The Research Of Artificial Neural Network Classification And The Application On The Personal Credit Evaluation

On the basis of research on Artificial Neural Network, this paper discusses a new method for classification——Support Vector Machine(SVM) and applies it to personal credit evaluation domain. Finally the paper establishes an accurate classifier.Support Vector Machine (SVM) is a kind of statistical learning theory for classification and regression, which is proposed by Vapnik in 1995. It's a learning system using linear function to assume space in high-dimension feature space. In the recent years, it has got breakthrough improvement in its theory research and algorithm implements, and therefore it becomes a powerful method of overcoming the traditional difficulties such as dimension disaster, over fitting, and so on. SVM is being paid more attention to because of many remarkable advantages and promising performance in experiments. It has been the enthusiastic of the machine learning research domain and it has got very ideal effects, such as face recognition, handwritten digit recognition, web classification, and so on.With the development of consumer credit in domestic commercial banks, personal credit is attached more and more value in our country. Personal credit evaluation has wide applicative foreground, and scholars inside and outside have done much research work on it. They have reported many preprocessing algorithms and pattern recognition algorithms, which improves the accuracy of personal credit evaluation in great measure. But up to now, the accuracy still needs to be improved and the problems such as selecting kernel functions and kernel parameters, and one class samples serious intermixed in another class still need to be solved.To improve the accuracy of personal credit evaluation, this paper applies improved support vector machine to personal credit evaluation and exploits a software system named MULTIEDIT-SVM-KNN. On the basis of concluding the research of the people of the past, this paper puts the emphasis on the factors that influence the performance of SVM classification. We flow process the method of choosing the best factors to validate the effectiveness of improved support vector machine for personal credit evaluation. In addition, firstly the paper prunes samples and then uses the classifier named SVM-KNN. The main work of this paper appears in the following aspects:⑴A nalyzing and comparing the methods of Artificial Neural Network classification. This paper mainly analyzes and compares three Neural Network classification methods: Back Propagation, Redial Basis Function Network and Support Vector Machine, and finds the most applicable Neural network classification method for personal credit evaluation. Finally the paper decides that SVM is the most appropriate method.⑵Comparing and analyzing the training algorithms.Comparing the three main training algorithms: Chunking algorithm, Osuna algorithm and SMO algorithm in speed, accuracy and memory saving. After comparison, we find that SMO algorithm has fast speed, high accuracy and needs the least memory, so it is appropriate to the solution of large-scale problems. Therefore, the SMO algorithm is selected for the problem.⑶Some researches are done on data preparation for personal credit evaluation. A method of selecting indexes is proposed based on Principal Component Analysis.⑷Proposing method named MULTIEDIT-SVM-KNN to improve SVM classifier.In the class process, the fact that the one class samples serious intermixed in another class leads to the complicated classification followed by lower generalization performance, and when the distance from the test sample to the optimal superplane of SVM in feature space is short, the precision of the method is not accuracy. So we prune the train samples by the method named Multiedit Nearest Neighbor, and then in the class phase, the algorithm computes the distance from the test sample to the optimal super plane of SVM in feature space, classes the test samples by SVM-KNN classifier, and finally improves SVM classifier.⑸Validating the effectiveness of MULTIEDIT-SVM-KNN for personal credit evaluation.This paper applies improved support vector machine to personal credit evaluation to improve the recognition accuracy because of its advantages. We flow process the procedure by applying the best factors in every step to MULTIEDIT-SVM-KNN software system. This paper shows a series of operations on personal credit database including preprocessing of data preparation, indexes selecting, data scaling, train samples pruning, the selection of best kernel functions and kernel parameters by grid search and k-fold cross validation, training and testing. Comparing the experiment results with those using other techniques on the same database can validate the effectiveness of MULTIEDIT–SVM- KNN classifier.Expecting to push the development of improving the accuracy of personal credit evaluation based on support vector machine.