Skip to content

Latest commit

 

History

History
executable file
·
15 lines (13 loc) · 795 Bytes

README.md

File metadata and controls

executable file
·
15 lines (13 loc) · 795 Bytes
Raw

Secure XGBoost Colloboration App

A Secure XGBoost application, created using Streamlit. Using this, you can securely train models and get inference. This app simulates HW Enclaves to demonstrate how secure collaboration works.

Pre requisites

  1. Linux (at least 18.04 version)
  2. Docker images for server and client

Guide

  1. Load docker images using docker load image.tar
  2. Start server from terminal using bash run_server.sh
  3. Add user names you want to run the server for (default: user1 & user2)
  4. Start the client containers using bash run_client1.sh and bash run_client2.sh in two different terminal windows
  5. Run the app by clicking on link present on terminal
  6. Generate keys, train collaboratively, and get predictions with classification metrics through the app