Overview of machine learning based side-channel analysis methods
Dirmanto Jap, Jakub Breier · 2014
Recent publications have shown that there is a possibility to apply machine learning methods for side-channel analysis, mostly for profiling based attacks. In this paper, we present a brief overview of those methods, and highlight what are the improvements that might be offered. It is shown that, in most cases, the performance of these methods could outperform the classical attacks. Here, we also discuss what could be the other potential applications of the learning algorithms, for example, as feature selection or for construction of leakage model.