Hierarchical support vector machines

Zhigang Liu; Wenzhong Shi; Qianqing Qin; Xiaowen Li; Donghui Xie

doi:10.1109/IGARSS.2005.1526138

Hierarchical support vector machines

Zhigang Liu
, Wenzhong Shi
, Qianqing Qin
, Xiaowen Li
, Donghui Xie

The Hong Kong Polytechnic University

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

17 Citations (Scopus)

Abstract

The speed and accuracy of a hierarchical SVM (H-SVM) depend on its tree structure. To achieve high performance, more separable classes should be separated at the upper nodes of a decision tree. Because SVM separates classes at feature space determined by the kernel function, separability in feature space should be considered. In this paper, a separability measure in feature space based on support vector data description is proposed. Based on this measure, we present two kinds of H-SVM, Binary Tree SVM and K-Tree SVM, the decision trees of which are constructed with two bottom-up agglomerative clustering algorithms respectively. Results of experimentation with remotely sensed data validate the effectiveness of the two proposed H-SVM.

Original language	English
Title of host publication	25th Anniversary IGARSS 2005
Subtitle of host publication	IEEE International Geoscience and Remote Sensing Symposium
Pages	186-189
Number of pages	4
Volume	1
DOIs	https://doi.org/10.1109/IGARSS.2005.1526138
Publication status	Published - 1 Dec 2005
Event	2005 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2005 - Seoul, Korea, Republic of Duration: 25 Jul 2005 → 29 Jul 2005

Conference

Conference	2005 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2005
Country/Territory	Korea, Republic of
City	Seoul
Period	25/07/05 → 29/07/05

ASJC Scopus subject areas

Computer Science Applications
General Earth and Planetary Sciences

More information

10.1109/IGARSS.2005.1526138

Cite this

@inproceedings{6d0f0ebcdd25404ea7dceb39c9ac0007,

title = "Hierarchical support vector machines",

abstract = "The speed and accuracy of a hierarchical SVM (H-SVM) depend on its tree structure. To achieve high performance, more separable classes should be separated at the upper nodes of a decision tree. Because SVM separates classes at feature space determined by the kernel function, separability in feature space should be considered. In this paper, a separability measure in feature space based on support vector data description is proposed. Based on this measure, we present two kinds of H-SVM, Binary Tree SVM and K-Tree SVM, the decision trees of which are constructed with two bottom-up agglomerative clustering algorithms respectively. Results of experimentation with remotely sensed data validate the effectiveness of the two proposed H-SVM.",

author = "Zhigang Liu and Wenzhong Shi and Qianqing Qin and Xiaowen Li and Donghui Xie",

year = "2005",

month = dec,

day = "1",

doi = "10.1109/IGARSS.2005.1526138",

language = "English",

isbn = "0780390504",

volume = "1",

pages = "186--189",

booktitle = "25th Anniversary IGARSS 2005",

note = "2005 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2005 ; Conference date: 25-07-2005 Through 29-07-2005",

}

TY - GEN

T1 - Hierarchical support vector machines

AU - Liu, Zhigang

AU - Shi, Wenzhong

AU - Qin, Qianqing

AU - Li, Xiaowen

AU - Xie, Donghui

PY - 2005/12/1

Y1 - 2005/12/1

N2 - The speed and accuracy of a hierarchical SVM (H-SVM) depend on its tree structure. To achieve high performance, more separable classes should be separated at the upper nodes of a decision tree. Because SVM separates classes at feature space determined by the kernel function, separability in feature space should be considered. In this paper, a separability measure in feature space based on support vector data description is proposed. Based on this measure, we present two kinds of H-SVM, Binary Tree SVM and K-Tree SVM, the decision trees of which are constructed with two bottom-up agglomerative clustering algorithms respectively. Results of experimentation with remotely sensed data validate the effectiveness of the two proposed H-SVM.

AB - The speed and accuracy of a hierarchical SVM (H-SVM) depend on its tree structure. To achieve high performance, more separable classes should be separated at the upper nodes of a decision tree. Because SVM separates classes at feature space determined by the kernel function, separability in feature space should be considered. In this paper, a separability measure in feature space based on support vector data description is proposed. Based on this measure, we present two kinds of H-SVM, Binary Tree SVM and K-Tree SVM, the decision trees of which are constructed with two bottom-up agglomerative clustering algorithms respectively. Results of experimentation with remotely sensed data validate the effectiveness of the two proposed H-SVM.

UR - https://www.scopus.com/pages/publications/33745727880

U2 - 10.1109/IGARSS.2005.1526138

DO - 10.1109/IGARSS.2005.1526138

M3 - Conference article published in proceeding or book

SN - 0780390504

SN - 9780780390508

VL - 1

SP - 186

EP - 189

BT - 25th Anniversary IGARSS 2005

T2 - 2005 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2005

Y2 - 25 July 2005 through 29 July 2005

ER -

Hierarchical support vector machines

Abstract

Conference

ASJC Scopus subject areas

More information

Other files and links

Fingerprint

Cite this