Research And Application Of Support Vector Clustering Algorithm

Clustering is an unsupervised learning algorithm,which is an important research content in pattern recognition and machine learning and has been widely used in practical problems.In 2015,a twin support vector clustering(TWSVC)algorithm based on plane clustering was proposed,which sought the central plane of clustering for each class.Least squares twin support vector clustering(LSTWSVC)is an improved algorithm of TWSVC.However,LSTWSVC does not fully consider the distribution information and structure information of data.The clustering effect is not good on data sets with irregular data distribution or noise.This paper attempts to introduce fuzzy membership,fully mining the distribution information and structure information of samples,and then carries out research on LSTWSVC and proposes new clustering algorithms to improve the clustering accuracy and reduce the complexity of the algorithm.Finally,some experiments are carried out on some artificial data sets and UCI data sets to verify the effectiveness of the proposed algorithm.Specific research contents are as follows:1.In order to make full use of the distribution information of data,a kind of fuzzy membership is constructed.Each sample is assigned a different weight,so that the contribution of the sample to the central plane of clustering is different.Fuzzy least squares twin support vector clustering(FLSTWSVC)is proposed,and the experimental results verify the effectiveness of the algorithm.2.In order to make full use of the structural information of the data,the intra-class covariance matrix and energy factor are introduced,and the equality constraint of LSTWSVC is transformed into equality constraint based on energy.And then,a novel energy-based structural least squares twin support vector clustering(ESLSTWSVC)is proposed,which can obtain more structural information and reduce the influence of noise and outliers on the clustering effect.Experimental results show that the proposed algorithm has fast training speed and relatively good robustness.