On the Evaluation of Deep Learning-Based Side-Channel Analysis

On the Evaluation of Deep Learning-Based Side-Channel Analysis

Wu, L. (TU Delft Cyber Security)
Perin, G. (TU Delft Cyber Security)
Picek, S. (TU Delft Cyber Security; Radboud Universiteit Nijmegen)

Balasch, Josep (editor)
O’Flynn, Colin (editor)

2022

Deep learning-based side-channel analysis is rapidly positioning itself as a de-facto standard for the most powerful profiling side-channel analysis.The results from the last few years show that deep learning techniques can efficiently break targets that are even protected with countermeasures. While there are constant improvements in making the deep learning-based attacks more powerful, little is done on evaluating the attacks’ performance. Indeed, how the evaluation process is done today is not different from what was done more than a decade ago from the perspective of evaluation metrics. This paper considers how to evaluate deep learning-based side-channel analysis and whether the commonly used approaches give the best results. To that end, we consider different summary statistics and the influence of algorithmic randomness on the stability of profiling models. Our results show that besides commonly used metrics like guessing entropy, one should also show the standard deviation results to assess the attack performance properly. Even more importantly, using the arithmetic mean for guessing entropy does not yield the best results, and instead, a median value should be used.

Deep Learning
Guessing Entropy
Median
Side-channel Analysis

http://resolver.tudelft.nl/uuid:af113cec-3d30-4c52-939b-3b2fe8f8e911

https://doi.org/10.1007/978-3-030-99766-3_3

Springer

2022-11-30

9783030997656

Constructive Side-Channel Analysis and Secure Design - 13th International Workshop, COSADE 2022, Proceedings

13th International Workshop on Constructive Side-Channel Analysis and Secure Design, COSADE 2022, 2022-04-11 → 2022-04-12, Leuven, Belgium

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 0302-9743, 13211 LNCS

Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Institutional Repository

conference paper

© 2022 L. Wu, G. Perin, S. Picek