东篱科研大数据发现系统（DRDS）

位置：成果数据库 > 期刊 > 期刊详情页

Training Robust Support Vector Machine Based on a New Loss Function

ISSN号：1672-5220
期刊名称：东华大学学报(英文版)
时间：2015.4.1
页码：261-263
分类：TP18[自动化与计算机技术—控制科学与工程;自动化与计算机技术—控制理论与控制工程]
作者机构：[1]School of Mathematics and Statistics, Henan University of Science & Technology, Luoyang 471003, China
相关基金：National Natural Science Foundations of China （Nos. 61272015, 11201123 ） ; the Scientific Research Foundation for the Doctor of Henan University of Science ＆Technology, China （No. 09001476） ; School Foundation of Hcnan University of Science ＆Technology, China （ No. 2012QN011 ）
相关项目：保险投资组合随机模型中风险控制及优化分红问题的研究

作者：刘叶青|

中文摘要：

To reduce the influences of outliers on support vector machine(SVM) classification problem,a new tangent loss function was constructed.Since the tangent loss function was not smooth in some interval,a smoothing function was used to approximate it in this interval.According to this loss function,the corresponding tangent SVM(TSVM) was got.The experimental results show that TSVM is less sensitive to outliers than SVM.So the proposed new loss function and TSVM are both effective.

英文摘要：

To reduce the influences of outliers on support vector machine（SVM） classification problem,a new tangent loss function was constructed.Since the tangent loss function was not smooth in some interval,a smoothing function was used to approximate it in this interval.According to this loss function,the corresponding tangent SVM（TSVM） was got.The experimental results show that TSVM is less sensitive to outliers than SVM.So the proposed new loss function and TSVM are both effective.

同期刊论文项目