Machine/Deep Learning for obfuscated malware Detection

The protection of digital systems and sensitive data has become paramount due to the appearance of new intelligent cybersecurity threats. The landscape of cyber threats is constantly evolving, becoming more sophisticated and diverse by the day. This dynamic environment challenges security measures, demanding constant vigilance and innovation to safeguard against emerging risks. This research represents an advancement in fortifying defenses against obfuscated malware. Its goal is to elevate the accuracy of detecting and classifying these sophisticated threats, bolstering our ability to identify and mitigate their impact effectively. The core objective is to differentiate between harmful and benign memory dumps using various deep/machine learning models. The models are designed to specialize in identifying obfuscated malware, especially malware that skillfully conceals its presence to evade detection and removal, thereby countering a range of threats, including spyware, ransomware, and Trojan horse malware. As a result, we could deploy various types of ML/DL algorithms with more than 85% precision in multiclass overlapping and imbalanced data. In addition, different types of data balancing were exploited, and each technique and its effect on the results were evaluated separately.

• malware classification
• memory dumps
• deep learning
• machine learning
• spyware
• ransomware
• class imbalance
• data balancing
• malware evasion

• Advanced Malware Detection Techniques
• Security and Verification in Computing
• Digital and Cyber Forensics
• malware classification
• memory dumps
• deep learning
• machine learning
• spyware
• ransomware
• class imbalance
• data balancing
• Signal Processing
• Artificial Intelligence
• Information Systems
• Computer Science