Classification of Malware Families Using Transformers and Auto Formers in Deep Learning

Malicious code in computer programming refers to programs intended to harm a system by exploiting vulnerabilities, potentially resulting in security breaches and data damage. The field of information security increasingly focuses on detecting malicious software threats. Recent studies aim to improve malware detection methods. As the cyber threat landscape continues to evolve, the detection and classification of malware have become increasingly challenging. Traditional signature-based methods struggle to keep up with the proliferation of new malware variants. In this context, deep learning has emerged as a powerful tool for classifying malware families. This paper presents a new method for categorizing virus into distinct families utilizing neural learning techniques. Antivirus is always creating new versions and evolving in the real world. Numerous academics have created solutions utilizing artificial intelligence methods to deal with this problem. But up until now, these algorithms have struggled to adapt to the ever-evolving virus landscape and had issues with incorrect data in the initial data set. The amount of malware is growing at a rapid pace. Certain types of malwares can evade detection by employing various obscuring strategies. Before it infects an immense number of machines, the malware must be found in order to safeguard the machines and the World Wide Web from it. Many studies on methods for detecting malware have been conducted lately. Virus identification, still remains a challenge. Both heuristic- and based on signatures detection techniques are quick and effective in finding known malware, but signature-based techniques in particular have not been able to find unidentified malware. However, for unfamiliar and complex malware, behavior-based, model-checking, and stored in the cloud approaches work well; deep learning-based, mobile device-based, and Internet of Things-based approaches also surface to identify a certain amount of recognized.