Double centroids-weighted support vector clustering algorithm for group-air grouping

Aiming at group-air grouping, a double centroids-weighted support vector clustering(D-SVC) algorithm is proposed. To effectively solve the Lagrange multipliers during the support vector machine training, the proposed algorithm introduces the maximum entropy principle. Different characteristics of samples lead to different clustering results, so the density-weighted centroids is introduced to the double centroids-weighted cluster labeling process, and the effectiveness of the algorithm is proved in classical sample sets. Through describing the group-air grouping model, the feature sets of a target point are proposed during group-air grouping. Besides, simulation experiment for group-air grouping is carried out. Simulation results show that the D-SVC cluster analysis of specific sample sets can be carried out quickly using the proposed algorithm which also ensures the effectiveness of the clustering results.