Abstract: Support vector machine can not directly deal with high dimension and large scale training set and it is sensifive to abnormal samples,an improved support vector classifier based on neighborhood rough set was proposed. In the paper, data preprocessing was done on training set from two different sides. On the one hand, neighborhood rough set was used to find these samples in boundary and obtain a reduced training set, at the same time, those abnormal samples which not only lead to over-learning but also decrease the generalization ability were deleted. On the other hand, attribute reduction was done and feature weight was imported based on attribute significance because different feature effects differently on classification. At last several comparative experiments using synthetic and real life data set show the performance and the effectivity of the method.

Key words: Support vector machine, Neighborhood rough set, Preprocess, Attribute reduction