Private Machine Learning

Table of Contents

• Overview
• Project Descriptions
• Publications

Overview

This repo summarizes the private machine learning work of Xtra group. Currently we work mainly on two areas: federated learning and differential privacy. Federated learning enables the collaborative learning of multiple parties without exchanging the local data.

Project Descriptions

We have worked/are working on the following projects.

(1) Federated Learning Survey: We conducted a survey on federated learning systems.

(2) Federated Gradient Boosting Decision Trees: We designed a novel federated learning framework for gradient boosting decision trees.

(3) Differentially Private Gradient Boosting Decision Trees: We designed a differentially private gradient boosting decision tree training algorithm.

(4) Federated Learning Benchmarks: We designed a benchmark for evaluating the components in different FL systems.

Publications

• A Survey on Federated Learning Systems: Vision, Hype and Reality for Data Privacy and Protection
  Qinbin Li, Zeyi Wen, Zhaomin Wu, Sixu Hu, Naibo Wang, Bingsheng He
  arXiv preprint

  • We conducted a comprehensive analysis against existing federated learning systems from different aspects (see details).

• Practical Federated Gradient Boosting Decision Trees
  Qinbin Li, Zeyi Wen, Bingsheng He
  Thirty-Fourth AAAI Conference on Artificial Intelligence. AAAI 2020.

  • We proposed a novel federated learning framework for gradient boosting decision trees by exploiting similarity (see details).

• Privacy-Preserving Gradient Boosting Decision Trees
  Qinbin Li, Zhaomin Wu, Zeyi Wen, Bingsheng He
  Thirty-Fourth AAAI Conference on Artificial Intelligence. AAAI 2020.

  • We designed a new differentially private gradient boosting decision trees training algorithm (see details).

• The OARF Benchmark Suite: Characterization and Implications for Federated Learning Systems
  Sixu Hu, Yuan Li, Xu Liu, Qinbin Li, Zhaomin Wu, Bingsheng He
  arXiv preprint.

  • We designed a benchmark for evaluating the components in different FL systems (see details, code).