106 0

클래스 불균형 데이터의 효과적인 분류를 위한 k-NN과 생성적 적대 신경망 기반의 오버 샘플링

Title
클래스 불균형 데이터의 효과적인 분류를 위한 k-NN과 생성적 적대 신경망 기반의 오버 샘플링
Other Titles
Oversampling Based on k-NN and GAN for Effective Classification of Class Imbalance Dataset
Author
허선
Keywords
Class Imbalance Dataset; Classifiers; Oversampling; k-Nearest Neighbor; GAN
Issue Date
2020-08
Publisher
대한산업공학회
Citation
대한산업공학회지, v. 46, No. 4, Page. 365-371
Abstract
Class imbalanced dataset is common in real world and may degrade the performance of the classifier. To address this, oversampling method that artificially creates the samples of minority class is adopted but is known to be ineffective for the high-dimensional dataset because it generates the samples whose distribution is far different from that of existing samples. Novel oversampling methods based on the generative adversarial networks (GAN) have been recently developed, but generated samples may have different degrees of impact on the performance of the classifier. Therefore, more efficient method that can capture the characteristics of the generated samples and select those samples that will be used to train and improve the performance of the classifier is necessary. This study proposes a GAN-based new oversampling method which generates artificial samples based on the distribution of existing minority class samples and extracts only those which are effective to expand the realm of samples of minority class using the k-nearest neighbor. We show the proposed method outperforms existing methods with respect to F1-measure by several illustrative datasets.
URI
https://www.dbpia.co.kr/journal/articleDetail?nodeId=NODE09415226https://repository.hanyang.ac.kr/handle/20.500.11754/164674
ISSN
1225-0988; 2234-6457
DOI
10.7232/JKIIE.2020.46.4.365
Appears in Collections:
COLLEGE OF ENGINEERING SCIENCES[E](공학대학) > INDUSTRIAL AND MANAGEMENT ENGINEERING(산업경영공학과) > Articles
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE