Optimization Research and Application of Unbalanced Data Set Multi-classification Algorithm
Leng Yi Ren, Weimin Zhou · 2016
This paper studied multi-classification algorithm of unbalanced data. The conventional classifier often directly expands dichotomy algorithm to multi-classification in dealing with unbalanced data multi-classification without paying attention to the relationship among data. According to the effect of data relationship on SVM algorithm improvement, this paper proposed a spatial extension-based SVM algorithm, which can optimize classifier and improve classification accuracy of minority sample data. Finally, the validity of modified algorithm was verified through data set.