Online Gaussian Process State-Space Models: Learning and Planning for Partially Observable Dynamical Systems.
Soon-Seo ParkYoung-Jin ParkHan-Lim ChoiPublished in: CoRR (2019)
Keyphrases
- partially observable
- dynamical systems
- dynamical models
- state space
- predictive state representations
- gaussian process
- reinforcement learning
- dynamic systems
- markov decision processes
- markov decision problems
- heuristic search
- gaussian processes
- decision problems
- belief state
- infinite horizon
- planning problems
- model selection
- partially observable markov decision processes
- bayesian framework
- dynamic programming
- learning tasks
- reward function
- semi supervised
- regression model
- data mining
- markov chain
- hyperparameters
- active learning
- prior knowledge
- latent variable models
- pairwise
- search algorithm
- bayesian networks