0x10: Solomon Sonya - Harnessing the Power of Machine Learning to Enhance Malware...
Provides 0x10: Solomon Sonya - Harnessing the Power of Machine Learning to Enhance Malware and evaluates key implications for security practitioners.
Summary
Motivation and scale
Solomon Sonya at SEC-T argues antivirus, SOC tooling, and compliance alone cannot keep pace with malware volume. He cites VirusTotal receiving over two million samples daily, with nearly two million distinct samples, and says new detection strategies must demystify how AI/ML/NLP convert artifacts into decisions.
Research gaps
His work targets four gaps: a uniform language for static, dynamic, and memory artifacts; an automated analysis pipeline producing standardized datasets; continuously updated benchmarks instead of decade-old corpora; and heterogeneous ensembles that raise detection precision without sacrificing accuracy.
ML foundations
Sonya walks through supervised versus unsupervised learning, labels, generalization, and basic model intuition using regression examples. He emphasizes that sufficiently advanced ML should not be treated as magic, and that personal research views do not represent the U.S. Air Force or DoD. The truncated session continues into classification and model families for malware-of-tomorrow detection.
Related talks
Black Hat·55 min
Analysts investigate attack surfaces and defenses concerning Harnessing Intelligence From Malware Repositories.
Black Hat·41 min
Marc Blanchou
Industry experts analyze emerging threats surrounding EU 2013 - Harnessing GP²Us - Building Better Browser Based Botnets.
Ekoparty Security Conference·40 min
C. Urcuqui · J. Angulo
Evaluates ransomware y Machine Learning: Un Viaje por el Lado Oscuro de la IA and evaluates key implications for security practitioners.