K-means recovers ICA filters when independent components are sparse.
Alon VinnikovShai Shalev-ShwartzPublished in: ICML (2014)
Keyphrases
- independent components
- independent component analysis
- mixing matrix
- k means
- independent components analysis
- blind source separation
- factor analysis
- principal components
- principal component analysis
- clustering algorithm
- signal processing
- sparse representation
- feature space
- cluster analysis
- feature vectors
- high dimensional
- clustering method
- spectral clustering
- filter responses
- data clustering
- feature extraction
- document clustering
- expectation maximization
- image classification
- edge detection
- principle component analysis
- source separation