The Algorithm And Application Of Linear Separable Support Vector Machine

The large data has the characteristics of high dimension and huge amount,which brings new development opportunities for modern society.The big data analysis can solve the country's many and important demand problem,in the future personalization medical,the scientific innovation and discovery,the innovation technology and the business management and the government management decision-making,etc.has brought the information tsunami,has driven another industrial revolution;Big data as new economic resources,new economic development engine,new economic output = capital + Labour + data,and data automatically become the new economic assets such as gold and currency,but also the new basis of scientific research and decision-making,profound impact on all aspects of social and economic life,and become a new impetus for social progress.Support vector machine(SVM)is one of the most important tools for big data analysis.It is suitable for high-dimensional small sample data analysis.It is a two classifier.The decision support model of SVM can be transformed into the solution of convex two degree programming.However,in large data background,massive training samples lead to the training of support vector machines with dimension disaster and huge computational cost.In this paper,the large sample training data are randomly grouped according to certain rules.The sample data of each group is a small sample of support vector machine.The separation Hyperplane of the support vector machine is obtained by means of the rotation and migration of the sensor.The initial Hyperplane of SVM is used as the initial Hyperplane of SVM iteration,and then the dynamic learning rate is used to continue the rotation and translation of the initial plane in the region formed by positive and negative classes until the geometric interval reaches maximum.The two algorithms presented in this paper are verified with actual examples and simulation data respectively.The results obtained from the support vector machine training package which directly calls R software are identical to those obtained.