# Differentially-Private "Draw and Discard" Machine Learning

arXiv: https://arxiv.org/abs/1807.04369 · HTML: https://mlsanity.com/paper/1807.04369/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Jul 2018.

## Built on by (most cited first, 7 of 7)

- [A Critical Review on the Use (and Misuse) of Differential Privacy in Machine Learning](https://mlsanity.com/paper/2206.04621/) (2022)
- [RoFL: Robustness of Secure Federated Learning](https://mlsanity.com/paper/2107.03311/) (2021)
- [DP-REC: Private & Communication-Efficient Federated Learning](https://mlsanity.com/paper/2111.05454/) (2021)
- [PRECAD: Privacy-Preserving and Robust Federated Learning via Crypto-Aided Differential Privacy](https://mlsanity.com/paper/2110.11578/) (2021)
- [On the Practicality of Differential Privacy in Federated Learning by Tuning Iteration Times](https://mlsanity.com/paper/2101.04163/) (2021)
- [Optimizing the Numbers of Queries and Replies in Federated Learning with Differential Privacy](https://mlsanity.com/paper/2107.01895/) (2021)
- [Facebook Report on Privacy of fNIRS data](https://mlsanity.com/paper/2401.00973/) (2024)

## Abstract (arXiv)

In this work, we propose a novel framework for privacy-preserving client-distributed machine learning. It is motivated by the desire to achieve differential privacy guarantees in the local model of privacy in a way that satisfies all systems constraints using asynchronous client-server communication and provides attractive model learning properties. We call it "Draw and Discard" because it relies on random sampling of models for load distribution (scalability), which also provides additional server-side privacy protections and improved model quality through averaging. We present the mechanics of client and server components of "Draw and Discard" and demonstrate how the framework can be applied to learning Generalized Linear models. We then analyze the privacy guarantees provided by our approach against several types of adversaries and showcase experimental results that provide evidence for the framework's viability in practical deployments.