Evolutionary nonnegative matrix factorization for data compression

Gong, Liyun, Mu, Tingting and Y. Goulermas, John (2015) Evolutionary nonnegative matrix factorization for data compression. Lecture Notes in Computer Science, 9225 . pp. 23-33. ISSN 0302-9743

Documents
ICIC2015.pdf
[img]
[Download]
19730 additional text.pdf
[img]
Preview
PDF
ICIC2015.pdf - Whole Document

1MB
[img] PDF
19730 additional text.pdf - Supplemental Material
Restricted to Repository staff only

8kB
Item Type:Article
Item Status:Live Archive

Abstract

This paper aims at improving non-negative matrix factor- ization (NMF) to facilitate data compression. An evolutionary updat- ing strategy is proposed to solve the NMF problem iteratively based on three sets of updating rules including multiplicative, firefly and sur- vival of the fittest rules. For data compression application, the quality of the factorized matrices can be evaluated by measurements such as spar- sity, orthogonality and factorization error to assess compression quality in terms of storage space consumption, redundancy in data matrix and data approximation accuracy. Thus, the fitness score function that drives the evolving procedure is designed as a composite score that takes into account all these measurements. A hybrid initialization scheme is per- formed to improve the rate of convergence, allowing multiple initial can- didates generated by different types of NMF initialization approaches. Effectiveness of the proposed method is demonstrated using Yale and ORL image datasets.

Keywords:Non-negative matrix factorization; data compression, evolutionary computation, NotOAChecked
Subjects:G Mathematical and Computer Sciences > G700 Artificial Intelligence
Divisions:College of Science
Related URLs:
ID Code:19730
Deposited On:08 Dec 2015 18:57

Repository Staff Only: item control page