Software source code vulnerability detection method based on artificial neural network

Abstract

The invention discloses a software source code vulnerability detection method based on an artificial neural network, and the method is based on deep learning and a word vector technology derived froma language model, so that a hidden mode of a code can be automatically learned by a deep learning algorithm, thereby saving the time consumption and tedious process of manual feature extraction. And the complexity of code processing and analysis is further weakened, so that the time investment of code processing and analysis is saved. The ELMo-based deep learning framework can directly accept a source code sequence as input and output a prediction result, so that end-to-end detection on a source code level is realized in a real sense.

Description

technical field [0001] The invention relates to the technical field of software source code loophole detection, in particular to an artificial neural network-based software source code loophole detection method. Background technique [0002] As a preventive attack defense measure, software vulnerability detection technology has attracted much attention in the field of computer security. The most cost-effective move is to conduct software vulnerability detection before the software is officially released. In this way, vulnerabilities that may be exploited by attackers can be discovered in advance and repaired in time, thereby preventing user data and services from being attacked. [0003] Traditional software vulnerability detection techniques can be divided into three categories: static, dynamic and hybrid. Static analysis techniques such as rule-matching-based filtering and symbolic execution are implemented by analyzing software source code. The disadvantage of these st...

AI Technical Summary

Problems solved by technology

But nothing can be done about the loopholes caused by non-library references

Therefore, the quality of feature engineering may limit the performance of algorithm prediction performance

In order to ensure the quality of feature extraction, feature engineering participants have to have an in-depth understanding of the software project itself, which greatly increases the threshold for feature participants

At the same time, extracting features is usually a labor-intensive task with high demand for manpower and time costs

Images