Evaluation of fast-convergence algorithm for ICA-based blind source separation of real convolutive mixture.
Hiroshi SaruwatariToshiya KawamuraKatsuyuki SawaiKiyohiro ShikanoAtsunobu KaminumaMasao SakataPublished in: EUSIPCO (2002)
Keyphrases
- blind source separation
- independent component analysis
- variable step size
- independent components
- convergence rate
- natural gradient
- mixing matrix
- k means
- source separation
- expectation maximization
- step size
- learning algorithm
- input data
- probabilistic model
- similarity measure
- particle swarm optimization
- algebraic equations
- cost function
- face recognition
- clustering method
- faster convergence
- blind separation
- data sets