基于类间距的径向基函数-支持向量机核参数评价方法分析

doi:10.3969/j.issn.1000-1093.2012.02.012

兵工学报 ›› 2012, Vol. 33 ›› Issue (2): 203-208.doi: 10.3969/j.issn.1000-1093.2012.02.012

基于类间距的径向基函数-支持向量机核参数评价方法分析

宋小杉¹，蒋晓瑜¹，罗建华²，姚军¹

(1.装甲兵工程学院控制工程系，北京 100072; 2.装甲兵工程学院科研部，北京 100072）

收稿日期:2010-08-27 修回日期:2010-08-27 上线日期:2014-03-04
作者简介:宋小杉（1980—），男，博士研究生，Email: sxsh029@yahoo.com.cn
基金资助:
“十一五”装备预研项目（2009YY02）

Analysis of the Inter-class Distance-based Kernel Parameter Evaluating Method for RBF-SVM

SONG Xiao-shan¹, JIANG Xiao-yu¹, LUO Jian-hua², YAO Jun¹

(1.Department of Control Engineering, Academy of Armored Force Engineering, Beijing 100072, China;2.Department of Science Research, Academy of Armored Force Engineering, Beijing 100072, China)

Received:2010-08-27 Revised:2010-08-27 Online:2014-03-04

摘要/Abstract

摘要： 分析了径向基函数（RBF）核参数γ对空间映射结果的影响，得出3条结论。在此基础上，找到了1种新的核参数评价方法，该方法通过计算特征空间中两类之间的平均距离（ICMD）来评价γ的优劣。文章分别从理论和实验两方面证明了ICMD最大值的存在性。为验证该方法的有效性，文中对7个样本集进行了两组参数选择实验：第一组实验通过ICMD找到最优核参数γ，再由10-折交叉验证得到最优惩罚因子C，称为“两步法”；第二组实验采用基于10-折交叉验证的网格搜索法进行参数选择。结果显示两种方法均选择出了适当的参数，但前者花费的时间比后者大大缩短，验证了ICMD方法的有效性。

关键词: 人工智能, 支持向量机, 高斯核, 核参数评价, 参数选择

Abstract: The effect of radial basis function (RBF) kernel parameter γ on the mapped space was analyzed. A novel kernel parameter evaluating method was proposed, which is based on the inter-class mean distance (ICMD). The theoretical and experimental analyses were made for the proposed method. Two sets of parameter selection experiments were made in order to prove the validity of the proposed method. In the first set, the optimal parameter γ was chosen by the ICMD method, and then the optimal cost parameter C was obtained by the 10-folds cross validation, which is called “Two-Stage Method”; In the second set, the 10-folds cross validation-based grid search method was adopted. The results show that the “Two-Stage Method” can select the optimal parameters with significantly decreased time cost, which proves the validity of ICMD method.

Key words: artificial intelligence, support vector machine, radial basis function kernel, kernel parameter evaluation, parameter selection

中图分类号:

TP181

宋小杉，蒋晓瑜，罗建华，姚军. 基于类间距的径向基函数-支持向量机核参数评价方法分析[J]. 兵工学报, 2012, 33(2): 203-208.

SONG Xiao-shan, JIANG Xiao-yu, LUO Jian-hua, YAO Jun. Analysis of the Inter-class Distance-based Kernel Parameter Evaluating Method for RBF-SVM[J]. Acta Armamentarii, 2012, 33(2): 203-208.

参考文献

［1］ Vapnik V. The nature of statistics learning theory [M]. New York: Springer Verlag, 1995.
[2] Hsu C W, Chang C C, Lin C J. A practical guide to support vector classification [EB/OL]. Available at: http: ∥www.csie.ntu.edu.tw/cjlin/ papers/ guide/guide.pdf, 2003.
[3] 阎辉, 张学工, 马云潜, 等. 基于变异函数的径向基核函数参数估计[J]. 自动化学报, 2002, 28 (3): 450-455.
YAN Hui, ZHANG Xue-gong, MA Yu-qian, et al. The parameter estimation of RBF kernel function based on variogram［J］. Acta Automatica Sinica, 2002, 28(3): 450-455. (in Chinese)
[4] Ayat N E, Cheriet M, Suen C Y. Automatic model selection for the optimization of SVM kernels [J]. Pattern Recognition, 2005, 38(10): 1733-1745.
[5] Huang C L, Wang C J. A GA-based feature selection and parameters optimization for support vector machines [J]. Expert Systems with Applications, 2006, 31(2): 231-240.
[6] Yu Q, Zhang B H, Wang J L. Parameter optimization of e-SVM by Genetic Algorithm [C]∥The Fifth International Conference on Natural Computation, Tian Jin: ICNC, 2009: 540-542.
[7] Guo X C, Yang J H , Wu C G, et al. A novel LS-SVM hyper-parameter selection based on particle swarm optimization[J]. Neurocomputing, 2008, 71(16-18): 3211-3215.
[8] Zhang X Y, Guo Y L. Optimization of SVM parameters based on PSO Algorithm [C]∥The Fifth International Conference on Natural Computation, Tian Jin: ICNC, 2009: 536-539.
[9] Mu T T, Nandi A K. Automatic tuning of L2-SVM parameters employing the Extended Kalman Filter[J]. Expert Systems, 2009, 26(2): 160-175.
[10] Wu K P, Wang S D. Choosing the kernel parameters for support vector machines by the inter-cluster distance in the feature space [J]. Pattern Recognition, 2009, 42(5): 710-717.
[11] Blake C L, Merz C J. UCI Repository of machine learning databases [EB/OL]. http:∥www.ics.uci.edu/~mlearn/MLRepository.html. 2003.

[1]	刘子昌, 李思雨, 裴模超, 刘洁, 孟硕, 吴巍屹. 基于纹理分析的柴油发动机故障诊断方法[J]. 兵工学报, 2024, 45(2): 684-694.
[2]	方秋雨, 张蕴霖, 麻壮壮, 邵晋梁. 未知环境下基于控制障碍函数的无人车轨迹规划[J]. 兵工学报, 2023, 44(S2): 90-102.
[3]	裘镓荣, 曾鹏飞, 邵伟平, 赵丽俊, 郝永平. 基于PSO-LSSVM弹药装配质量预测方法[J]. 兵工学报, 2022, 43(9): 2379-2387.
[4]	纪广，郝建国，张振伟. 面向无人机作战的虚拟孪生系统设计方案[J]. 兵工学报, 2022, 43(8): 1902-1912.
[5]	段浩，陈晖，翟兆阳，韩雨，马凡华，崔亚辉. 基于支持向量机的氢混天然气发动机性能预测[J]. 兵工学报, 2022, 43(5): 1002-1011.
[6]	杨建新，兰小平，冯亚东，杨一铭，郭志明. 基于改进樽海鞘群和最小二乘支持向量机算法的新型弹药质量评估方法[J]. 兵工学报, 2022, 43(5): 1012-1022.
[7]	李良钰，苏铁熊，马富康，吴小军，徐春龙. 基于集合经验模态分解-支持向量机的高压共轨系统故障诊断方法[J]. 兵工学报, 2022, 43(5): 992-1001.
[8]	马春生, 尹晓秦, 马振东, 米文博. 下肢外骨骼的膝关节轴线自适应设计与尺度综合[J]. 兵工学报, 2022, 43(3): 653-660.
[9]	徐雨芯，顾楚梅，曹建军，许金勇，魏志虎. 基于蚁群特征选择并行分类集成学习的孪生辐射源个体识别[J]. 兵工学报, 2022, 43(12): 3132-3141.
[10]	密俊霞，于会龙，席军强. 基于MLP-SVM的驾驶员换道行为预测[J]. 兵工学报, 2022, 43(12): 3020-3029.
[11]	陈雨荻，熊智，刘建业，杨闯，晁丽君，彭杨. 基于海马体的面向未知复杂环境类脑导航技术综述[J]. 兵工学报, 2022, 43(11): 2965-2980.
[12]	马润波，董丽虹，王海斗. 基于支持向量机的热喷涂层接触疲劳寿命预测[J]. 兵工学报, 2021, 42(12): 2743-2752.
[13]	雷震烁，刘松涛，温镇铭，葛杨. 复杂电磁环境下末制导雷达工作状态实时判别方法[J]. 兵工学报, 2021, 42(1): 83-90.
[14]	王成，吴岩，杨廷飞. 利用改进单分类支持向量机提升舰船尾流目标的检测准确率[J]. 兵工学报, 2020, 41(9): 1887-1893.
[15]	吕国俊，曹建军，郑奇斌，常宸，翁年凤，彭琮. 基于多目标蚁群优化的单类支持向量机相似重复记录检测[J]. 兵工学报, 2020, 41(2): 324-331.

基于类间距的径向基函数-支持向量机核参数评价方法分析

Analysis of the Inter-class Distance-based Kernel Parameter Evaluating Method for RBF-SVM

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价