MALWARE DETECTION BASED ON DEEP LEARNING AND BEHAVIOR GRAPHS

MALWARE DETECTION BASED ON DEEP LEARNING AND BEHAVIOR GRAPHS

Dr. Thayyaba Khatoon, V.Sravanthi, Sri Likhita.G, Abhilash.G, Akshay.G, Anil Kumar.G

Artificial Intelligence & Machine Learning

Malla Reddy University,

Hyderabad

Abstract
One of the most common cyberattacks is malware, and it is becoming more and more common across the network every day. In contrast to benign traffic, which is always symmetrical, malware traffic is always asymmetrical. Fortunately, a variety of artificial intelligence approaches are available for identifying malware and differentiating it from everyday operations. The goal of this project is to develop a reliable and effective malware detection system for computer networks. The increasingly complex and polymorphic malware threats cannot be detected using conventional signature-based malware detection techniques. Instead of depending on specific signatures, the suggested deep learning-based behaviour graph technique seeks to solve this issue by examining the behavioural patterns of malware. The project models the behaviour of malware using graph neural networks. By examining the behaviour graphs of both good and bad software, the model picks up on the traits of malware. The system calls and API functions utilised by the software are represented by behaviour graphs, which the model utilises to find malicious activity.

Keywords: Deep Learning, Malware detection, Behavioural graphs, Malware, Benign.