Block-iterative algorithms for non-negative matrix approximation

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

8 Scopus citations

Abstract

In this paper we present new algorithms for non-negative matrix approximation (NMA), commonly known as the NMF problem. Our methods improve upon the well-known methods of Lee & Seung [12] for both the Frobenius norm as well the Kullback-Leibler divergence versions of the problem. For the latter problem, our results are especially interesting because it seems to have witnessed much lesser algorithmic progress as compared to the Frobenius norm NMA problem. Our algorithms are based on a particular block-iterative acceleration technique for EM, which pre- serves the multiplicative nature of the updates and also en- sures monotonicity. Furthermore, our algorithms also nat- urally apply to the Bregman-divergence NMA algorithms of [6]. Experimentally, we show that our algorithms outper- form the traditional Lee/Seung approach most of the time.

Original languageEnglish
Title of host publicationProceedings - 8th IEEE International Conference on Data Mining, ICDM 2008
Pages1037-1042
Number of pages6
DOIs
StatePublished - 2008
Externally publishedYes
Event8th IEEE International Conference on Data Mining, ICDM 2008 - Pisa, Italy
Duration: 15 Dec 200819 Dec 2008

Publication series

NameProceedings - IEEE International Conference on Data Mining, ICDM
ISSN (Print)1550-4786

Conference

Conference8th IEEE International Conference on Data Mining, ICDM 2008
Country/TerritoryItaly
CityPisa
Period15/12/0819/12/08

Fingerprint

Dive into the research topics of 'Block-iterative algorithms for non-negative matrix approximation'. Together they form a unique fingerprint.

Cite this