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 language | English (US) |
|---|---|
| Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
| Pages | 166-175 |
| Number of pages | 10 |
| Volume | 8228 LNCS |
| Edition | PART 3 |
| DOIs | |
| State | Published - 2013 |
| Externally published | Yes |
| Event | 20th International Conference on Neural Information Processing, ICONIP 2013 - Daegu, Korea, Republic of Duration: Nov 3 2013 → Nov 7 2013 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Number | PART 3 |
| Volume | 8228 LNCS |
| ISSN (Print) | 03029743 |
| ISSN (Electronic) | 16113349 |
Other
| Other | 20th International Conference on Neural Information Processing, ICONIP 2013 |
|---|---|
| Country | Korea, Republic of |
| City | Daegu |
| Period | 11/3/13 → 11/7/13 |
Fingerprint
Keywords
- Group lasso
- Image affect
- Image classification and retrieval
- Low-level image features
- Multiple kernel learning
ASJC Scopus subject areas
- Computer Science(all)
- Theoretical Computer Science
Cite this
Affective abstract image classification and retrieval using multiple kernel learning. / Zhang, He; Yang, Zhirong; Gonen, Mehmet; Koskela, Markus; Laaksonen, Jorma; Honkela, Timo; Oja, Erkki.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8228 LNCS PART 3. ed. 2013. p. 166-175 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8228 LNCS, No. PART 3).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Affective abstract image classification and retrieval using multiple kernel learning
AU - Zhang, He
AU - Yang, Zhirong
AU - Gonen, Mehmet
AU - Koskela, Markus
AU - Laaksonen, Jorma
AU - Honkela, Timo
AU - Oja, Erkki
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.
KW - Group lasso
KW - Image affect
KW - Image classification and retrieval
KW - Low-level image features
KW - Multiple kernel learning
UR - http://www.scopus.com/inward/record.url?scp=84893413775&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84893413775&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-42051-1_22
DO - 10.1007/978-3-642-42051-1_22
M3 - Conference contribution
AN - SCOPUS:84893413775
SN - 9783642420504
VL - 8228 LNCS
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 166
EP - 175
BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ER -