A Deep Non-negative Matrix Factorization Model for Big Data Representation Learning

Nowadays, deep representations have been attracting much attention owing to the great performance in various tasks. However, the interpretability of deep representations poses a vast challenge on real-world applications. To alleviate the challenge, a deep matrix factorization method with non-negativ...

Full description

Saved in:
Bibliographic Details
Published inFrontiers in neurorobotics Vol. 15; p. 701194
Main Authors Chen, Zhikui, Jin, Shan, Liu, Runze, Zhang, Jianing
Format Journal Article
LanguageEnglish
Published Lausanne Frontiers Research Foundation 20.07.2021
Frontiers Media S.A
Subjects
Online AccessGet full text
ISSN1662-5218
1662-5218
DOI10.3389/fnbot.2021.701194

Cover

More Information
Summary:Nowadays, deep representations have been attracting much attention owing to the great performance in various tasks. However, the interpretability of deep representations poses a vast challenge on real-world applications. To alleviate the challenge, a deep matrix factorization method with non-negative constraints is proposed to learn deep part-based representations of interpretability for big data in this paper. Specifically, a deep architecture with a supervisor network suppressing noise in data and a student network learning deep representations of interpretability is designed, which is an end-to-end framework for pattern mining. Furthermore, to train the deep matrix factorization architecture, an interpretability loss is defined, including a symmetric loss, an apposition loss, and a non-negative constraint loss, which can ensure the knowledge transfer from the supervisor network to the student network, enhancing the robustness of deep representations. Finally, extensive experimental results on two benchmark datasets demonstrate the superiority of the deep matrix factorization method.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
Edited by: Qingchen Zhang, St. Francis Xavier University, Canada
Reviewed by: Asif Ali Laghari, Sindh Madressatul Islam University, Pakistan; Zhenchao Ma, University of British Columbia, Canada
ISSN:1662-5218
1662-5218
DOI:10.3389/fnbot.2021.701194