Soteria: Preserving Privacy in Distributed Machine Learning

Cláudia Brito; Pedro Ferreira; Bernardo Portela; Rui Oliveira; João Paulo

Paper 2021/966

Soteria: Preserving Privacy in Distributed Machine Learning

Cláudia Brito

Pedro Ferreira

Bernardo Portela

Rui Oliveira

João Paulo

Abstract

In more detail, we propose Soteria, a system for distributed privacy-preserving ML that leverages Trusted Execution Environments (e.g., Intel SGX) to run computations over sensitive information in isolated containers (enclaves). Unlike previous work, where all ML-related computation is performed at trusted enclaves, we introduce a hybrid scheme, combining computation done inside and outside these enclaves. The experimental evaluation validates that our approach reduces the runtime of ML algorithms by up to 41% compared to previous related work. Our protocol is accompanied by a security proof and a discussion regarding resilience against a wide spectrum of ML attacks

Metadata

Available format(s): PDF
Publication info: Published elsewhere. The 38th ACM/SIGAPP Symposium On Applied Computing (SAC'23)
DOI: 10.1145/3555776.3578591
Keywords: Privacy-preserving Machine Learning Apache Spark Confidential Computing Intel SGX
Contact author(s): claudia v brito @ inesctec pt
History: 2023-07-21: last of 5 revisions; 2021-07-22: received; See all versions
Short URL: https://ia.cr/2021/966
License: CC BY

BibTeX

@misc{cryptoeprint:2021/966,
      author = {Cláudia Brito and Pedro Ferreira and Bernardo Portela and Rui Oliveira and João Paulo},
      title = {Soteria: Preserving Privacy in Distributed Machine Learning},
      howpublished = {Cryptology ePrint Archive, Paper 2021/966},
      year = {2021},
      doi = {10.1145/3555776.3578591},
      note = {\url{https://eprint.iacr.org/2021/966}},
      url = {https://eprint.iacr.org/2021/966}
}