SPALS: Fast Alternating Least Squares via Implicit Leverage Scores Sampling.
Dehua ChengRichard PengYan LiuIoakeim PerrosPublished in: NIPS (2016)
Keyphrases
- alternating least squares
- low rank approximation
- negative matrix factorization
- nonnegative matrix factorization
- singular value decomposition
- low rank
- subspace learning
- stochastic gradient descent
- random sampling
- latent semantic indexing
- kernel matrix
- monte carlo
- spectral clustering
- matrix factorization
- machine learning
- iterative algorithms
- sample size