Paper 2025/1970
Delving into Cryptanalytic Extraction of PReLU Neural Networks
, Institute for Advanced Study, Tsinghua University, Beijing, China, Institute for Network Sciences and Cyberspace, Tsinghua University, Beijing, China, Department of Computer Science and Technology, Tsinghua University, Beijing, China, Department of Computer Science and Technology, Tsinghua University, Beijing, China, Institute for Advanced Study, Tsinghua University, Beijing, China, Department of Computer Science and Technology, Tsinghua University, Beijing, China, Institute for Advanced Study, Tsinghua University, Beijing, ChinaAbstract
The machine learning problem of model extraction was first introduced in 1991 and gained prominence as a cryptanalytic challenge starting with Crypto 2020. For over three decades, research in this field has primarily focused on ReLU-based neural networks. In this work, we take the first step towards the cryptanalytic extraction of PReLU neural networks, which employ more complex nonlinear activation functions than their ReLU counterparts. We propose a raw output-based parameter recovery attack for PReLU networks and extend it to more restrictive scenarios where only the top-m probability scores are accessible. Our attacks are rigorously evaluated through end-to-end experiments on diverse PReLU neural networks, including models trained on the MNIST dataset. To the best of our knowledge, this is the first practical demonstration of PReLU neural network extraction across three distinct attack scenarios.
Metadata
- Available format(s)
-
PDF
- Category
- Attacks and cryptanalysis
- Publication info
- A minor revision of an IACR publication in ASIACRYPT 2025
- Keywords
- cryptanalysismodel parameter recovery attackneural networkPReLU activation
- Contact author(s)
-
chenyi2023 @ tsinghua edu cn
xiaoyangdong @ tsinghua edu cn
marj21 @ mails tsinghua edu cn
shenyt22 @ mails tsinghua edu cn
anyuwang @ tsinghua edu cn
yuhongbo @ tsinghua edu cn
xiaoyunwang @ tsinghua edu cn - History
- 2025-10-24: approved
- 2025-10-21: received
- See all versions
- Short URL
- https://ia.cr/2025/1970
- License
-
CC BY
BibTeX
@misc{cryptoeprint:2025/1970,
author = {Yi Chen and Xiaoyang Dong and Ruijie Ma and Yantian Shen and Anyu Wang and Hongbo Yu and Xiaoyun Wang},
title = {Delving into Cryptanalytic Extraction of {PReLU} Neural Networks},
howpublished = {Cryptology {ePrint} Archive, Paper 2025/1970},
year = {2025},
url = {https://eprint.iacr.org/2025/1970}
}
