Adaptive structure discovery for multimedia analysis using multiple features

Kun Zhan, Xiaojun Chang, Junpeng Guan, Ling Chen, Zhigang Ma, Yi Yang

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)

Abstract

Multifeature learning has been a fundamental research problem in multimedia analysis. Most existing multifeature learning methods exploit graph, which must be computed beforehand, as input to uncover data distribution. These methods have two major problems confronted. First, graph construction requires calculating similarity based on nearby data pairs by a fixed function, e.g., the RBF kernel, but the intrinsic correlation among different data pairs varies constantly. Therefore, feature learning based on such predefined graphs may degrade, especially when there is dramatic correlation variation between nearby data pairs. Second, in most existing algorithms, each single-feature graph is computed independently and then combine them for learning, which ignores the correlation between multiple features. In this paper, a new unsupervised multifeature learning method is proposed to make the best utilization of the correlation among different features by jointly optimizing data correlation from multiple features in an adaptive way. As opposed to computing the affinity weight of data pairs by a fixed function, the weight of affinity graph is learned by a well-designed optimization problem. Additionally, the affinity graph of data pairs from different features is optimized in a global level to better leverage the correlation among different channels. In this way, the adaptive approach correlates the features of all features for a better learning process. Experimental results on real-world datasets demonstrate that our approach outperforms the state-of-the-art algorithms on leveraging multiple features for multimedia analysis.

Original languageEnglish
Pages (from-to)1826-1834
Number of pages9
JournalIEEE Transactions on Cybernetics
Volume49
Issue number5
DOIs
Publication statusPublished - May 2019
Externally publishedYes

Keywords

  • Clustering algorithms
  • Correlation
  • Dimensionality reduction
  • Graph learning
  • Laplace equations
  • Learning systems
  • Linear programming
  • multimedia analysis
  • Multimedia communication
  • multiview clustering
  • unsupervised learning

Cite this

Zhan, Kun ; Chang, Xiaojun ; Guan, Junpeng ; Chen, Ling ; Ma, Zhigang ; Yang, Yi. / Adaptive structure discovery for multimedia analysis using multiple features. In: IEEE Transactions on Cybernetics. 2019 ; Vol. 49, No. 5. pp. 1826-1834.
@article{4bb0f2aa09944f3abc91c58959556bb0,
title = "Adaptive structure discovery for multimedia analysis using multiple features",
abstract = "Multifeature learning has been a fundamental research problem in multimedia analysis. Most existing multifeature learning methods exploit graph, which must be computed beforehand, as input to uncover data distribution. These methods have two major problems confronted. First, graph construction requires calculating similarity based on nearby data pairs by a fixed function, e.g., the RBF kernel, but the intrinsic correlation among different data pairs varies constantly. Therefore, feature learning based on such predefined graphs may degrade, especially when there is dramatic correlation variation between nearby data pairs. Second, in most existing algorithms, each single-feature graph is computed independently and then combine them for learning, which ignores the correlation between multiple features. In this paper, a new unsupervised multifeature learning method is proposed to make the best utilization of the correlation among different features by jointly optimizing data correlation from multiple features in an adaptive way. As opposed to computing the affinity weight of data pairs by a fixed function, the weight of affinity graph is learned by a well-designed optimization problem. Additionally, the affinity graph of data pairs from different features is optimized in a global level to better leverage the correlation among different channels. In this way, the adaptive approach correlates the features of all features for a better learning process. Experimental results on real-world datasets demonstrate that our approach outperforms the state-of-the-art algorithms on leveraging multiple features for multimedia analysis.",
keywords = "Clustering algorithms, Correlation, Dimensionality reduction, Graph learning, Laplace equations, Learning systems, Linear programming, multimedia analysis, Multimedia communication, multiview clustering, unsupervised learning",
author = "Kun Zhan and Xiaojun Chang and Junpeng Guan and Ling Chen and Zhigang Ma and Yi Yang",
year = "2019",
month = "5",
doi = "10.1109/TCYB.2018.2815012",
language = "English",
volume = "49",
pages = "1826--1834",
journal = "IEEE Transactions on Cybernetics",
issn = "2168-2267",
number = "5",

}

Adaptive structure discovery for multimedia analysis using multiple features. / Zhan, Kun; Chang, Xiaojun; Guan, Junpeng; Chen, Ling; Ma, Zhigang; Yang, Yi.

In: IEEE Transactions on Cybernetics, Vol. 49, No. 5, 05.2019, p. 1826-1834.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Adaptive structure discovery for multimedia analysis using multiple features

AU - Zhan, Kun

AU - Chang, Xiaojun

AU - Guan, Junpeng

AU - Chen, Ling

AU - Ma, Zhigang

AU - Yang, Yi

PY - 2019/5

Y1 - 2019/5

N2 - Multifeature learning has been a fundamental research problem in multimedia analysis. Most existing multifeature learning methods exploit graph, which must be computed beforehand, as input to uncover data distribution. These methods have two major problems confronted. First, graph construction requires calculating similarity based on nearby data pairs by a fixed function, e.g., the RBF kernel, but the intrinsic correlation among different data pairs varies constantly. Therefore, feature learning based on such predefined graphs may degrade, especially when there is dramatic correlation variation between nearby data pairs. Second, in most existing algorithms, each single-feature graph is computed independently and then combine them for learning, which ignores the correlation between multiple features. In this paper, a new unsupervised multifeature learning method is proposed to make the best utilization of the correlation among different features by jointly optimizing data correlation from multiple features in an adaptive way. As opposed to computing the affinity weight of data pairs by a fixed function, the weight of affinity graph is learned by a well-designed optimization problem. Additionally, the affinity graph of data pairs from different features is optimized in a global level to better leverage the correlation among different channels. In this way, the adaptive approach correlates the features of all features for a better learning process. Experimental results on real-world datasets demonstrate that our approach outperforms the state-of-the-art algorithms on leveraging multiple features for multimedia analysis.

AB - Multifeature learning has been a fundamental research problem in multimedia analysis. Most existing multifeature learning methods exploit graph, which must be computed beforehand, as input to uncover data distribution. These methods have two major problems confronted. First, graph construction requires calculating similarity based on nearby data pairs by a fixed function, e.g., the RBF kernel, but the intrinsic correlation among different data pairs varies constantly. Therefore, feature learning based on such predefined graphs may degrade, especially when there is dramatic correlation variation between nearby data pairs. Second, in most existing algorithms, each single-feature graph is computed independently and then combine them for learning, which ignores the correlation between multiple features. In this paper, a new unsupervised multifeature learning method is proposed to make the best utilization of the correlation among different features by jointly optimizing data correlation from multiple features in an adaptive way. As opposed to computing the affinity weight of data pairs by a fixed function, the weight of affinity graph is learned by a well-designed optimization problem. Additionally, the affinity graph of data pairs from different features is optimized in a global level to better leverage the correlation among different channels. In this way, the adaptive approach correlates the features of all features for a better learning process. Experimental results on real-world datasets demonstrate that our approach outperforms the state-of-the-art algorithms on leveraging multiple features for multimedia analysis.

KW - Clustering algorithms

KW - Correlation

KW - Dimensionality reduction

KW - Graph learning

KW - Laplace equations

KW - Learning systems

KW - Linear programming

KW - multimedia analysis

KW - Multimedia communication

KW - multiview clustering

KW - unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=85044375956&partnerID=8YFLogxK

U2 - 10.1109/TCYB.2018.2815012

DO - 10.1109/TCYB.2018.2815012

M3 - Article

VL - 49

SP - 1826

EP - 1834

JO - IEEE Transactions on Cybernetics

JF - IEEE Transactions on Cybernetics

SN - 2168-2267

IS - 5

ER -