Modelling transition dynamics in MDPs with RKHS embeddings
Steffen GrünewälderGuy LeverLuca BaldassarreMassimiliano PontilArthur GrettonPublished in: CoRR (2012)
Keyphrases
- markov decision processes
- reproducing kernel hilbert space
- kernel methods
- reinforcement learning
- transition model
- dynamic model
- factored mdps
- dimensionality reduction
- dynamical systems
- state space
- agent based models
- euclidean space
- random projections
- manifold learning
- low dimensional
- gaussian process latent variable models
- vector space
- finite horizon
- state transitions
- machine learning
- domain adaptation
- kernel matrix
- partially observable
- hilbert space
- distance measure
- markov decision problems
- least squares
- decision theoretic planning
- special case