Learning Environment Properties in Partially Observable Monte Carlo Planning.
Maddalena ZuccottoAlberto CastelliniMarco PiccinelliEnrico MarchesiniAlessandro FarinelliPublished in: AIRO@AI*IA (2021)
Keyphrases
- monte carlo
- partially observable
- learning environment
- state space
- partial observability
- dynamical systems
- reinforcement learning
- decision problems
- markov decision processes
- markov chain
- particle filter
- partial observations
- monte carlo simulation
- belief space
- belief state
- importance sampling
- infinite horizon
- markov decision problems
- planning problems
- optimal strategy
- reward function
- monte carlo tree search
- machine learning
- temporal difference
- optimal policy
- learning algorithm
- multi agent
- dynamic programming
- variance reduction
- partially observable markov decision processes
- single agent
- kalman filter
- planning domains