Weighted nonnegative tensor factorization: on monotonicity of multiplicative update rules and application to user-guided audio source separation - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports (Research Report) Year : 2013

Weighted nonnegative tensor factorization: on monotonicity of multiplicative update rules and application to user-guided audio source separation

Alexey Ozerov
  • Function : Author
  • PersonId : 930358
Ngoc Q. K. Duong
  • Function : Author
  • PersonId : 946470
Louis Chevallier
  • Function : Author
  • PersonId : 940855

Abstract

This report focuses on so-called weighted variants of nonnnegative matrix factorization (NMF) and more generally nonnnegative tensor factorization (NTF) approximations. First, we consider multiplicative update (MU) rules to optimize these approximations, and we prove that some results on monotonicity of MU rules for NMF generalize without restrictions to both the NTF and the WNTF cases. Second, we propose new weighting strategies for an existing NTF-based user-guided audio source separation method. Experimental evaluation shows that these weightings lead to better source separation than the same model without using the weights. The best configuration of the proposed method was entered into the fourth community-based Signal Separation Evaluation Campaign (SiSEC 2013).
Fichier principal
Vignette du fichier
Ozerov_et_al_TechRepTEC_2013.pdf (210.04 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-00878685 , version 1 (30-10-2013)
hal-00878685 , version 2 (21-03-2014)

Identifiers

  • HAL Id : hal-00878685 , version 2

Cite

Alexey Ozerov, Ngoc Q. K. Duong, Louis Chevallier. Weighted nonnegative tensor factorization: on monotonicity of multiplicative update rules and application to user-guided audio source separation. [Research Report] 2013, pp.10. ⟨hal-00878685v2⟩

Collections

LARA
244 View
344 Download

Share

Gmail Facebook X LinkedIn More