Sphynx: A Deep Neural Network Design for Private Inference

Minsu Cho; Zahra Ghodsi; Brandon Reagen; Siddharth Garg; Chinmay Hegde

doi:10.1109/MSEC.2022.3165475

Sphynx: A Deep Neural Network Design for Private Inference

Minsu Cho, Zahra Ghodsi, Brandon Reagen, Siddharth Garg, Chinmay Hegde

Electrical and Computer Engineering

Research output: Contribution to journal › Article › peer-review

Abstract

Private inference (PI) methods involving deep networks enable cryptographically secure machine learning but incur severe latency costs. We outline how neural architecture searches are used to design networks with fewer nonlinearities, enabling efficient PI.

Original language	English (US)
Pages (from-to)	22-34
Number of pages	13
Journal	IEEE Security and Privacy
Volume	20
Issue number	5
DOIs	https://doi.org/10.1109/MSEC.2022.3165475
State	Published - 2022

ASJC Scopus subject areas

Computer Networks and Communications
Electrical and Electronic Engineering
Law

Access to Document

10.1109/MSEC.2022.3165475

Cite this

@article{fa2dd23a1b1d4a398f9c44dae6aff4ac,

title = "Sphynx: A Deep Neural Network Design for Private Inference",

abstract = "Private inference (PI) methods involving deep networks enable cryptographically secure machine learning but incur severe latency costs. We outline how neural architecture searches are used to design networks with fewer nonlinearities, enabling efficient PI.",

author = "Minsu Cho and Zahra Ghodsi and Brandon Reagen and Siddharth Garg and Chinmay Hegde",

note = "Publisher Copyright: {\textcopyright} 2003-2012 IEEE.",

year = "2022",

doi = "10.1109/MSEC.2022.3165475",

language = "English (US)",

volume = "20",

pages = "22--34",

journal = "IEEE Security and Privacy",

issn = "1540-7993",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "5",

}

TY - JOUR

T1 - Sphynx

T2 - A Deep Neural Network Design for Private Inference

AU - Cho, Minsu

AU - Ghodsi, Zahra

AU - Reagen, Brandon

AU - Garg, Siddharth

AU - Hegde, Chinmay

PY - 2022

Y1 - 2022

N2 - Private inference (PI) methods involving deep networks enable cryptographically secure machine learning but incur severe latency costs. We outline how neural architecture searches are used to design networks with fewer nonlinearities, enabling efficient PI.

AB - Private inference (PI) methods involving deep networks enable cryptographically secure machine learning but incur severe latency costs. We outline how neural architecture searches are used to design networks with fewer nonlinearities, enabling efficient PI.

UR - http://www.scopus.com/inward/record.url?scp=85131759988&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85131759988&partnerID=8YFLogxK

U2 - 10.1109/MSEC.2022.3165475

DO - 10.1109/MSEC.2022.3165475

M3 - Article

AN - SCOPUS:85131759988

SN - 1540-7993

VL - 20

SP - 22

EP - 34

JO - IEEE Security and Privacy

JF - IEEE Security and Privacy

IS - 5

ER -

Sphynx: A Deep Neural Network Design for Private Inference

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this