Deep learning based automatic modulation classification (AMC) has received significant attention owing to its potential applications in both military and civilian use cases. Recently, data-driven subsampling techniques have been utilized to overcome the challenges associated with computational complexity and training time for AMC. Beyond these direct advantages of data-driven subsampling, these methods also have regularizing properties that may improve the adversarial robustness of the modulation classifier. In this paper, we investigate the effects of an adversarial attack on an AMC system that employs deep learning models both for AMC and for subsampling. Our analysis shows that subsampling itself is an effective deterrent to adversarial attacks. We also uncover the most efficient subsampling strategy when an adversarial attack on both the classifier and the subsampler is anticipated.

ML can help provide more comprehensive context-rich detections of the few bad actors already in your network. Compromises will continue in 2018, and machine learning will continue to grow in intelligent sifting through alert information to detect them. And in some cases, the ML can help the security team automatically or semi-automatically resolve them. Unfortunately, I think this is one area in which an adversarial actor using ML has the upper-hand: ML might be creating some of the problems here in 2018, but also, ML will be used more for detecting social media manipulation and automated phishing and spear-phishing attacks. This is one area where ML isn't required to get pretty good detection.