Towards efficient sparse coding for scalable image annotation

Junshi Huang, Hairong Liu, Jialie Shen, Shuicheng Yan

Research output: Chapter in Book/Report/Conference proceedingConference Paper published in ProceedingsResearchpeer-review

Abstract

Nowadays, content-based retrieval methods are still the de- velopment trend of the traditional retrieval systems. Image labels, as one of the most popular approaches for the semantic representation of images, can fully capture the representative information of images. To achieve the high performance of retrieval systems, the precise annotation for images becomes inevitable. However, as the massive number of images in the Internet, one cannot annotate all the images without a scalable and exible (i.e., training-free) annotation method. In this paper, we particularly investi- gate the problem of accelerating sparse coding based scal- Able image annotation, whose off-the-shelf solvers are gen- erally inefficient on large-scale dataset. By leveraging the prior that most reconstruction coefficients should be zero, we develop a general and efficient framework to derive an accurate solution to the large-scale sparse coding problem through solving a series of much smaller-scale subproblems. In this framework, an active variable set, which expands and shrinks iteratively, is maintained, with each snapshot of the active variable set corresponding to a subproblem. Mean- while, the convergence of our proposed framework to global optimum is theoretically provable. To further accelerate the proposed framework, a sub-linear time complexity hashing strategy, e.g. Locality-Sensitive Hashing, is seamlessly integrated into our framework. Extensive empirical experiments on NUS-WIDE and IMAGENET datasets demonstrate that the orders-of-magnitude acceleration is achieved by the proposed framework for large-scale image annotation, along with zero/negligible accuracy loss for the cases without/with hashing speed-up, compared to the expensive off-the-shelf solvers.

Original languageEnglish
Title of host publicationMM 2013 - Proceedings of the 2013 ACM Multimedia Conference
Place of PublicationBarcelona; Spain
PublisherAssociation for Computing Machinery (ACM)
Pages947-956
Number of pages10
ISBN (Print)9781450324045
DOIs
Publication statusPublished - 18 Nov 2013
Externally publishedYes
Event21st ACM International Conference on Multimedia, MM 2013 - Barcelona, Spain
Duration: 21 Oct 201325 Oct 2013

Conference

Conference21st ACM International Conference on Multimedia, MM 2013
CountrySpain
CityBarcelona
Period21/10/1325/10/13

Fingerprint

Content based retrieval
Labels
Semantics
Internet
Experiments

Cite this

Huang, J., Liu, H., Shen, J., & Yan, S. (2013). Towards efficient sparse coding for scalable image annotation. In MM 2013 - Proceedings of the 2013 ACM Multimedia Conference (pp. 947-956). Barcelona; Spain: Association for Computing Machinery (ACM). https://doi.org/10.1145/2502081.2502127
Huang, Junshi ; Liu, Hairong ; Shen, Jialie ; Yan, Shuicheng. / Towards efficient sparse coding for scalable image annotation. MM 2013 - Proceedings of the 2013 ACM Multimedia Conference. Barcelona; Spain : Association for Computing Machinery (ACM), 2013. pp. 947-956
@inproceedings{6bcc2f511ac148ba8d72e72b22338473,
title = "Towards efficient sparse coding for scalable image annotation",
abstract = "Nowadays, content-based retrieval methods are still the de- velopment trend of the traditional retrieval systems. Image labels, as one of the most popular approaches for the semantic representation of images, can fully capture the representative information of images. To achieve the high performance of retrieval systems, the precise annotation for images becomes inevitable. However, as the massive number of images in the Internet, one cannot annotate all the images without a scalable and exible (i.e., training-free) annotation method. In this paper, we particularly investi- gate the problem of accelerating sparse coding based scal- Able image annotation, whose off-the-shelf solvers are gen- erally inefficient on large-scale dataset. By leveraging the prior that most reconstruction coefficients should be zero, we develop a general and efficient framework to derive an accurate solution to the large-scale sparse coding problem through solving a series of much smaller-scale subproblems. In this framework, an active variable set, which expands and shrinks iteratively, is maintained, with each snapshot of the active variable set corresponding to a subproblem. Mean- while, the convergence of our proposed framework to global optimum is theoretically provable. To further accelerate the proposed framework, a sub-linear time complexity hashing strategy, e.g. Locality-Sensitive Hashing, is seamlessly integrated into our framework. Extensive empirical experiments on NUS-WIDE and IMAGENET datasets demonstrate that the orders-of-magnitude acceleration is achieved by the proposed framework for large-scale image annotation, along with zero/negligible accuracy loss for the cases without/with hashing speed-up, compared to the expensive off-the-shelf solvers.",
keywords = "Hash-accelerated sparsity induced scalable optimization, Large-scale image annotation, Sparse coding",
author = "Junshi Huang and Hairong Liu and Jialie Shen and Shuicheng Yan",
year = "2013",
month = "11",
day = "18",
doi = "10.1145/2502081.2502127",
language = "English",
isbn = "9781450324045",
pages = "947--956",
booktitle = "MM 2013 - Proceedings of the 2013 ACM Multimedia Conference",
publisher = "Association for Computing Machinery (ACM)",
address = "United States",

}

Huang, J, Liu, H, Shen, J & Yan, S 2013, Towards efficient sparse coding for scalable image annotation. in MM 2013 - Proceedings of the 2013 ACM Multimedia Conference. Association for Computing Machinery (ACM), Barcelona; Spain, pp. 947-956, 21st ACM International Conference on Multimedia, MM 2013, Barcelona, Spain, 21/10/13. https://doi.org/10.1145/2502081.2502127

Towards efficient sparse coding for scalable image annotation. / Huang, Junshi; Liu, Hairong; Shen, Jialie; Yan, Shuicheng.

MM 2013 - Proceedings of the 2013 ACM Multimedia Conference. Barcelona; Spain : Association for Computing Machinery (ACM), 2013. p. 947-956.

Research output: Chapter in Book/Report/Conference proceedingConference Paper published in ProceedingsResearchpeer-review

TY - GEN

T1 - Towards efficient sparse coding for scalable image annotation

AU - Huang, Junshi

AU - Liu, Hairong

AU - Shen, Jialie

AU - Yan, Shuicheng

PY - 2013/11/18

Y1 - 2013/11/18

N2 - Nowadays, content-based retrieval methods are still the de- velopment trend of the traditional retrieval systems. Image labels, as one of the most popular approaches for the semantic representation of images, can fully capture the representative information of images. To achieve the high performance of retrieval systems, the precise annotation for images becomes inevitable. However, as the massive number of images in the Internet, one cannot annotate all the images without a scalable and exible (i.e., training-free) annotation method. In this paper, we particularly investi- gate the problem of accelerating sparse coding based scal- Able image annotation, whose off-the-shelf solvers are gen- erally inefficient on large-scale dataset. By leveraging the prior that most reconstruction coefficients should be zero, we develop a general and efficient framework to derive an accurate solution to the large-scale sparse coding problem through solving a series of much smaller-scale subproblems. In this framework, an active variable set, which expands and shrinks iteratively, is maintained, with each snapshot of the active variable set corresponding to a subproblem. Mean- while, the convergence of our proposed framework to global optimum is theoretically provable. To further accelerate the proposed framework, a sub-linear time complexity hashing strategy, e.g. Locality-Sensitive Hashing, is seamlessly integrated into our framework. Extensive empirical experiments on NUS-WIDE and IMAGENET datasets demonstrate that the orders-of-magnitude acceleration is achieved by the proposed framework for large-scale image annotation, along with zero/negligible accuracy loss for the cases without/with hashing speed-up, compared to the expensive off-the-shelf solvers.

AB - Nowadays, content-based retrieval methods are still the de- velopment trend of the traditional retrieval systems. Image labels, as one of the most popular approaches for the semantic representation of images, can fully capture the representative information of images. To achieve the high performance of retrieval systems, the precise annotation for images becomes inevitable. However, as the massive number of images in the Internet, one cannot annotate all the images without a scalable and exible (i.e., training-free) annotation method. In this paper, we particularly investi- gate the problem of accelerating sparse coding based scal- Able image annotation, whose off-the-shelf solvers are gen- erally inefficient on large-scale dataset. By leveraging the prior that most reconstruction coefficients should be zero, we develop a general and efficient framework to derive an accurate solution to the large-scale sparse coding problem through solving a series of much smaller-scale subproblems. In this framework, an active variable set, which expands and shrinks iteratively, is maintained, with each snapshot of the active variable set corresponding to a subproblem. Mean- while, the convergence of our proposed framework to global optimum is theoretically provable. To further accelerate the proposed framework, a sub-linear time complexity hashing strategy, e.g. Locality-Sensitive Hashing, is seamlessly integrated into our framework. Extensive empirical experiments on NUS-WIDE and IMAGENET datasets demonstrate that the orders-of-magnitude acceleration is achieved by the proposed framework for large-scale image annotation, along with zero/negligible accuracy loss for the cases without/with hashing speed-up, compared to the expensive off-the-shelf solvers.

KW - Hash-accelerated sparsity induced scalable optimization

KW - Large-scale image annotation

KW - Sparse coding

UR - http://www.scopus.com/inward/record.url?scp=84887453045&partnerID=8YFLogxK

U2 - 10.1145/2502081.2502127

DO - 10.1145/2502081.2502127

M3 - Conference Paper published in Proceedings

SN - 9781450324045

SP - 947

EP - 956

BT - MM 2013 - Proceedings of the 2013 ACM Multimedia Conference

PB - Association for Computing Machinery (ACM)

CY - Barcelona; Spain

ER -

Huang J, Liu H, Shen J, Yan S. Towards efficient sparse coding for scalable image annotation. In MM 2013 - Proceedings of the 2013 ACM Multimedia Conference. Barcelona; Spain: Association for Computing Machinery (ACM). 2013. p. 947-956 https://doi.org/10.1145/2502081.2502127