Divergence optimization in nonnegative matrix factorization with spectrogram restoration for multichannel signal separation.
Daichi KitamuraHiroshi SaruwatariSatoshi NakamuraYu TakahashiKazunobu KondoHirokazu KameokaPublished in: HSCMA (2014)
Keyphrases
- nonnegative matrix factorization
- single channel
- data representation
- probabilistic latent semantic indexing
- multi channel
- negative matrix factorization
- nonnegative matrix
- wiener filter
- matrix factorization
- frequency domain
- least squares
- image restoration
- signal processing
- image processing
- principal component analysis
- sound source
- signal reconstruction
- objective function
- document clustering
- spectral clustering
- machine learning
- nearest neighbor
- active learning
- relational databases
- pattern recognition
- co occurrence