Efficient Pruning for Machine Learning Under Homomorphic Encryption

Ehud Aharoni, Moran Baruch, Pradip Bose, Alper Buyuktosunoglu, Nir Drucker, Subhankar Pal, Tomer Pelleg, Kanthi Sarpatwar, Hayim Shaul, Omri Soceanu, Roman Vaculin

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Privacy-preserving machine learning (PPML) solutions are gaining widespread popularity. Among these, many rely on homomorphic encryption (HE) that offers confidentiality of the model and the data, but at the cost of large latency and memory requirements. Pruning neural network (NN) parameters improves latency and memory in plaintext ML but has little impact if directly applied to HE-based PPML. We introduce a framework called HE-PEx that comprises new pruning methods, on top of a packing technique called tile tensors, for reducing the latency and memory of PPML inference. HE-PEx uses permutations to prune additional ciphertexts, and expansion to recover inference loss. We demonstrate the effectiveness of our methods for pruning fully-connected and convolutional layers in NNs on PPML tasks, namely, image compression, denoising, and classification, with autoencoders, multilayer perceptrons (MLPs) and convolutional neural networks (CNNs). We implement and deploy our networks atop a framework called HElayers, which shows a 10–35% improvement in inference speed and a 17–35% decrease in memory requirement over the unpruned network, corresponding to 33–65% fewer ciphertexts, within a 2.5% degradation in inference accuracy over the unpruned network. Compared to the state-of-the-art pruning technique for PPML, our techniques generate networks with 70% fewer ciphertexts, on average, for the same degradation limit.

Original languageEnglish
Title of host publicationComputer Security – ESORICS 2023 - 28th European Symposium on Research in Computer Security, The Hague, The Netherlands, September 25–29, 2023, Proceedings
EditorsGene Tsudik, Mauro Conti, Kaitai Liang, Georgios Smaragdakis
PublisherSpringer Science and Business Media Deutschland GmbH
Pages204-225
Number of pages22
ISBN (Print)9783031514814
DOIs
StatePublished - 2024
Externally publishedYes
Event28th European Symposium on Research in Computer Security, ESORICS 2023 - The Hague, Netherlands
Duration: 25 Sep 202329 Sep 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14347 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference28th European Symposium on Research in Computer Security, ESORICS 2023
Country/TerritoryNetherlands
CityThe Hague
Period25/09/2329/09/23

Bibliographical note

Publisher Copyright:
© 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Keywords

  • Homomorphic encryption
  • Machine learning
  • Neural networks
  • Privacy-preserving computation
  • Pruning
  • Tile tensors

Fingerprint

Dive into the research topics of 'Efficient Pruning for Machine Learning Under Homomorphic Encryption'. Together they form a unique fingerprint.

Cite this