Support Vector Machine Based On Rough Fuzzy And Fuzzy Rough Clustering

The shortcomings of the original clustering methods are analyzed. Moreover, the rough theory and fuzzy theory are combined together. Firstly, the improvement of rough fuzzy K-means clustering algorithm is given. Secondly, a fuzzy rough K-means clustering algorithm is designed, and the validity of fuzzy rough K-means clustering algorithm is verified. The proposed clustering algorithms are applied to support vector machine. In the above applications, the training samples are pre-processed to reduce the number of samples and improve the training speed and the classification accuracy. At last, on the basis of considering the tightness of samples, a new fuzzy membership is confirmed, and a new fuzzy support vector machine is constructed. The experimental result shows that this method can be given to non-support vectors small membership. Therefore, the impact of these non-support vectors on the classification boundary can be reduced. At the same time, it can assure that the support vectors can obtain greater membership degree and the classification accuracy of fuzzy support vector machine can be improved.