The Financial Freud Recognition Model Based On Support Vector Machine

Because the growth of economic is rapidly,and the complex of the economic activities is increasing,concealment of the financial fraud is also increasing.There are urgent need for effective recognition of the financial fraud.With the development of science and technology.artificial intelligence method is relatively mature,more and more machine learning methods are introduced to the field of financial fraud recognition.Support vector machine(Support Vector Machine,referred to as SVM)is a new generation machine learning technique based on statistical learning theory,and widely applied to studies on financial fraud recognition field.Now,most of the literature focus on the financial fraud recognition model established by SVM,ignores the method of selecting model parameters,and proper selection of parameters is directly related to he prediction accuracy and generalization ability.This paper selects the genetic algorithm(genetic algorithm,referred to as GA),the particle swarm optimization(particle swarm optimization,referred to as PSO)to find the optimal parameters of support vector machine.This paper establishes the model of SVM which to find the optimal parameters by using the genetic algorithm(genetic algorithm,referred to as GA),the particle swarm optimization(particle swarm optimization,referred to as PSO)on the basis of listing corporation financial data and other important information from 2006 to 2015 of the public offering,at the same time,this study selecting 18 financial recognition index and 8 non-financial recognition index,by T independence test,this study obtain 8 recognition index which can make the recognition result better.Then obtain the training accuracy,training time and the grid search algorithm to search the optimal parameters in the modeling process,through recognizes the same test set to obtain the recognition rate.The comparison of the results shows that the accuracy rate and the training time of the particle swarm optimization is better than the other two methods.After the simulation.we found that the difference of the two recognition rate of PSO is the smallest,so the extension ability of the PSO is better than the other two methods.