计算机应用与软件Issue(6)：236-240,266,6.DOI:10.3969/j.issn.1000-386x.2015.06.058
一种结合共享最近邻法和粒度支持向量机的混合模型
A HYBRID MODEL INTEGRATING SHARED NEAREST NEIGHBOURHOOD AND GRANULAR SUPPORT VECTOR MACHINE
摘要
Abstract
When granular support vector machine (GSVM)dealing with large-scale data set,the division of granules has significant im-pact on training efficiency and generalisation ability of its model.However,the randomness of traditional division method seriously affects the training effect of GSVMmodel.In light of this issue,in this paper we propose a hybrid model which integrates the shared nearest neighbour-hood and the granular support vector machine (GSVM-SNN).By using the shared nearest neighbour it automatically partitions the sample point into several information granules,and extracts from them the key information.Since most of the support vector points are distributed on the edges of information granules,we propose a k-NN connectivity in the paper,by calculating the connectivity we pick up pure points on granule edges and fuse all key information to build final decision model.Experimental result demonstrates that comparing with traditional GS-VM,this new method has certain advantages in classification time and classification accuracy.关键词
支持向量机/SNN/粒度支持向量机/KNNKey words
Support vector machine/Shared nearest neighbourhood (SNN)/Granular support vector machine/k-nearest neighbourhood/(k-NN)分类
信息技术与安全科学引用本文复制引用
王建国,范凯,张文兴..一种结合共享最近邻法和粒度支持向量机的混合模型[J].计算机应用与软件,2015,(6):236-240,266,6.基金项目
国家自然科学基金项目（21366017）；内蒙古自然科学基金重大项目（2011ZD08）；内蒙教育局项目（NJZY11150）；包头市科技局重大科技发展项目（2011Z1006）。 （）
