Skip to content

mkskeller/EzPC

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CrypTFlow: An End-to-end System for Secure TensorFlow Inference.

Nishant Kumar, Mayank Rathee, Nishanth Chandran, Divya Gupta, Aseem Rastogi, Rahul Sharma

CrypTFlow paper link: eprint, arXiv.

EzPC paper link: eprint

Project webpage: https://aka.ms/ezpc.

Build Status License: MIT contributions welcome

HitCount GitHub contributors GitHub All Releases

GitHub repo size GitHub language count Maintenance

GitHub search hit counter GitHub search hit counter GitHub issues

Introduction

CrypTFlow consists of 3 components:

  • Athos: an end-to-end compiler from TensorFlow to a variety of semi-honest MPC protocols. Athos leverages EzPC as a low-level intermediate language.
  • Porthos: a semi-honest 3 party computation protocol which is geared towards TensorFlow-like applications.
  • Aramis: a novel technique that uses hardware with integrity guarantees to convert any semi-honest MPC protocol into an MPC protocol that provides malicious security.

With these components in place, we are able to run for the first time secure inference on the ImageNet dataset with the pre-trained models of the following deep neural nets: ResNet-50, DenseNet-121 and SqueezeNet for ImageNet.

In particular, this repository contains the code for the following components:

  • EzPC
  • Athos (part of CrypTFlow)
  • Porthos (part of CrypTFlow)
  • Aramis (part of CrypTFlow)

Each one of the above is independent and usable in their own right and more information can be found in the readme of each of the components. But together these combine to make CrypTFlow a powerful system for end-to-end secure inference of deep neural networks written in TensorFlow.

For setup instructions, please refer to each of the components' readme. We plan to release a docker version of the system as well which will make the system easier to setup.

Wiki

Wiki section of this repository provides coding practices and examples to get started with EzPC.

Issues/Bugs

For bugs and support, please create an issue on the issues page.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 92.2%
  • Python 3.9%
  • OCaml 1.6%
  • C 1.6%
  • Shell 0.4%
  • Makefile 0.3%