SuperLectures.com

MAJORIZATION-MINIMIZATION ALGORITHM FOR SMOOTH ITAKURA-SAITO NONNEGATIVE MATRIX FACTORIZATION

Non-negative Tensor Factorization and Blind Separation

Full Paper at IEEE Xplore

Presented by: Cédric Févotte, Author(s): Cédric Févotte, CNRS LTCI / Télécom ParisTech, France

Nonnegative matrix factorization (NMF) with the Itakura-Saito divergence has proven efficient for audio source separation and music transcription, where the signal power spectrogram is factored into a ``dictionary'' matrix times an ``activation'' matrix. Given the nature of audio signals it is expected that the activation coefficients exhibit smoothness along time frames. This may be enforced by penalizing the NMF objective function with an extra term reflecting smoothness of the activation coefficients. We propose a novel regularization term that solves some deficiencies of our previous work and leads to an efficient implementation using a majorization-minimization procedure.


  Speech Transcript

|

  Slides

Enlarge the slide | Show all slides in a pop-up window

0:00:16

  1. slide

0:00:24

  2. slide

0:01:04

  3. slide

0:01:50

  4. slide

0:05:14

  5. slide

0:06:06

  6. slide

0:08:49

  7. slide

0:09:38

  8. slide

0:10:42

  9. slide

0:11:36

 10. slide

0:12:25

 11. slide

0:13:28

 12. slide

0:14:23

 13. slide

0:15:05

 14. slide

0:15:54

 15. slide

0:17:09

 16. slide

0:18:16

 17. slide

0:18:41

 18. slide

0:18:58

 19. slide

  Comments

Please sign in to post your comment!

  Lecture Information

Recorded: 2011-05-26 16:35 - 16:55, Club B
Added: 21. 6. 2011 17:20
Number of views: 123
Video resolution: 1024x576 px, 512x288 px
Video length: 0:23:19
Audio track: MP3 [7.90 MB], 0:23:19