Polynomial Time and Sample Complexity for Non-Gaussian Component Analysis: Spectral Methods.
Yan Shuo TanRoman VershyninPublished in: COLT (2018)
Keyphrases
- component analysis
- sample complexity
- blind source separation
- special case
- factor analysis
- independent component analysis
- learning problems
- upper bound
- non stationary
- active learning
- learning algorithm
- spectral clustering
- lower bound
- supervised learning
- generalization error
- worst case
- sample size
- computational complexity
- learning process
- negative matrix factorization
- sparse representation
- cluster analysis
- matrix factorization
- cross validation
- speech signal
- training examples
- np hard
- training data
- image segmentation