Adaptive Semi-supervised Feature Selection for cross-modal retrieval

En Yu, Jiande Sun, Jing Li, Xiaojun Chang, Xian-Hua Han, Alexander G. Hauptmann

Research output: Contribution to journalArticleResearchpeer-review

Abstract

In order to exploit the abundant potential information of the unlabeled data and contribute to analyze the correlation among heterogeneous data, we propose the semi-supervised model named Adaptive Semi-supervised Feature Selection (ASFS) for cross-modal retrieval. Firstly, we utilize the semantic regression to strengthen the neighbor relationship between the data with the same semantic. And the correlation between heterogeneous data can be optimized via keeping the pairwise closeness when learning the common latent space. Secondly, we adopt the graph-based constraint to predict accurate labels for unlabeled data, and it can also keep the geometric structure consistency between the label space and feature space of heterogeneous data in the common latent space. Finally, an efficient joint optimization algorithm is proposed to update the mapping matrices and label matrix for unlabeled data simultaneously and iteratively. It makes samples from different classes to be far apart while the samples from same class lie as close as possible. Meanwhile, the <formula><tex>${l_{2,1}}$</tex></formula>-norm constraint is used for feature selection and outliers reduction when the mapping matrices are learned. In addition, we propose learning different mapping matrices corresponding to different sub-tasks to emphasize the semantic and structural information of query data. Experiment results on three datasets demonstrate that our method performs better than the state-of-the-art methods.

Original languageEnglish
Pages (from-to)1276-1288
Number of pages13
JournalIEEE Transactions on Multimedia
Volume21
Issue number5
DOIs
Publication statusPublished - May 2019

Keywords

  • Bicycles
  • Correlation
  • cross-modal retrieval
  • Feature extraction
  • feature selection
  • Learning systems
  • Optimization
  • Semantics
  • semi-supervised
  • Task analysis

Cite this

Yu, En ; Sun, Jiande ; Li, Jing ; Chang, Xiaojun ; Han, Xian-Hua ; Hauptmann, Alexander G. / Adaptive Semi-supervised Feature Selection for cross-modal retrieval. In: IEEE Transactions on Multimedia. 2019 ; Vol. 21, No. 5. pp. 1276-1288.
@article{fb8a590b4f1c40c7bb155100666303bd,
title = "Adaptive Semi-supervised Feature Selection for cross-modal retrieval",
abstract = "In order to exploit the abundant potential information of the unlabeled data and contribute to analyze the correlation among heterogeneous data, we propose the semi-supervised model named Adaptive Semi-supervised Feature Selection (ASFS) for cross-modal retrieval. Firstly, we utilize the semantic regression to strengthen the neighbor relationship between the data with the same semantic. And the correlation between heterogeneous data can be optimized via keeping the pairwise closeness when learning the common latent space. Secondly, we adopt the graph-based constraint to predict accurate labels for unlabeled data, and it can also keep the geometric structure consistency between the label space and feature space of heterogeneous data in the common latent space. Finally, an efficient joint optimization algorithm is proposed to update the mapping matrices and label matrix for unlabeled data simultaneously and iteratively. It makes samples from different classes to be far apart while the samples from same class lie as close as possible. Meanwhile, the ${l_{2,1}}$-norm constraint is used for feature selection and outliers reduction when the mapping matrices are learned. In addition, we propose learning different mapping matrices corresponding to different sub-tasks to emphasize the semantic and structural information of query data. Experiment results on three datasets demonstrate that our method performs better than the state-of-the-art methods.",
keywords = "Bicycles, Correlation, cross-modal retrieval, Feature extraction, feature selection, Learning systems, Optimization, Semantics, semi-supervised, Task analysis",
author = "En Yu and Jiande Sun and Jing Li and Xiaojun Chang and Xian-Hua Han and Hauptmann, {Alexander G.}",
year = "2019",
month = "5",
doi = "10.1109/TMM.2018.2877127",
language = "English",
volume = "21",
pages = "1276--1288",
journal = "IEEE Transactions on Multimedia",
issn = "1520-9210",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
number = "5",

}

Adaptive Semi-supervised Feature Selection for cross-modal retrieval. / Yu, En; Sun, Jiande; Li, Jing; Chang, Xiaojun; Han, Xian-Hua; Hauptmann, Alexander G.

In: IEEE Transactions on Multimedia, Vol. 21, No. 5, 05.2019, p. 1276-1288.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Adaptive Semi-supervised Feature Selection for cross-modal retrieval

AU - Yu, En

AU - Sun, Jiande

AU - Li, Jing

AU - Chang, Xiaojun

AU - Han, Xian-Hua

AU - Hauptmann, Alexander G.

PY - 2019/5

Y1 - 2019/5

N2 - In order to exploit the abundant potential information of the unlabeled data and contribute to analyze the correlation among heterogeneous data, we propose the semi-supervised model named Adaptive Semi-supervised Feature Selection (ASFS) for cross-modal retrieval. Firstly, we utilize the semantic regression to strengthen the neighbor relationship between the data with the same semantic. And the correlation between heterogeneous data can be optimized via keeping the pairwise closeness when learning the common latent space. Secondly, we adopt the graph-based constraint to predict accurate labels for unlabeled data, and it can also keep the geometric structure consistency between the label space and feature space of heterogeneous data in the common latent space. Finally, an efficient joint optimization algorithm is proposed to update the mapping matrices and label matrix for unlabeled data simultaneously and iteratively. It makes samples from different classes to be far apart while the samples from same class lie as close as possible. Meanwhile, the ${l_{2,1}}$-norm constraint is used for feature selection and outliers reduction when the mapping matrices are learned. In addition, we propose learning different mapping matrices corresponding to different sub-tasks to emphasize the semantic and structural information of query data. Experiment results on three datasets demonstrate that our method performs better than the state-of-the-art methods.

AB - In order to exploit the abundant potential information of the unlabeled data and contribute to analyze the correlation among heterogeneous data, we propose the semi-supervised model named Adaptive Semi-supervised Feature Selection (ASFS) for cross-modal retrieval. Firstly, we utilize the semantic regression to strengthen the neighbor relationship between the data with the same semantic. And the correlation between heterogeneous data can be optimized via keeping the pairwise closeness when learning the common latent space. Secondly, we adopt the graph-based constraint to predict accurate labels for unlabeled data, and it can also keep the geometric structure consistency between the label space and feature space of heterogeneous data in the common latent space. Finally, an efficient joint optimization algorithm is proposed to update the mapping matrices and label matrix for unlabeled data simultaneously and iteratively. It makes samples from different classes to be far apart while the samples from same class lie as close as possible. Meanwhile, the ${l_{2,1}}$-norm constraint is used for feature selection and outliers reduction when the mapping matrices are learned. In addition, we propose learning different mapping matrices corresponding to different sub-tasks to emphasize the semantic and structural information of query data. Experiment results on three datasets demonstrate that our method performs better than the state-of-the-art methods.

KW - Bicycles

KW - Correlation

KW - cross-modal retrieval

KW - Feature extraction

KW - feature selection

KW - Learning systems

KW - Optimization

KW - Semantics

KW - semi-supervised

KW - Task analysis

UR - http://www.scopus.com/inward/record.url?scp=85055194628&partnerID=8YFLogxK

U2 - 10.1109/TMM.2018.2877127

DO - 10.1109/TMM.2018.2877127

M3 - Article

VL - 21

SP - 1276

EP - 1288

JO - IEEE Transactions on Multimedia

JF - IEEE Transactions on Multimedia

SN - 1520-9210

IS - 5

ER -