Confidential Computing of Machine Learning using Intel SGX

Master's Thesis


View Demo · Report Bug

Table of Contents

• About the Project
  • Built With
• Getting Started
  • Prerequisites
  • Installation

About The Project

A TEE-based application to securely upload and train multiparty data-sets with confidential computing.

Thesis reprort

Built With

This platform was built with

• c++
• WinAPI
• Intel SGX

Getting Started

Steps to run the software locally

Prerequisites

• Intel SGX capable systems or Visual Studio 2017 with Intel SGX SDK
• Microsoft Windows 10 64-bit for Intel SGX support
• Intel processor that supports Intel Data Protection Technology(DPT) with Secure Key

Installation

1. Install Visual Studio 2017 along with Intel SGX SDK
2. Clone the repo

git clone https://github.com/prasadkjose/Confidential_ML_SGX.git

3. Import the project in Visual Studio.
4. Run the application in debug mode or simulation mode.

Usage

There are two applications

• Encrypter

  • Open the Encrypter application

    

  • Login

    

  • Upload the Datasets

    

  • Once the files are uploaded, they can be viewed in the console and also listed under "Encrypted Files"

    

• ML Uploader

  • Open the ML Uploader application.

    

  • Log in and upload the datasets.

    

  • Click Train to train the model