| 模糊超球支持向量机 |
| |
| 引用本文： | 张金泽,单甘霖.模糊超球支持向量机[J].军械工程学院学报,2005,17(3):65-67. |
| |
| 作者姓名： | 张金泽 单甘霖 |
| |
| 作者单位： | 军械工程学院光学与电子工程系 河北石家庄050003
(张金泽)，军械工程学院光学与电子工程系 河北石家庄050003(单甘霖) |
| |
| 摘 要： | 提出了模糊超球支持向量机算法。首先根据样本分布特性将样本空间划分成有限个超球子空间,超球半径经归一化处理后作为各球心样本点的模糊属性值,超球球心组成新的训练样本集,在构造决策超平面的过程中对应不同模糊属性值的各输入样本增加了不同训练权重。仿真实验结果表明:该算法减少了训练样本数,有利于提高支持向量机的训练速度和测试速度,有效地防止了训练样本中噪声点对训练结果的负面影响。
|
| 关 键 词： | 支持向量机 超球 模糊隶属度 |
| 文章编号： | 1008-2956(2005)03-0065-03 |
| 修稿时间： | 2005年3月22日 |
Fuzzy Hypersphere Support Vector Machine |
| |
| ZHANG Jin-ze,SHAN Gan-lin.Fuzzy Hypersphere Support Vector Machine[J].Journal of Ordnance Engineering College,2005,17(3):65-67. |
| |
| Authors: | ZHANG Jin-ze SHAN Gan-lin |
| |
| Abstract: | Fuzzy Hypersphere SVM algorithm is presented in the paper.First of all,the sample space is partitioned into hypersphere subspaces according to the distributing characteristics of training samples.After processed,the radiuses are regarded as fuzzy memberships of the samples located on the center of hyperspheres.Those samples as new training set with different fuzzy memberships make different contributions to decision hyperplane constructing.Experimental results show that the number of training samples is reduced, which means a little of the training and testing time.It can also avoid the bad effect of noises in the training samples. |
| |
| Keywords: | support vector machine (SVM) hypersphere fuzzy membership |
| 本文献已被 CNKI 维普 等数据库收录！ |
