Few-Shot Machine Learning for Side-Channel Analysis

Brodin, Robert

Student Work

Few-Shot Machine Learning for Side-Channel Analysis

Público

Side-channel Attacks (SCA) is a fast growing subfield of Machine Learning. Machine Learning models have been shown to be effective classifiers when applied to SCA. One of the issues of SCA is the difficulty of data collection, which is often remedied in the field of Machine Learning with Few-shot Learning models. We propose using Prototypical Networks, a few-shot learning algorithm, to implement Side-channel attacks on hardware. The Prototypical Network will be trained on the ASCAD dataset, which contains power trace data collected from a Side-channel attack. We then modify the inputs of the training and testing steps in the prototypical network to be the same size as the ASCAD dataset. Lastly, we will observe the performance of prototypical networks in the scope of SCA, through a ranking function.

This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator