Skip to content
Snippets Groups Projects
README.md 2.88 KiB
Newer Older
Deevashwer's avatar
Deevashwer committed
# CrypTFlow: An End-to-end System for Secure TensorFlow Inference [![contributions welcome](https://img.shields.io/badge/contributions-welcome-brightgreen.svg?style=flat)](https://github.com/mpc-msri/EzPC/issues) [![HitCount](http://hits.dwyl.io/mpc-msri/EzPC.svg)](http://hits.dwyl.io/mpc-msri/EzPC)
Deevashwer's avatar
Deevashwer committed
**Reference Papers:**  
Mayank's avatar
Mayank committed

Deevashwer's avatar
Deevashwer committed
[CrypTFlow2: Practical 2-Party Secure Inference](https://eprint.iacr.org/2020/1002)  
Deevashwer's avatar
Deevashwer committed
Deevashwer Rathee, Mayank Rathee, Nishant Kumar, Nishanth Chandran, Divya Gupta, Aseem Rastogi, Rahul Sharma  
*ACM CCS 2020*
Nishant Kumar's avatar
Nishant Kumar committed

Deevashwer's avatar
Deevashwer committed
[CrypTFlow: Secure TensorFlow Inference](https://eprint.iacr.org/2019/1049)  
Nishant Kumar, Mayank Rathee, Nishanth Chandran, Divya Gupta, Aseem Rastogi, Rahul Sharma  
*IEEE S&P 2020*
Nishant Kumar's avatar
Nishant Kumar committed

Deevashwer's avatar
Deevashwer committed
[EzPC: Programmable, Efficient, and Scalable Secure Two-Party Computation for Machine Learning](https://eprint.iacr.org/2017/1109.pdf)  
Nishanth Chandran, Divya Gupta, Aseem Rastogi, Rahul Sharma, Shardul Tripathi  
*IEEE EuroS&P 2019*
Mayank's avatar
Mayank committed

Deevashwer's avatar
Deevashwer committed
**Project webpage:** <https://aka.ms/ezpc>
Mayank's avatar
Mayank committed

Nishant Kumar's avatar
Nishant Kumar committed
## Introduction
Deevashwer's avatar
Deevashwer committed
This repository has the following components:  
Deevashwer's avatar
Deevashwer committed
- **EzPC**: a language for secure machine learning.
- **Athos** (part of **CrypTFlow**): 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** (part of **CrypTFlow**): a semi-honest 3 party computation protocol which is geared towards TensorFlow-like applications.
- **Aramis** (part of **CrypTFlow**): a novel technique that uses hardware with integrity guarantees to convert any semi-honest MPC protocol into an MPC protocol that provides malicious security.
- **SCI** (part of **CrypTFlow2**): a semi-honest 2-party computation library for secure inference on deep neural networks.
Mayank's avatar
Mayank committed
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.
Deevashwer's avatar
Deevashwer committed
With these components in place, we are able to run for the first time secure inference on the [ImageNet dataset]([http://www.image-net.org) with the pre-trained models of the following deep neural nets: ResNet-50, DenseNet-121 and SqueezeNet for ImageNet.

For setup instructions, please refer to each of the components' readme.

Alternatively you can use the **setup_env_and_build.sh** script. It installs dependencies and builds each component. It also creates a virtual environment in a *mpc_venv* folder with all the required packages.

Please do ``source mpc_venv/bin/activate`` before using the toolchain.

We plan to release a docker version of the system as well which will make the system easier to setup.

Mayank's avatar
Mayank committed
## Wiki
Wiki section of this repository provides coding practices and examples to get started with EzPC.

## Issues/Bugs
Nishant Kumar's avatar
Nishant Kumar committed
For bugs and support, please create an issue on the issues page.