Differential privacy

Pre-training differentially private models with limited public data

Zhiqi Bu, Xinwei Zhang, Sheng Zha, Mingyi Hong

NeurIPS 2024

2024

The superior performance of large foundation models relies on the use of massive amounts of high-quality data, which often contain sensitive, private and copyrighted material that requires formal protection. While differential privacy (DP) is a prominent method to gauge the degree of security provided to the models, its application is commonly limited to the model fine-tuning stage, due to the performance

Machine learning

Differentially private bias-term fine-tuning of foundation models

Zhiqi Bu, Yu-Xiang Wang, Sheng Zha, George Karypis

ICML 2024

2024

We study the problem of differentially private (DP) fine-tuning of large pre-trained models — a recent privacy-preserving approach suitable for solving downstream tasks with sensitive data. Existing work has demonstrated that high accuracy is possible under strong privacy constraint, yet requires significant computational overhead or modifications to the network architecture. We propose differentially private

Machine learning

Fast differential privacy

Zhiqi Bu, Yu-Xiang Wang, Sheng Zha, George Karypis

2024

Fast Differential Privacy (fastDP) is a library that allows differentially private optimization of PyTorch models, with a few additional lines of code. The goal of this library is to make DP deep learning as similar to the standard non-private learning as possible, in terms of speed, memory cost, scalability, accuracy and hyperparameter-tuning. It supports all PyTorch optimizers, popular models in TIMM,

Machine learning

Improved differentially private regression via gradient boosting

Shuai Tang, Sergul Aydore, Michael Kearns, Saeyoung Rho, Aaron Roth, Yichen Wang, Yu-Xiang Wang, Zhiwei Steven Wu

IEEE SaTML 2024

2024

We revisit the problem of differentially private squared error linear regression. We observe that existing state- of-the-art methods are sensitive to the choice of hyperparameters — including the “clipping threshold” that cannot be set optimally in a data-independent way. We give a new algorithm for private linear regression based on gradient boosting. We show that our method consistently improves over

Security, privacy, and abuse prevention

A quick guide to Amazon’s papers at ICML 2023

Larry Hardesty

July 21, 2023

Across a range of topics, Amazon research blends the theoretical and the practical.

Machine learning

Computing on private data

Joan Feigenbaum, Mahnush Movahedi, Parker Newton

June 01, 2023

Both secure multiparty computation and differential privacy protect the privacy of data used in computation, but each has advantages in different contexts.

Security, privacy, and abuse prevention

Differential privacy for deep learning at GPT scale

Zhiqi Bu, Yu-Xiang Wang

May 11, 2023

Two papers from Amazon Web Services AI present algorithms that alleviate the intensive hyperparameter search and fine-tuning required by privacy-preserving deep learning at very large scales.

Security, privacy, and abuse prevention

Fast private kernel density estimation via locality sensitive quantization

Tal Wagner, Yonatan Naamad, Nina Mishra

ICML 2023

2023

We study efficient mechanisms for differentially private kernel density estimation (DP-KDE). Prior work for the Gaussian kernel described algorithms that run in time exponential in the number of dimensions d. This paper breaks the exponential barrier, and shows how the KDE can privately be approximated in time linear in d, making it feasible for high-dimensional data. We also present improved bounds for

Machine learning

Differentially private optimization on large model at small cost

Zhiqi Bu, Yu-Xiang Wang, Sheng Zha, George Karypis

ICML 2023

2023

Differentially private (DP) optimization is the standard paradigm to learn large neural networks that are accurate and privacy-preserving. The computational cost for DP deep learning, however, is notoriously heavy due to the per-sample gradient clipping. Existing DP implementations are 2 ∼ 1000× more costly in time and space complexity than the standard (non-private) training. In this work, we develop a

Machine learning

Better differential privacy for end-to-end speech recognition

Huck Yang, Andreas Stolcke

January 11, 2023

Private aggregation of teacher ensembles (PATE) leads to word error rate reductions of more than 26% relative to standard differential-privacy techniques.

Conversational AI

Differential privacy

Work with us