Sparse nonnegative matrix factorization based on a hyperbolic tangent approximation of L0-Norm and neurodynamic optimization

Xinqi Li, Jun Wang, Sam Kwong

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

4 Citations (Scopus)

Abstract

Sparse nonnegative matrix factorization (SNMF) attracts much attention in the past two decades because its sparse and part-based representations are desirable in many machine learning applications. Due to the combinatorial nature of the sparsity constraint in form of l0, the problem is hard to solve. In this paper, a hyperbolic tangent function is introduced to approximate the l0-norm. A discrete-time neurodynamic approach is developed for solving the proposed formulation. The stability and the convergence behavior are shown for the state vectors. Experiment results are discussed to demonstrate the superiority of the approach. The results show that this approach outperforms other sparse NMF approaches with the smallest relative reconstruction error and the required level of sparsity.

Original languageEnglish
Title of host publication12th International Conference on Advanced Computational Intelligence, ICACI 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages542-549
Number of pages8
ISBN (Electronic)9781728142487
DOIs
Publication statusPublished - Aug 2020
Externally publishedYes
Event12th International Conference on Advanced Computational Intelligence, ICACI 2020 - Dali, Yunnan, China
Duration: 14 Aug 202016 Aug 2020

Publication series

Name12th International Conference on Advanced Computational Intelligence, ICACI 2020

Conference

Conference12th International Conference on Advanced Computational Intelligence, ICACI 2020
Country/TerritoryChina
CityDali, Yunnan
Period14/08/2016/08/20

Keywords

  • Neurodynamic optimization
  • Sparse nonnegative matrix factorization

Fingerprint

Dive into the research topics of 'Sparse nonnegative matrix factorization based on a hyperbolic tangent approximation of L0-Norm and neurodynamic optimization'. Together they form a unique fingerprint.

Cite this