Nonnegative Diagonals and High Performance on Low-Profile Matrices from Householder QR.
James DemmelMark HoemmenYozo HidaE. Jason RiedyPublished in: SIAM J. Sci. Comput. (2009)
Keyphrases
- singular value decomposition
- data matrix
- linear programming
- objective function
- scientific computing
- original data
- singular values
- high efficiency
- newly defined
- projection matrices
- tensor factorization
- graph embedding
- nonnegative matrix factorization
- discriminant analysis
- user profiles
- principal component analysis
- least squares
- pairwise