On-line Reinforcement Learning Using Incremental Kernel-Based Stochastic Factorization.
André da Motta Salles BarretoDoina PrecupJoelle PineauPublished in: NIPS (2012)
Keyphrases
- reinforcement learning
- direct policy search
- stochastic approximation
- learning automata
- batch mode
- control policies
- incremental learning
- reinforcement learning algorithms
- function approximation
- model free
- state space
- kernel methods
- temporal difference learning
- monte carlo
- machine learning
- continuous state
- robotic control
- stochastic optimization
- kronecker product
- multi agent
- control problems
- temporal difference
- singular value decomposition
- markov decision processes
- support vector
- kernel pca
- active learning
- dynamic programming
- incremental version
- policy search
- continuous state spaces
- approximate dynamic programming
- support vector machine
- tensor factorization
- supervised learning
- reinforcement learning methods
- optimal control
- control policy
- action space
- multibody