Bo Jin
Personal Homepage
Paper Publications
Learning a Distance Metric by Balancing KL-Divergence for Imbalanced Datasets
Hits:

Indexed by:Journal Papers

First Author:Feng, Lin

Correspondence Author:Jin, B (reprint author), Dalian Univ Technol, Sch Innovat & Entrepreneurship, Fac Elect Informat & Elect Engn, Dalian 116024, Peoples R China.; Jin, B (reprint author), Dalian Univ Technol, Sch Comp Sci & Technol, Fac Elect Informat & Elect Engn, Dalian 116024, Peoples R China.

Co-author:Wang, Huibing,Jin, Bo,Li, Haohao,Xue, Mingliang,Wang, Le

Date of Publication:2019-12-01

Journal:IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS

Included Journals:EI、SCIE

Volume:49

Issue:12

Page Number:2384-2395

ISSN No.:2168-2216

Key Words:Distance metric by balancing KL-divergence (DMBK); distance metric learning (DML); geometric mean; imbalanced dataset

Abstract:In many real-world domains, datasets with imbalanced class distributions occur frequently, which may confuse various machine learning tasks. Among all these tasks, learning classifiers from imbalanced datasets is an important topic. To perform this task well, it is crucial to train a distance metric which can accurately measure similarities between samples from imbalanced datasets. Unfortunately, existing distance metric methods, such as large margin nearest neighbor, information-theoretic metric learning, etc., care more about distances between samples and fail to take imbalanced class distributions into consideration. Traditional distance metrics have natural tendencies to favor the majority classes, which can more easily satisfy their objective function. Those important minority classes are always neglected during the construction process of distance metrics, which severely affects the decision system of most classifiers. Therefore, how to learn an appropriate distance metric which can deal with imbalanced datasets is of vital importance, but challenging. In order to solve this problem, this paper proposes a novel distance metric learning method named distance metric by balancing KL-divergence (DMBK). DMBK defines normalized divergences using KL-divergence to describe distinctions between different classes. Then it combines geometric mean with normalized divergences and separates samples from different classes simultaneously. This procedure separates all classes in a balanced way and avoids inaccurate similarities incurred by imbalanced class distributions. Various experiments on imbalanced datasets have verified the excellent performance of our novel method.

Personal information

Professor
Supervisor of Doctorate Candidates
Supervisor of Master's Candidates

Gender:Male

Alma Mater:Dalian University of Technology

Degree:Doctoral Degree

School/Department:Dalian University of Technology

Discipline:Computer Applied Technology

Business Address:816 Yanjiao Building, Dalian University of Technology

Click:

Open time:..

The Last Update Time:..


Address: No.2 Linggong Road, Ganjingzi District, Dalian City, Liaoning Province, P.R.C., 116024

MOBILE Version