Learning unions of orthonormal bases with thresholded singular value decomposition.
Sylvain LesageRémi GribonvalFrédéric BimbotLaurent BenaroyaPublished in: ICASSP (5) (2005)
Keyphrases
- singular value decomposition
- singular values
- least squares
- reconstruction error
- basis functions
- dimension reduction
- dimensionality reduction
- latent semantic indexing
- low rank
- principal component analysis
- low rank matrix
- low rank approximation
- correlation matrix
- singular vectors
- linear combination
- feature space
- latent semantic analysis
- data matrix
- null space