期刊
NEURAL PROCESSING LETTERS
卷 51, 期 1, 页码 41-66出版社
SPRINGER
DOI: 10.1007/s11063-019-10078-w
关键词
Pattern classification; Information entropy; Least squares twin support vector machine; Fuzzy membership
资金
- National Natural Science Foundation of China [61702012]
- University Outstanding Young Talent Support Project of Anhui Province of China [gxyq2017026]
- University Natural Science Research Project of Anhui Province of China [KJ2016A431, KJ2017A361, KJ2017A368]
- Program for Innovative Research Team in Anqing Normal University
Least squares twin support vector machine (LSTSVM) is a new machine learning method, as opposed to solving two quadratic programming problems in twin support vector machine (TWSVM), which generates two nonparallel hyperplanes by solving a pair of linear system of equations. However, LSTSVM obtains the resultant classifier by giving same importance to all training samples which may be important for classification performance. In this paper, by considering the fuzzy membership value for each sample, we propose an entropy-based fuzzy least squares twin support vector machine where fuzzy membership values are assigned based on the entropy values of all training samples. The proposed method not only retains the superior characteristics of LSTSVM which is simple and fast algorithm, but also implements the structural risk minimization principle to overcome the possible over-fitting problem. Experiments are performed on several synthetic as well as benchmark datasets and the experimental results illustrate the effectiveness of our method.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据