Machine Learning in Computer Security is Difficult to Fix
Battista Biggio · Communications of the ACM · 2024
This study is timely and exciting as it clearly points out some common issues hindering the design of ML models for computer security and how to overcome them. This witnesses once again that, despite the impressive performance reported in many of the published papers in this area, the reality is quite different, and applying ML in computer security is much more challenging than it may seem.