Research on Software Vulnerability Detection Method Based on Improved CNN Model

A software construction detection algorithm based on improved CNN model is proposed. Firstly, extract the vulnerability characteristics of software, from static code by using program slicing technology, establish library, standardize language, and vectorize it as input data. Gru used to optimize neural network. The organic combination two can quickly process feature data retain calling relationship between codes. Compared with single RNN model, has stronger ability higher accuracy. In contrast, strong terms training loss rate, DNN + 17.2% lower than 10.5% 7% VulDeePecker model.