Reinforcement Learning Transfer via Common Subspaces.
Haitham Bou-AmmarMatthew E. TaylorPublished in: ALA (2011)
Keyphrases
- reinforcement learning
- transfer learning
- knowledge transfer
- function approximation
- mutual subspace method
- neural network
- state space
- learning algorithm
- transferring knowledge
- reinforcement learning algorithms
- cross domain
- low dimensional
- high dimensional data
- model free
- parameter space
- learning process
- support vector
- previously learned
- linear subspace
- high dimensional
- dynamic programming
- principal component analysis
- least squares
- dimensionality reduction
- original data
- action selection
- lower dimensional
- learning problems
- learning tasks
- robot control
- subspace clusters
- policy search
- optimal policy
- multi agent