Uncommon Sense - Detecting Exploits with Novel Hardware Performance Counters and ML Magic

Explore cutting-edge exploit detection techniques using novel hardware performance counters (HPCs) and machine learning in this 44-minute Black Hat conference talk. Delve into the limitations of current HPC-based detection methods for attacks like speculative execution, Rowhammer, and Return Oriented Programming (ROP). Discover how relying solely on well-known counters makes detection techniques vulnerable to bypass attempts. Learn about the untapped potential of lesser-known HPCs and how they can be leveraged to create more robust and resilient exploit detection systems. Gain insights from speakers Nick Gregory and Harini Kannan on innovative approaches to cybersecurity that combine advanced hardware monitoring with machine learning algorithms.