Rank Selection in Nonnegative Matrix Factorization using Minimum Description Length.
Steven SquiresAdam Prügel-BennettMahesan NiranjanPublished in: Neural Comput. (2017)
Keyphrases
- nonnegative matrix factorization
- minimum description length
- probabilistic latent semantic indexing
- matrix factorization
- negative matrix factorization
- mdl principle
- information theoretic
- data representation
- nonnegative matrix
- principal component analysis
- spectral clustering
- least squares
- document clustering
- data sets
- original data
- mutual information
- digital libraries
- objective function
- keywords
- computer vision