| Since the introduction of Vapnik in the 1990s, support vector machine has become a hotspot of machine learning with its excellent learning ability. The traditional support vector machine is to study two kinds of classification, and the practical application of multi-class classification is more common, while the actual application will encounter a lot of fuzzy information, how to better apply it to multi-class classification and overcome the fuzzy information Interference is the focus of support vector machine research.In this paper, the existing fuzzy binary tree support vector machine is improved in the construction of tree structure and the construction method of membership function. The main work is as follows:1.Aiming at the shortcomings of the existing tree structure construction method of fuzzy binary tree support vector machine, a construction method based on k-means clustering algorithm is proposed,which is a bottom-up method of constructing tree structure. The combination of different classes makes the combination of different classes more reasonable, and the structure of the tree structure is not forced to be completely normal, more in line with the actual situation,while numerical experiments verify the effectiveness of this method.2.In this paper, the membership function of fuzzy binary tree support vector machine is studied, and a membership function method based on correlation coefficient is proposed. According to the correlation degree of the sample points, different values are given to realize the fuzzy processing of the samples, thus avoiding the And the feasibility of the experiment is verified by numerical experiments.3.An improved fuzzy binary tree support vector machine is proposed and compared with 1-v-rSVMs and 1-v-1SVMs. The results show that the improved fuzzy binary tree support vector machine has higher overall performance than 1-v-rSVMs and 1-v-1SVMs.Finally, summed up the main work of this article and makes a prospect for the future work. |
PDF Full Text Request