Homomorphic ReLU with Full-Domain Bootstrapping
Cryptography · 24.03.2026 · 10 (2) · 21
DOI: 10.3390/cryptography10020021 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cryptography, online erschienen am 24.03.2026. Verfasst von Yuqun Lin und 7 weiteren Autor:innen. Die Arbeit nennt 18 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Shenzhen Science and Technology Program, National Natural Science Foundation of China https://ror.org/01h0zpd94, Shenzhen Social Science Fund.
Abstract (Originalsprache)
Fully homomorphic encryption (FHE) offers a promising solution for privacy-preserving machine learning by enabling arbitrary computations on encrypted data. However, the efficient evaluation of non-linear functions—such as the ReLU activation function over large integers—remains a major obstacle in practical deployments, primarily due to high bootstrapping overhead and limited precision support in existing schemes. In this paper, we propose LargeIntReLU, a novel framework that enables efficient homomorphic ReLU evaluation over large integers (7–11 bits) via full-domain bootstrapping. Central to our approach is a signed digit decomposition algorithm, SignedDecomp, that partitions a large integer ciphertext into signed 6-bit segments using three new low-level primitives: LeftShift, HomMod, and CipherClean. This decomposition preserves arithmetic consistency, avoids cross-segment carry propagation, and allows parallelized bootstrapping. By segmenting the large integer and processing each chunk independently with optimized small-integer bootstrapping, we achieve homomorphic ReLU with full-domain bootstrapping, which significantly reduces the total number of sequential bootstrapping operations required. The security of our scheme is guaranteed by TFHE. Experimental results demonstrate that the proposed method reduces the bootstrapping cost by an average of 28.58% compared to state-of-the-art approaches while maintaining 95.2% accuracy. With execution times ranging from 1.16 s to 1.62 s across 7–11 bit integers, our work bridges a critical gap toward a scalable and efficient homomorphic ReLU function, which is useful in privacy-preserving machine learning. Furthermore, an end-to-end encrypted inference test on a CNN model with the MNIST dataset confirms its practicality, achieving 88.85% accuracy and demonstrating a complete pipeline for privacy-preserving neural network evaluation.
Förderung
- Shenzhen Science and Technology Program
- National Natural Science Foundation of China https://ror.org/01h0zpd94
- Shenzhen Social Science Fund
- PCL-CMCC Foundation for Science and Innovation
Zitieren
APA
Lin, Y., Huang, Y., Tang, X., Fan, J., Xu, Q., Jiang, Z., Zhang, X., & Fang, J. (2026). Homomorphic ReLU with Full-Domain Bootstrapping. Cryptography, 10(2), 21. https://doi.org/10.3390/cryptography10020021
Vancouver
Lin Y, Huang Y, Tang X, Fan J, Xu Q, Jiang Z, et al. Homomorphic ReLU with Full-Domain Bootstrapping. Cryptography. 2026;10(2):21. doi:10.3390/cryptography10020021
BibTeX
@article{Lin2026,
author = {Lin, Yuqun and Huang, Yi and Tang, Xiaomeng and Fan, Jingjing and Xu, Qifei and Jiang, Zoe-Lin and Zhang, Xiaosong and Fang, Junbin},
title = {Homomorphic ReLU with Full-Domain Bootstrapping},
journal = {Cryptography},
year = {2026},
volume = {10},
number = {2},
pages = {21},
doi = {10.3390/cryptography10020021}
}
EUVIMED