[1]张浩然,汪晓东,张长江,等.一种鲁棒回归支持向量机及其学习算法[J].南京理工大学学报(自然科学版),2006,(03):311-314.
 ZHANG Hao-ran,WANG Xiao-dong,ZHANG Chang-jiang.Robust Regression SVM and Its Learning Algorithm[J].Journal of Nanjing University of Science and Technology,2006,(03):311-314.
点击复制

一种鲁棒回归支持向量机及其学习算法
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
期数:
2006年03期
页码:
311-314
栏目:
出版日期:
2006-06-30

文章信息/Info

Title:
Robust Regression SVM and Its Learning Algorithm
作者:
张浩然;汪晓东;张长江;
浙江师范大学信息科学与工程学院, 浙江金华321004
Author(s):
ZHANG Hao-ranWANG Xiao-dongZHANG Chang-jiang
College of Information Science and Engineering,Zhejiang Normal University,Jinhua 321004,China
关键词:
结构风险最小化 支持向量机 鲁棒损失函数 局部梯度法
Keywords:
structura l r iskm inim ization support vecto rm ach ine robust loss funct ion local g rad-i ent a lgorithm
分类号:
TP 183
摘要:
为了提高支持向量机的泛化能力,给出了一个鲁棒损失函数,利用它建立了鲁棒支持向量机,并利用对偶原理推导出其对偶优化问题的形式,在此基础上设计了局部梯度算法,在这种算法中每次迭代只改变两个优化变量的值。随后分析了算法的收敛性条件,给出了学习步长的选择依据,最后用一个仿真实例来说明所提出的支持向量机的学习性能,比标准支持向量机具有更好的鲁棒性。
Abstract:
In order to increase the genera lizat ion ability o f SVM ( support vectormach ine) , a robust loss function is given. A robust support vector machine is put fo rw ard. The dual optima l problem formation is deduced using a dua l theo ry. A loca l grad ient algorithm is designed, and on ly tw o opt-i ma l variables are updated in every iterat ion. The convergence condit ion o f the a lgorithm is ana lyzed, and a formu la to select learning step size is g iven using the convergence condit ion. The simulation resu lts show tha t the robust support vectormachine perform s significan tly w ell and it possesses stronger robustness than that of the orig inal support vectormach ine.

参考文献/References:

[ 1] Vapn ik V N. The nature o f statistica l lea rn ing theory [M ]. New York: Springer, 1995.
[ 2] Cherkassky V, Mu lier F. Learn ing from data- Concepts, theory and methods [M ]. New York: JohnW -i ley Sons, 1998.
[ 3 ] Joach im s T. Tex t ca tego rization w ith suppo rt vec to r m ach ines: learn ing w ith m any re levant features [ EB / OL] . http: / /c iteseer. ist. psu. edu / joachim s98 tex t. htm ,l 2006- 03- 29.
[ 4] Guyon I, W eston J, Barnh ill S. Gene selection fo r cancer class ification using support vecto r m achines [ J]. M achine Learning, 2002, 46 ( 3): 389- 422.
[ 5 ] M ukhe rjee S, Osuna E, G iro si F. Nonlinear pred iction o f chao tic tim e series us ing suppo rt vecto r mach ines [ A ]. Neu ra l Netwo rks for S ignal Process ing [ C ]. W ash ington D C: IEEE Press, 1997. 511 - 520.
[ 6] V ijayakum ar S. Sequentia l support vec to r c lassifiers and regress ion [ EB / OL]. http: / / citesee r. ist. psu. edu /v ijayakum ar99sequentia.l htm ,l 2006- 03- 29.
[ 7] M angasa rian O L, Musicant D R. Lag rang ian support vector m ach ines [ J]. Journa l of M ach ine Learn ing Research, 2001, 1( 1): 161- 177.

相似文献/References:

[1]邓敏,韩玉启.基于支持向量机的大学财务困境预警模型[J].南京理工大学学报(自然科学版),2012,36(03):551.
 DENG Min,HAN Yu-qi.Financial Crisis Early Warning Model for Universities Based on Support Vector Machine[J].Journal of Nanjing University of Science and Technology,2012,36(03):551.
[2]张玉珍,何新,王建宇,等.一种基于SVM的高效球门检测方法[J].南京理工大学学报(自然科学版),2010,(01):13.
 ZHANG Yu-zhen,HE Xin,WANG Jian-yu,et al.Efficient Goal-mouth Detection Method Based on SVM[J].Journal of Nanjing University of Science and Technology,2010,(03):13.
[3]朱宁,冯志刚,王祁,等.基于小生境遗传算法的支持向量机分类器参数优化[J].南京理工大学学报(自然科学版),2009,(01):16.
 ZHU Ning,FENG Zhi-gang,WANG Qi.Parameter Optimization of Support Vector Machine for Classification Using Niche Genetic Algorithm[J].Journal of Nanjing University of Science and Technology,2009,(03):16.
[4]栾英宏,李跃华.基于模糊支持向量机的一维距离像识别[J].南京理工大学学报(自然科学版),2009,(03):375.
 LUAN Ying-hong,LI Yue-hua.One-dimensional Range Profile Recognition Based on FSVM[J].Journal of Nanjing University of Science and Technology,2009,(03):375.
[5]张培林,等.基于蚁群算法的支持向量机参数优化[J].南京理工大学学报(自然科学版),2009,(04):464.
 ZHANG Bei-lin,QIAN Lin-fang,CAO Jian-jun,et al.Parameter Optimization of Support Vector Machine Based on Ant Colony Optimization Algorithm[J].Journal of Nanjing University of Science and Technology,2009,(03):464.
[6]张琨,曹宏鑫,衷宜,等.网络入侵检测中一种新型SVM特征加权分类方法[J].南京理工大学学报(自然科学版),2008,(02):231.
 ZHANG Kun,CAO Hong-xin,ZHONG Yi,et al.New SVM Weighted Feature Classification Method in Network Intrusion Detection[J].Journal of Nanjing University of Science and Technology,2008,(03):231.
[7]冯志刚,王祁,徐涛,等.基于小波包和支持向量机的传感器故障诊断方法[J].南京理工大学学报(自然科学版),2008,(05):609.
 FENG Zhi-gang,WANG Qi,XU Tao,et al.Sensor Fault Diagnosis Based on Wavelet Packet and Support Vector Machines[J].Journal of Nanjing University of Science and Technology,2008,(03):609.
[8]杨静宇,魏兴国,孙怀江.一种快速SVM学习算法[J].南京理工大学学报(自然科学版),2003,(05):530.
 YangJingyu WeiXingguo SunHuaijiang.A Fast SVM Learning Algorithm[J].Journal of Nanjing University of Science and Technology,2003,(03):530.
[9]张 琨,等.一种基于SVM的网络入侵检测模型[J].南京理工大学学报(自然科学版),2007,(04):403.
 ZHANG Kun,CAO Hong-xin,LIU Feng-yu,et al.SVM-based Network Intrusion Detection Model[J].Journal of Nanjing University of Science and Technology,2007,(03):403.
[10]臧希喆,赵杰,蔡鹤皋,等.面向网络遥操作的多传感器集成手爪的设计[J].南京理工大学学报(自然科学版),2006,(01):38.
 ZANG Xi-zhe,ZHAO Jie,CAI He-gao.Design of Multisensory Gripper for Internet-based Teleoperation[J].Journal of Nanjing University of Science and Technology,2006,(03):38.

备注/Memo

备注/Memo:
基金项目: 浙江省自然科学基金( Y105281) 作者简介: 张浩然( 1972- ), 男, 安徽灵璧人, 副教授, 博士, 主要研究方向: 机器学习、模式识别及其在信号处理中的应用, E-m ail:hylt@zjnu. cn。
更新日期/Last Update: 2006-06-30