Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning

He Zhang, Zhirong Yang, Mehmet Gönen, Markus Koskela, Jorma Laaksonen, Timo Honkela, Erkki Oja

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Abstract

Emotional semantic image retrieval systems aim at incorporating the user's affective states for responding adequately to the user's interests. One challenge is to select features specific to image affect detection. Another challenge is to build effective learning models or classifiers to bridge the so-called "affective gap". In this work, we study the affective classification and retrieval of abstract images by applying multiple kernel learning framework. An image can be represented by different feature spaces and multiple kernel learning can utilize all these feature representations simultaneously (i.e., multiview learning), such that it jointly learns the feature representation weights and corresponding classifier in an intelligent manner. Our experimental results on two abstract image datasets demonstrate the advantage of the multiple kernel learning framework for image affect detection in terms of feature selection, classification performance, and interpretation.
Original languageEnglish
Title of host publicationUnknown host publication
EditorsM. Lee, A. Hirose, Z.-G. Hou, R.M. Kil
Number of pages10
Publication date2013
Pages166-175
Publication statusPublished - 2013
MoE publication typeA4 Article in conference proceedings

Cite this

Zhang, H., Yang, Z., Gönen, M., Koskela, M., Laaksonen, J., Honkela, T., & Oja, E. (2013). Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning. In M. Lee, A. Hirose, Z-G. Hou, & R. M. Kil (Eds.), Unknown host publication (pp. 166-175)
Zhang, He ; Yang, Zhirong ; Gönen, Mehmet ; Koskela, Markus ; Laaksonen, Jorma ; Honkela, Timo ; Oja, Erkki. / Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning. Unknown host publication. editor / M. Lee ; A. Hirose ; Z.-G. Hou ; R.M. Kil. 2013. pp. 166-175
@inproceedings{38ce6f2c605949d0943f2696e1e41f77,
title = "Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning",
abstract = "Emotional semantic image retrieval systems aim at incorporating the user's affective states for responding adequately to the user's interests. One challenge is to select features specific to image affect detection. Another challenge is to build effective learning models or classifiers to bridge the so-called {"}affective gap{"}. In this work, we study the affective classification and retrieval of abstract images by applying multiple kernel learning framework. An image can be represented by different feature spaces and multiple kernel learning can utilize all these feature representations simultaneously (i.e., multiview learning), such that it jointly learns the feature representation weights and corresponding classifier in an intelligent manner. Our experimental results on two abstract image datasets demonstrate the advantage of the multiple kernel learning framework for image affect detection in terms of feature selection, classification performance, and interpretation.",
author = "He Zhang and Zhirong Yang and Mehmet G{\"o}nen and Markus Koskela and Jorma Laaksonen and Timo Honkela and Erkki Oja",
note = "Volume: Proceeding volume:",
year = "2013",
language = "English",
pages = "166--175",
editor = "M. Lee and A. Hirose and Z.-G. Hou and R.M. Kil",
booktitle = "Unknown host publication",

}

Zhang, H, Yang, Z, Gönen, M, Koskela, M, Laaksonen, J, Honkela, T & Oja, E 2013, Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning. in M Lee, A Hirose, Z-G Hou & RM Kil (eds), Unknown host publication. pp. 166-175.

Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning. / Zhang, He; Yang, Zhirong; Gönen, Mehmet; Koskela, Markus; Laaksonen, Jorma; Honkela, Timo; Oja, Erkki.

Unknown host publication. ed. / M. Lee; A. Hirose; Z.-G. Hou; R.M. Kil. 2013. p. 166-175.

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

TY - GEN

T1 - Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning

AU - Zhang, He

AU - Yang, Zhirong

AU - Gönen, Mehmet

AU - Koskela, Markus

AU - Laaksonen, Jorma

AU - Honkela, Timo

AU - Oja, Erkki

N1 - Volume: Proceeding volume:

PY - 2013

Y1 - 2013

N2 - Emotional semantic image retrieval systems aim at incorporating the user's affective states for responding adequately to the user's interests. One challenge is to select features specific to image affect detection. Another challenge is to build effective learning models or classifiers to bridge the so-called "affective gap". In this work, we study the affective classification and retrieval of abstract images by applying multiple kernel learning framework. An image can be represented by different feature spaces and multiple kernel learning can utilize all these feature representations simultaneously (i.e., multiview learning), such that it jointly learns the feature representation weights and corresponding classifier in an intelligent manner. Our experimental results on two abstract image datasets demonstrate the advantage of the multiple kernel learning framework for image affect detection in terms of feature selection, classification performance, and interpretation.

AB - Emotional semantic image retrieval systems aim at incorporating the user's affective states for responding adequately to the user's interests. One challenge is to select features specific to image affect detection. Another challenge is to build effective learning models or classifiers to bridge the so-called "affective gap". In this work, we study the affective classification and retrieval of abstract images by applying multiple kernel learning framework. An image can be represented by different feature spaces and multiple kernel learning can utilize all these feature representations simultaneously (i.e., multiview learning), such that it jointly learns the feature representation weights and corresponding classifier in an intelligent manner. Our experimental results on two abstract image datasets demonstrate the advantage of the multiple kernel learning framework for image affect detection in terms of feature selection, classification performance, and interpretation.

M3 - Conference contribution

SP - 166

EP - 175

BT - Unknown host publication

A2 - Lee, M.

A2 - Hirose, A.

A2 - Hou, Z.-G.

A2 - Kil, R.M.

ER -

Zhang H, Yang Z, Gönen M, Koskela M, Laaksonen J, Honkela T et al. Affective Abstract Image Classification and Retrieval Using Multiple Kernel Learning. In Lee M, Hirose A, Hou Z-G, Kil RM, editors, Unknown host publication. 2013. p. 166-175