摘要
为解决核主成分分析(KPCA)和支持向量机(SVM)融合算法分类精度差的问题,提出基于差空间融合特征选择的SVM算法。利用主成分分析(PCA)处理原始数据,得到差空间数据;分别对原数据和差空间数据进行KPCA,得到融合特征;用ReliefF算法得到对应特征的权重,根据初步分类评价指标选择最优的特征组合;对得到的数据利用SVM进行分类。该算法在UCI数据集上的测试结果表明,它能够有效提高分类精度,在高维数据中减小分类过程的计算复杂度。
To solve the problem that the fusion algorithm of kernel principle component analysis (KPCA) and support vector machine (SVM) show poor classification accuracy,SVM algorithm based on feature selection of differential space fusion was proposed.The original data were processed using principle component analysis (PCA) to get differential space data.KPCA algorithm was performed on the original data and differential space data respectively to get the mixed features.ReliefF algorithm was used to get the weight of features,and the optimal combination of features was selected using preliminary classification evaluation index.SVM algorithm was used to classify the dimensionality reduction data.Testing of the proposed algorithm on the UCI data sets shows that it can not only improve the classification accuracy effectively,but reduce the computational complexity of the classification process in high dimension data.
作者
景炜
丁卫平
JING Wei;DING Wei-ping(School of Electronics and Information,Nantong University,Nantong 226019,China;School of Computer Science and Technology,Nantong University,Nantong 226019,China)
出处
《计算机工程与设计》
北大核心
2019年第8期2235-2241,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(61300167)
江苏省自然科学基金项目(BK20151274)
江苏省六大人才高峰基金项目(XYDXXJS-048)
南通市应用基础研究基金项目(GY12016014)
江苏高校“青蓝工程”基金项目(苏教师[2019]3号)
