Recent advances in differentially private deep learning have demonstrated that application of differential privacy, specifically the DP-SGD algorithm, has a disparate impact on different sub-groups in the population, which leads to a significantly high drop-in model utility for sub-populations that are under-represented (minorities), compared to well-represented ones. In this work, we aim to compare PATE, another mechanism for training deep learning models using differential privacy, with DP-SGD in terms of fairness. We show that PATE does have a disparate impact too, however, it is much less severe than DP-SGD. We draw insights from this observation on what might be promising directions in achieving better fairness-privacy trade-offs.
@article{arxiv.2106.12576,
title = {DP-SGD vs PATE: Which Has Less Disparate Impact on Model Accuracy?},
author = {Archit Uniyal and Rakshit Naidu and Sasikanth Kotti and Sahib Singh and Patrik Joslin Kenfack and Fatemehsadat Mireshghallah and Andrew Trask},
journal= {arXiv preprint arXiv:2106.12576},
year = {2022}
}