|
Title:
|
AN INCREMENTAL SPARSE LS-SVM CLASSIFICATION METHOD FOR IMBALANCED DATA SETS |
|
Author(s):
|
Leichen Chen, Zhihua Cai, Shuang Ao |
|
ISBN:
|
978-972-8939-23-6 |
|
Editors:
|
António Palma dos Reis and Ajith P. Abraham |
|
Year:
|
2010 |
|
Edition:
|
Single |
|
Keywords:
|
Imbalanced data sets, incremental learning, sparse least squares support vector machine |
|
Type:
|
Full Paper |
|
First Page:
|
45 |
|
Last Page:
|
52 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In this paper, a new classification method (ISLS-SVM) for imbalanced data sets is proposed. In this method, the original training data set is constituted by all the minority samples and the same amount of randomly selected majority samples, and the incremental data set is consisted of the rest majority samples. The classifier tests the incremental data and the misclassification samples replace some small value of support vectors with the same class label, then both the training data set and the incremental data set are reconstructed. It stops until the whole incremental data are correct classification or the misclassification data are never changed. A series of experiments on both UCI standard data sets and an engineering data set of coal and gas outburst have shown that the new classification method (ISLS-SVM) performs better under the criterion of F-measure and ROC Area (AUC) than the existing methods of Adaboost, LS-SVM, Tomek Links and SMOTE |
|
|
|
|
|
|
