Side-Channel Attacks are attacks against implementations of cryptographic algorithms. These attacks exploit physical properties of a device under attack. For example an attacker can measure the execution time or power consumption of a device while it executes a cryptographic algorithm.
Based on neural network, deep learning represents an active research in machine learning that allows producing automatic attacks requiring no a priori information on the underlying phenomenon. The purpose of this work is to shed new light on the capabilities of deep learning in side-channel attacks.
This work is in collaboration with RISCURE (www.riscure.com), a company working on security evaluation of embedded devices.