基于加权最小二乘双支持向量机的含噪声分类  

Classification with Noise via Weighted Least Squares Twin Support Vector Machine

作　　者：穆晓霞[1,2] 李钧涛[3] 陈留院[4] 

机构地区：[1]河南师范大学计算机与信息工程学院,河南新乡453007 [2]武汉理工大学信息工程学院,湖北武汉430070 [3]河南师范大学数学与信息科学学院,河南新乡453007 [4]河南师范大学学报编辑部,河南新乡453007

出　　处：《计算机仿真》2014年第5期288-292,共5页Computer Simulation

基　　金：国家自然科学基金(61203293);河南省重点科技攻关计划(122102210131;122102210132);河南省基础与前沿技术研究(132300410390;132300410389);河南省高校科技创新人才支持计划(13HASTIT040);河南省教育厅科技攻关计划(13A120524);河南省高校青年骨干教师资助计划(2012GGJS-063)

摘　　要：针对最小二乘双支持向量机对噪声样本敏感的问题,依据给含有大噪声的样本赋予较小权重、给较小噪声的样本赋予较大权重的原则,通过评估训练样本点到两个非平行分类超平面的距离,构造了能反映样本噪声程度的权重,提出了线性和非线性加权最小二乘双支持向量机,并发展了两种加权支持向量机的求解算法,解决了对含噪声样本的高精度分类问题。将所提两种加权最小二乘双支持向量机分别应用到Heart-statlog和Two-moons数据集上进行仿真,结果表明所提方法有效消除了噪声的影响,提高了分类精度。This paper aimed to overcome the influence of the noise on the least squares twin support vector machine. Following the principle that gives the samples containing large noise smaller weights and the samples containing small noise larger weights, the weights reflecting the noise level in samples were first proposed by evaluating the distances from the training samples to the two nonparallel separating hyperplanes. By using the proposed weights, the linear and nonlinear weighted least squares twin support vector machines were presented and the corresponding solving algorithms were developed. The proposed weighted least squares twin support vector machines were applied to the Heart-statlog data and Two-moons data. The simulation results show that the proposed methods can remove the influence of the noise and improve the classification accuracy.

关 键 词：支持向量机 最小二乘双支持向量机 加权支持向量机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]