| Presentation | 2021-04-12 On the Performance Evaluation of Deep-Learning Based Side-Channel Attacks Akira Ito, Rei Ueno, Naofumi Homma, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | This paper presents a method for estimating the lower bound of success rate (SR) and the upper bound of guessing entropy (GE) on deep-learning-based side-channel attacks (DL-SCAs) for the purpose of direct and quantitative performance evaluation. In conventional side-channel attacks, SR and GE are widely used as indicators for measuring the efficiency of attacks. On the other hand, in DL-SCA, it is pointed out that performance evaluation metrics generally used in machine learning such as Accuracy and Precision are not effective in estimating SR and GE. In this paper, we consider that the negative log-likelihood used in DL-SCA can be reduced to the sum of independent random variables, and derive a tighter GE upper bound and SR lower bound using a probability concentration inequality. Through attack experiments on different data sets, we confirm the effectiveness of the upper bound of GE and the lower bound of SR derived by the proposed method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Side-channel attacks / Deep learning / Probability concentration inequalities |
| Paper # | HWS2021-8 |
| Date of Issue | 2021-04-05 (HWS) |
| Conference Information | |
| Committee | HWS |
|---|---|
| Conference Date | 2021/4/12(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Tokyo University/Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Hardware Security |
| Chair | Makoto Ikeda(Univ. of Tokyo) |
| Vice Chair | Yasuhisa Shimazaki(Renesas Electronics) / Makoto Nagata(Kobe Univ.) |
| Secretary | Yasuhisa Shimazaki(Kyushu Univ.) / Makoto Nagata(NTT) |
| Assistant | |
| Paper Information | |
| Registration To | Technical Committee on Hardware Security |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | On the Performance Evaluation of Deep-Learning Based Side-Channel Attacks |
| Sub Title (in English) | |
| Keyword(1) | Side-channel attacks |
| Keyword(2) | Deep learning |
| Keyword(3) | Probability concentration inequalities |
| 1st Author's Name | Akira Ito |
| 1st Author's Affiliation | Tohoku University(Tohoku Univ.) |
| 2nd Author's Name | Rei Ueno |
| 2nd Author's Affiliation | Tohoku University(Tohoku Univ.) |
| 3rd Author's Name | Naofumi Homma |
| 3rd Author's Affiliation | Tohoku University(Tohoku Univ.) |
| Date | 2021-04-12 |
| Paper # | HWS2021-8 |
| Volume (vol) | vol.121 |
| Number (no) | HWS-1 |
| Page | pp.pp.33-38(HWS), |
| #Pages | 6 |
| Date of Issue | 2021-04-05 (HWS) |