GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces

Yuchai Wan; Xiabi Liu; Kunqi Tong; Xue Wei; Yi Wu; Fei Guan; Kunpeng Pang

doi:10.1007/978-3-642-34481-7_26

GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces

Yuchai Wan^*, Xiabi Liu, Kunqi Tong, Xue Wei, Yi Wu, Fei Guan, Kunpeng Pang

^*Corresponding author for this work

School of Computer Science and Technology

Beijing Institute of Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

5 Citations (Scopus)

Abstract

This paper proposes a novel clustering based indexing approach called GMM-ClusterForest for supporting multi-features based similarity search in high-dimensional spaces. We fit a Gaussian Mixture Model (GMM) to data through the Expectation-Maximization (EM) algorithm for estimating GMM parameters and the Minimum Description Length (MDL) criterion for selecting GMM structure. Each Gaussian component in the GMM is taken as a cluster center and each data point is assigned to the cluster according to the Bayesian decision rule. By performing this clustering method hierarchically, an index tree is constructed and the corresponding similarity search method is developed for a type of features. Then multi-features based similarity search is fulfilled by fusing the index trees for all the types of features considered. We evaluated the proposed indexing approach through applying it to example-based image retrieval and conducting the experiments on Corel 1000 dataset and self-collected large dataset. The experimental results show that our approach is effective and promising.

Original language	English
Title of host publication	Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings
Pages	210-217
Number of pages	8
Edition	PART 2
DOIs	https://doi.org/10.1007/978-3-642-34481-7_26
Publication status	Published - 2012
Event	19th International Conference on Neural Information Processing, ICONIP 2012 - Doha, Qatar Duration: 12 Nov 2012 → 15 Nov 2012

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 2
Volume	7664 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	19th International Conference on Neural Information Processing, ICONIP 2012
Country/Territory	Qatar
City	Doha
Period	12/11/12 → 15/11/12

Keywords

Clustering
Content-Based Image Retrieval (CBIR)
Gaussian Mixture Models (GMM)
High-dimensional data indexing
Similarity search

Access to Document

10.1007/978-3-642-34481-7_26

Cite this

Wan, Y., Liu, X., Tong, K., Wei, X., Wu, Y., Guan, F., & Pang, K. (2012). GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces. In Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings (PART 2 ed., pp. 210-217). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7664 LNCS, No. PART 2). https://doi.org/10.1007/978-3-642-34481-7_26

Wan, Yuchai ; Liu, Xiabi ; Tong, Kunqi et al. / GMM-ClusterForest : A novel indexing approach for multi-features based similarity search in high-dimensional spaces. Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings. PART 2. ed. 2012. pp. 210-217 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).

@inproceedings{04a47b52c6f1471abeae3d8ebda1d6dd,

title = "GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces",

abstract = "This paper proposes a novel clustering based indexing approach called GMM-ClusterForest for supporting multi-features based similarity search in high-dimensional spaces. We fit a Gaussian Mixture Model (GMM) to data through the Expectation-Maximization (EM) algorithm for estimating GMM parameters and the Minimum Description Length (MDL) criterion for selecting GMM structure. Each Gaussian component in the GMM is taken as a cluster center and each data point is assigned to the cluster according to the Bayesian decision rule. By performing this clustering method hierarchically, an index tree is constructed and the corresponding similarity search method is developed for a type of features. Then multi-features based similarity search is fulfilled by fusing the index trees for all the types of features considered. We evaluated the proposed indexing approach through applying it to example-based image retrieval and conducting the experiments on Corel 1000 dataset and self-collected large dataset. The experimental results show that our approach is effective and promising.",

keywords = "Clustering, Content-Based Image Retrieval (CBIR), Gaussian Mixture Models (GMM), High-dimensional data indexing, Similarity search",

author = "Yuchai Wan and Xiabi Liu and Kunqi Tong and Xue Wei and Yi Wu and Fei Guan and Kunpeng Pang",

year = "2012",

doi = "10.1007/978-3-642-34481-7_26",

language = "English",

isbn = "9783642344800",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

number = "PART 2",

pages = "210--217",

booktitle = "Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings",

edition = "PART 2",

note = "19th International Conference on Neural Information Processing, ICONIP 2012 ; Conference date: 12-11-2012 Through 15-11-2012",

}

Wan, Y, Liu, X, Tong, K, Wei, X, Wu, Y, Guan, F & Pang, K 2012, GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces. in Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 7664 LNCS, pp. 210-217, 19th International Conference on Neural Information Processing, ICONIP 2012, Doha, Qatar, 12/11/12. https://doi.org/10.1007/978-3-642-34481-7_26

GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces. / Wan, Yuchai; Liu, Xiabi; Tong, Kunqi et al.
Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings. PART 2. ed. 2012. p. 210-217 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7664 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - GMM-ClusterForest

T2 - 19th International Conference on Neural Information Processing, ICONIP 2012

AU - Wan, Yuchai

AU - Liu, Xiabi

AU - Tong, Kunqi

AU - Wei, Xue

AU - Wu, Yi

AU - Guan, Fei

AU - Pang, Kunpeng

PY - 2012

Y1 - 2012

N2 - This paper proposes a novel clustering based indexing approach called GMM-ClusterForest for supporting multi-features based similarity search in high-dimensional spaces. We fit a Gaussian Mixture Model (GMM) to data through the Expectation-Maximization (EM) algorithm for estimating GMM parameters and the Minimum Description Length (MDL) criterion for selecting GMM structure. Each Gaussian component in the GMM is taken as a cluster center and each data point is assigned to the cluster according to the Bayesian decision rule. By performing this clustering method hierarchically, an index tree is constructed and the corresponding similarity search method is developed for a type of features. Then multi-features based similarity search is fulfilled by fusing the index trees for all the types of features considered. We evaluated the proposed indexing approach through applying it to example-based image retrieval and conducting the experiments on Corel 1000 dataset and self-collected large dataset. The experimental results show that our approach is effective and promising.

AB - This paper proposes a novel clustering based indexing approach called GMM-ClusterForest for supporting multi-features based similarity search in high-dimensional spaces. We fit a Gaussian Mixture Model (GMM) to data through the Expectation-Maximization (EM) algorithm for estimating GMM parameters and the Minimum Description Length (MDL) criterion for selecting GMM structure. Each Gaussian component in the GMM is taken as a cluster center and each data point is assigned to the cluster according to the Bayesian decision rule. By performing this clustering method hierarchically, an index tree is constructed and the corresponding similarity search method is developed for a type of features. Then multi-features based similarity search is fulfilled by fusing the index trees for all the types of features considered. We evaluated the proposed indexing approach through applying it to example-based image retrieval and conducting the experiments on Corel 1000 dataset and self-collected large dataset. The experimental results show that our approach is effective and promising.

KW - Clustering

KW - Content-Based Image Retrieval (CBIR)

KW - Gaussian Mixture Models (GMM)

KW - High-dimensional data indexing

KW - Similarity search

UR - http://www.scopus.com/inward/record.url?scp=84869072309&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-34481-7_26

DO - 10.1007/978-3-642-34481-7_26

M3 - Conference contribution

AN - SCOPUS:84869072309

SN - 9783642344800

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 210

EP - 217

BT - Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings

Y2 - 12 November 2012 through 15 November 2012

ER -

Wan Y, Liu X, Tong K, Wei X, Wu Y, Guan F et al. GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces. In Neural Information Processing - 19th International Conference, ICONIP 2012, Proceedings. PART 2 ed. 2012. p. 210-217. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). doi: 10.1007/978-3-642-34481-7_26

GMM-ClusterForest: A novel indexing approach for multi-features based similarity search in high-dimensional spaces

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this