GPU Acceleration of Sparse Fully Homomorphic Encrypted DNNs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: D'Agata, Lara, Agulló-Domingo, Carlos, Vera-López, Óscar, Shivdikar, Kaustubh, Yudha, Ardhi W. B., Yaman, Ferhat, Kaeli, David, Abellán, José L., Colbert, Ian, Cano, José
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913026183004160
author D'Agata, Lara
Agulló-Domingo, Carlos
Vera-López, Óscar
Shivdikar, Kaustubh
Yudha, Ardhi W. B.
Yaman, Ferhat
Kaeli, David
Abellán, José L.
Colbert, Ian
Cano, José
author_facet D'Agata, Lara
Agulló-Domingo, Carlos
Vera-López, Óscar
Shivdikar, Kaustubh
Yudha, Ardhi W. B.
Yaman, Ferhat
Kaeli, David
Abellán, José L.
Colbert, Ian
Cano, José
contents Fully homomorphic encryption (FHE) has recently attracted significant attention as both a cryptographic primitive and a systems challenge. Given the latest advances in accelerated computing, FHE presents a promising opportunity for progress, with applications ranging from machine learning to information security. We target the most computationally intensive operation in deep neural networks from a hardware perspective, matrix multiplication (matmul), and adapt it for execution on AMD GPUs. We propose a new optimized method that improves the runtime and complexity of ciphertext matmul by using FIDESlib, a recent open-source FHE library designed specifically for GPUs. By exploiting sparsity in both operands, our sparse matmul implementation outperforms its CPU counterpart by up to $3.0\times$ and reduces the time complexity from cubic to semi-linear, demonstrating an improvement over existing FHE matmul implementations.
format Preprint
id arxiv_https___arxiv_org_abs_2604_11659
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GPU Acceleration of Sparse Fully Homomorphic Encrypted DNNs
D'Agata, Lara
Agulló-Domingo, Carlos
Vera-López, Óscar
Shivdikar, Kaustubh
Yudha, Ardhi W. B.
Yaman, Ferhat
Kaeli, David
Abellán, José L.
Colbert, Ian
Cano, José
Cryptography and Security
Distributed, Parallel, and Cluster Computing
Data Structures and Algorithms
Machine Learning
Performance
Fully homomorphic encryption (FHE) has recently attracted significant attention as both a cryptographic primitive and a systems challenge. Given the latest advances in accelerated computing, FHE presents a promising opportunity for progress, with applications ranging from machine learning to information security. We target the most computationally intensive operation in deep neural networks from a hardware perspective, matrix multiplication (matmul), and adapt it for execution on AMD GPUs. We propose a new optimized method that improves the runtime and complexity of ciphertext matmul by using FIDESlib, a recent open-source FHE library designed specifically for GPUs. By exploiting sparsity in both operands, our sparse matmul implementation outperforms its CPU counterpart by up to $3.0\times$ and reduces the time complexity from cubic to semi-linear, demonstrating an improvement over existing FHE matmul implementations.
title GPU Acceleration of Sparse Fully Homomorphic Encrypted DNNs
topic Cryptography and Security
Distributed, Parallel, and Cluster Computing
Data Structures and Algorithms
Machine Learning
Performance
url https://arxiv.org/abs/2604.11659