Abstract

User-adaptive image retrieval/recommendation has drawn a lot of research interests in recent years, owing to fast development of various Web applications where retrieving images is a key enabling task. Existing challenges include the lack of user-adaptive training data, the ambiguity of user query and the real-time interactivity of a system. This paper proposes a hybrid learning strategy that fuses knowledge from both pointwise and pairwise training data into one framework for attribute-based, user-adaptive image retrieval. Under this framework, we develop an online learning algorithm for updating the ranking performance based on user feedback. Furthermore, we derive the framework into a kernel form, allowing easy application of kernel techniques. The proposed approach is evaluated on two image datasets and experimental results show that it achieves obvious performance gains over ranking and zero-shot learning from either type of training data independently. In addition, the online learning algorithm is able to deliver much better performance than batch learning, given the same elapsed running time, or can achieve better performance in much less time.

Original languageEnglish (US)
Title of host publicationICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval
PublisherAssociation for Computing Machinery, Inc
Pages67-74
Number of pages8
ISBN (Print)9781450332743
DOIs
StatePublished - Jun 22 2015
Event5th ACM International Conference on Multimedia Retrieval, ICMR 2015 - Shanghai, China
Duration: Jun 23 2015Jun 26 2015

Other

Other5th ACM International Conference on Multimedia Retrieval, ICMR 2015
CountryChina
CityShanghai
Period6/23/156/26/15

Fingerprint

Image retrieval
Learning algorithms
Labels
Electric fuses
Feedback

Keywords

  • Adaptive image retrieval
  • Attribute learning
  • Learning to rank

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Software
  • Computer Vision and Pattern Recognition

Cite this

Chen, L., Zhang, P., & Li, B. (2015). Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval. In ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval (pp. 67-74). Association for Computing Machinery, Inc. https://doi.org/10.1145/2671188.2749358

Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval. / Chen, Lin; Zhang, Peng; Li, Baoxin.

ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval. Association for Computing Machinery, Inc, 2015. p. 67-74.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chen, L, Zhang, P & Li, B 2015, Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval. in ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval. Association for Computing Machinery, Inc, pp. 67-74, 5th ACM International Conference on Multimedia Retrieval, ICMR 2015, Shanghai, China, 6/23/15. https://doi.org/10.1145/2671188.2749358
Chen L, Zhang P, Li B. Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval. In ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval. Association for Computing Machinery, Inc. 2015. p. 67-74 https://doi.org/10.1145/2671188.2749358
Chen, Lin ; Zhang, Peng ; Li, Baoxin. / Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval. ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval. Association for Computing Machinery, Inc, 2015. pp. 67-74
@inproceedings{3e0f3dfa99c749119149e97a63a97655,
title = "Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval",
abstract = "User-adaptive image retrieval/recommendation has drawn a lot of research interests in recent years, owing to fast development of various Web applications where retrieving images is a key enabling task. Existing challenges include the lack of user-adaptive training data, the ambiguity of user query and the real-time interactivity of a system. This paper proposes a hybrid learning strategy that fuses knowledge from both pointwise and pairwise training data into one framework for attribute-based, user-adaptive image retrieval. Under this framework, we develop an online learning algorithm for updating the ranking performance based on user feedback. Furthermore, we derive the framework into a kernel form, allowing easy application of kernel techniques. The proposed approach is evaluated on two image datasets and experimental results show that it achieves obvious performance gains over ranking and zero-shot learning from either type of training data independently. In addition, the online learning algorithm is able to deliver much better performance than batch learning, given the same elapsed running time, or can achieve better performance in much less time.",
keywords = "Adaptive image retrieval, Attribute learning, Learning to rank",
author = "Lin Chen and Peng Zhang and Baoxin Li",
year = "2015",
month = "6",
day = "22",
doi = "10.1145/2671188.2749358",
language = "English (US)",
isbn = "9781450332743",
pages = "67--74",
booktitle = "ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - Fusing pointwise and pairwise labels for supporting user-adaptive image retrieval

AU - Chen, Lin

AU - Zhang, Peng

AU - Li, Baoxin

PY - 2015/6/22

Y1 - 2015/6/22

N2 - User-adaptive image retrieval/recommendation has drawn a lot of research interests in recent years, owing to fast development of various Web applications where retrieving images is a key enabling task. Existing challenges include the lack of user-adaptive training data, the ambiguity of user query and the real-time interactivity of a system. This paper proposes a hybrid learning strategy that fuses knowledge from both pointwise and pairwise training data into one framework for attribute-based, user-adaptive image retrieval. Under this framework, we develop an online learning algorithm for updating the ranking performance based on user feedback. Furthermore, we derive the framework into a kernel form, allowing easy application of kernel techniques. The proposed approach is evaluated on two image datasets and experimental results show that it achieves obvious performance gains over ranking and zero-shot learning from either type of training data independently. In addition, the online learning algorithm is able to deliver much better performance than batch learning, given the same elapsed running time, or can achieve better performance in much less time.

AB - User-adaptive image retrieval/recommendation has drawn a lot of research interests in recent years, owing to fast development of various Web applications where retrieving images is a key enabling task. Existing challenges include the lack of user-adaptive training data, the ambiguity of user query and the real-time interactivity of a system. This paper proposes a hybrid learning strategy that fuses knowledge from both pointwise and pairwise training data into one framework for attribute-based, user-adaptive image retrieval. Under this framework, we develop an online learning algorithm for updating the ranking performance based on user feedback. Furthermore, we derive the framework into a kernel form, allowing easy application of kernel techniques. The proposed approach is evaluated on two image datasets and experimental results show that it achieves obvious performance gains over ranking and zero-shot learning from either type of training data independently. In addition, the online learning algorithm is able to deliver much better performance than batch learning, given the same elapsed running time, or can achieve better performance in much less time.

KW - Adaptive image retrieval

KW - Attribute learning

KW - Learning to rank

UR - http://www.scopus.com/inward/record.url?scp=84962376481&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84962376481&partnerID=8YFLogxK

U2 - 10.1145/2671188.2749358

DO - 10.1145/2671188.2749358

M3 - Conference contribution

SN - 9781450332743

SP - 67

EP - 74

BT - ICMR 2015 - Proceedings of the 2015 ACM International Conference on Multimedia Retrieval

PB - Association for Computing Machinery, Inc

ER -