Prediction of Illegal Fundraising Behavior Based on Convolutional Neural Network

We proposed a hybrid machine learning algorithm to predict the risk of illegal fundraising for companies. Then, the intention of criminals to disguise the company's business and financial situation can be prevented for victims who may invest in illegal fundraising. We collected risk data from 355 companies on the Internet "Enterprise Inspection" platform. With data preprocessing, combined with the features of bank statements, the features for the company's risk assessment are obtained. Then, a feature prediction model for user transaction behavior was trained using a convolutional neural network (CNN). Based on this model, the users of each company were screened, and the ratio of the number of users who complied with illegal fundraising and transactions in the company was obtained, The prediction of the risk of illegal fundraising showed a satisfactory effect.