AI-Powered Malware Analysis: a Comparative Study of Traditional vs. AI-Based Approaches

EasyChair Preprint 14528

Abstract

This study explores the comparative effectiveness of traditional and AI-powered approaches to malware analysis. Traditional methods, including signature-based and heuristic-based techniques, have long been used to detect and mitigate malware threats. However, the rapid evolution of malware, including polymorphic and metamorphic variants, poses significant challenges to these conventional methods. In response, AI-powered approaches, such as machine learning and deep learning, have emerged as promising solutions due to their ability to identify complex patterns and adapt to new threats.

The study evaluates the strengths and limitations of both traditional and AI-based malware analysis techniques. Key aspects considered include detection accuracy, adaptability to new threats, and operational efficiency. Traditional methods are evaluated for their reliance on known signatures and predefined rules, while AI-based approaches are assessed for their capacity to learn from vast datasets, recognize novel threats, and provide dynamic defense mechanisms.

By analyzing case studies and performance metrics, the study highlights the advantages of AI-powered solutions in enhancing malware detection rates, reducing false positives, and improving overall system resilience. The findings suggest that while traditional methods remain relevant, AI-based approaches offer significant advancements in addressing the evolving malware landscape. The study concludes with recommendations for integrating AI into existing malware analysis frameworks to optimize threat detection and response.

Keyphrases: Advanced Persistent Threats (APTs), Automated Threat Response, Cyber Threat Intelligence Feeds, Machine Learning Models, Scalability in Cybersecurity, Security Incident Response, Threat Prediction, Zero-day exploits, data correlation

@booklet{EasyChair:14528,
  author    = {Edwin Frank},
  title     = {AI-Powered Malware Analysis: a Comparative Study of Traditional vs. AI-Based Approaches},
  howpublished = {EasyChair Preprint 14528},
  year      = {EasyChair, 2024}}