The Implementation of Improved Convolutional Neural Network Model in Art Pattern Design

Abstract

In order to study the identification of abnormal financial commodities in e-business, this article proposes an algorithm for identifying financial risks in ebusiness commodity transactions based on DL and CAD models. Unsupervised learning is trained by using the model gradient descent method of random automatic learning. At the same time, the algorithm uses automated distributed tables for data distribution, so as to identify the abnormal behavior in e-business transactions. The simulation results show that the accuracy of this algorithm can reach more than 90%, and the highest can reach about 95%. Its accuracy is 9.14% higher than that of traditional NN (Neural network) algorithm and 5.17% higher than that of BPNN (Back Propagation Neural Network) algorithm. Using the model constructed in this article to classify and identify abnormal behaviors in ebusiness transactions has good results, which can realize rapid and accurate identification requirements. It provides an effective and reliable method for identifying abnormal financial commodities of e-business.