giulio MazziAlberto CastelliniAlessandro FarinelliPublished in: ICAPS (2021)
Keyphrases
- monte carlo
- partially observable
- state space
- dynamical systems
- reinforcement learning
- markov chain
- markov decision processes
- decision problems
- partial observability
- markov decision problems
- importance sampling
- planning domains
- infinite horizon
- monte carlo simulation
- belief state
- belief space
- planning problems
- particle filter
- partial observations
- transition probabilities
- reward function
- partially observable markov decision processes
- optimal policy
- optimal strategy
- machine learning
- game tree
- monte carlo tree search
- domain independent