Non-convex regularized self-representation for unsupervised feature selection

Weizhi Wang, Hongzhi Zhang, Pengfei Zhu, Dapeng Zhang, Wangmeng Zuo

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

5 Citations (Scopus)

Abstract

Feature selection aims to select a subset of features to decrease time complexity, reduce storage burden and improve the generalization ability of classification or clustering. For the countless unlabeled high dimensional data, unsupervised feature selection is effective in alleviating the curse of dimension-ality and can find applications in various fields. In this paper, we propose a non-convex regularized self-representation (RSR) model where features can be represented by a linear combination of other features, and propose to impose L2,p norm (0 < p < 1) regularization on self-representation coefficients for unsupervised feature selection. Compared with the conventional L2, 1 norm regularization, when p < 1, much sparser solution is obtained on the self-representation coefficients, and it is also more effective in selecting salient features. To solve the non-convex RSR model, we further propose an efficient iterative reweighted least squares (IRLS) algorithm with guaranteed convergence to fixed point. Extensive experimental results on nine datasets show that our feature selection method with small p is more effective. It mostly outperforms features selected at p = 1 and other state-of-the-art unsupervised feature selection methods in terms of classification accuracy and clustering result.
Original languageEnglish
Title of host publicationIntelligence Science and Big Data Engineering
Subtitle of host publicationBig Data and Machine Learning Techniques - 5th International Conference, IScIDE 2015, Revised Selected Papers
PublisherSpringer Verlag
Pages55-65
Number of pages11
ISBN (Print)9783319238616
DOIs
Publication statusPublished - 1 Jan 2015
Event5th International Conference on Intelligence Science and Big Data Engineering, IScIDE 2015 - Suzhou, China
Duration: 14 Jun 201516 Jun 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9243
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference5th International Conference on Intelligence Science and Big Data Engineering, IScIDE 2015
Country/TerritoryChina
CitySuzhou
Period14/06/1516/06/15

Keywords

  • L norm 2p
  • Self-representation
  • Sparse representation
  • Unsupervised feature selection

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this