Research On Linear Programming Support Vector Machine With Path Tracking Method

Support Vector Machine(SVM)is a new type of machine learning method based on structural risk minimization principle and statistical learning theory(SLT),which has excellent learning performance and generalization performance.The traditional standard support vector machine is a quadratic programming(QP)problem.When the training set is large,the training speed will be slowing down and the training algorithm will become very complex,inefficient and maybe exhibit other issues.At the same time,support vector machine does not have a set of mature theory as a support to select suitable parameters,which bring a lot of inconvenience of the application of support vector machine.In order to solve the above problems,this paper has done the following two aspects and the main work are as follows:(1)A path follow linear programming support vector machine(PF-LPSVM)has been proposed.Firstly,the linear programming support vector machine model is constructed,and then the model will be trained by the path following internal point method to improve the training efficiency.Experiments were carried out on the random data set and the UCI data set.Lastly the experimental results were be compared and analyzed.From that we can conclude that the improved support vector machine model was improved both in classification efficiency and in classification accuracy.(2)In order to Choose the Best Combination Parameter of support vector machine(SVM),this paper applies a Low-Discrepancy Sequence Quantum Genetic Algorithm(LDQGA)to find more appropriate parameters of the SVM model.The optimal combination parameters will be applied to the optimized FP-LPSVM model and compared with the Quantum Genetic Algorithm and the 3-fold Cross-verifier.It can be seen from the experimental results that both in classification accuracy and classification efficiency have been improved,which we can concluded that the method we used in this paper to find a more suitable combination parameters of support vector machine(SVM)is feasible.