Research And Application Of Heterogeneous Weighted Support Vector Machine Algorithm

With the continuous development of science and technology,people’s daily production and life will produce a large amount of data information.In the face of massive data information,how to accurately and efficiently analyze the relationship between data,discover the knowledge contained in data,and obtain the hidden information between data has become a new problem.The support vector machine algorithm in statistical learning theory has attracted extensive attention because of its good generalization ability and the advantages of avoiding dimension disaster.In the traditional support vector machine algorithm,all sample points are considered to be equally important,that is,each sample point has the same weight in the algorithm.To solve this problem,this paper first proposes an adaptive penalty factor for unbalanced data and introduces it into the fuzzy support vector classification algorithm.The theoretical derivation and proof of the algorithm and the least-squares solution are given under the condition that the linearity can be divided into two cases,and its approximate correctness in the sense of probability is proved.Secondly,the self weighted support vector regression algorithm is proposed to prove the existence and uniqueness of the solution,and the dual problem of the self weighted support vector regression algorithm is given.It is proved that the solution of the dual problem has a unified form and is the solution of the original problem.Finally,simulation experiments are given for the proposed improved support vector classification algorithm and regression algorithm to verify the effectiveness and superiority of the model.