EvadeSafe-XAI: A Robust Malware Defense System using Explainable AI
Pendli Ramya, Kundaram Divya, Bhavya Sri Kurmala, Vaishnavi Maitreyi Poranki, Medikonda Asha Kiran, Peddada Nagamani · 2025
Adopting machine learning (ML) techniques in malware detection has significantly improved the automated identification of malicious software. However, these systems remain susceptible to evasion attacks, where adversaries modify malware to bypass detection mechanisms. The EvadeSafe-XAI framework integrates adversarial training and robust optimization techniques to address this challenge, thereby enhancing detection accuracy and resilience. EvadeSafe-XAI employs deep learning-based malware and adversary detectors to effectively counter sophisticated attacks while maintaining high detection rates for benign and malicious files. This approach strengthens malware detection systems against adversarial threats, ensuring more reliable cybersecurity defenses.