Novel alternating least squares algorithm for nonnegative matrix and tensor factorizations

Anh Huy Phan, Andrzej Cichocki, Rafal Zdunek, Thanh Vu Dinh

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Citations (Scopus)

Abstract

Alternative least squares (ALS) algorithm is considered as a "work-horse" algorithm for general tensor factorizations. For nonnegative tensor factorizations (NTF), we usually use a nonlinear projection (rectifier) to remove negative entries during the iteration process. However, this kind of ALS algorithm often fails and cannot converge to the desired solution. In this paper, we proposed a novel algorithm for NTF by recursively solving nonnegative quadratic programming problems. The validity and high performance of the proposed algorithm has been confirmed for difficult benchmarks, and also in an application of object classification.

Original languageEnglish
Title of host publicationNeural Information Processing
Subtitle of host publicationTheory and Algorithms - 17th International Conference, ICONIP 2010, Proceedings
Pages262-269
Number of pages8
EditionPART 1
DOIs
Publication statusPublished - 2010
Externally publishedYes
Event17th International Conference on Neural Information Processing, ICONIP 2010 - Sydney, NSW, Australia
Duration: 22 Nov 201025 Nov 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume6443 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th International Conference on Neural Information Processing, ICONIP 2010
Country/TerritoryAustralia
CitySydney, NSW
Period22/11/1025/11/10

Keywords

  • ALS
  • NMF
  • nonnegative quadratic programming
  • nonnegative tensor factorization
  • object classification
  • PARAFAC
  • parallel computing

Fingerprint

Dive into the research topics of 'Novel alternating least squares algorithm for nonnegative matrix and tensor factorizations'. Together they form a unique fingerprint.

Cite this