Fine-grained vulnerability detection method based on depth features

Abstract

The invention discloses a fine-grained vulnerability detection method based on depth characteristics. The fine-grained vulnerability detection method comprises the following two stages: a training stage and a detection stage. The training stage comprises the steps that a large number of programs with vulnerabilities and without vulnerabilities are collected; Preprocessing the programs, and extracting program slices from the program dependency graph; Labeling the generated program slice according to the vulnerability type; Extracting a program focus point from the program slice according to a security analysis rule; Converting the program slice and the program focus point into vectors; Building a vulnerability detection model based on deep learning, and using vectors to train model parameters to be optimal; And finally, a trained vulnerability detection model based on deep learning is obtained. The detection stage comprises the following steps: extracting a program slice and a program focus point from a program to be detected according to a source code processing mode of the training stage, and respectively converting the program slice and the program focus point into vectors; And classifying the vectors by using the trained vulnerability detection model, and finally generating a vulnerability detection report according to a classification result.

Description

technical field [0001] The invention belongs to the field of cyberspace security, and more specifically relates to a fine-grained vulnerability detection method based on deep features. Background technique [0002] Most network attacks are caused by some software vulnerabilities, although many security technologies have emerged to avoid software vulnerabilities. But due to many reasons, for example, the complexity of software makes software vulnerabilities inevitable. Since software vulnerabilities cannot be avoided, another defense is to detect and patch them as soon as possible. Detecting vulnerabilities in software source code is an indispensable method to solve the problem of software vulnerabilities. Under this approach, many solutions use manually defined vulnerability patterns to detect vulnerabilities. However, manually defining vulnerability patterns is a subjective and tedious task, which usually cannot guarantee the false positive and false negative rates of vu...

AI Technical Summary

Problems solved by technology

[0003] In view of the above defects or improvement needs of the prior art, the purpose of the present invention is to solve the problem that the existing manually defined detection loophole mode is a subjective and cumbersome task, which usually cannot guarantee the false positive rate and false negative rate of the loophole detection results question

Images