Tensor and Matrix Low-Rank Value-Function Approximation in Reinforcement Learning.
Sergio RozadaAntonio G. MarquesPublished in: CoRR (2022)
Keyphrases
- low rank
- trace norm
- reinforcement learning
- frobenius norm
- high order
- temporal difference
- state space
- tensor decomposition
- linear combination
- convex optimization
- singular value decomposition
- missing data
- matrix factorization
- basis functions
- matrix completion
- matrix decomposition
- semi supervised
- kernel matrix
- function approximation
- low rank matrix
- high dimensional data
- factorization methods
- reinforcement learning algorithms
- model free
- low rank matrices
- higher order
- rank minimization
- nuclear norm
- data matrix
- function approximators
- multi task
- markov decision processes
- optimal policy
- singular values
- tensor factorization
- norm minimization
- dimensionality reduction
- affinity matrix
- robust principal component analysis
- machine learning
- active learning
- pairwise