Search for potentially dangerous constructs in the source code of the program using neural networks

This work is devoted to the study of artificial neural networks to search for potentially dangerous constructs in the source code of the program. A modification of the existing analyzer programs is proposed for a better search for vulnerabilities in software. The use of neural networks will allow programmers and analysts to find those vulnerabilities that have not yet been added to the database. As a result of the research, it was found that the use of neural networks to identify specified constructs is a more effective solution than searching through a database of vulnerabilities, since it allows you to identify not only those constructs that are in the database, but also suspicious code sections.