paper-with-me

Papers

Does Differential Privacy Impact Bias in Pretrained NLP Models?

2024-10-24 · Md. Khairul Islam, Andrew Wang, Tianhao Wang, Yangfeng Ji, Judy Fox, Jieyu Zhao

Differential privacy (DP) is applied when fine-tuning pre-trained large language models (LLMs) to limit leakage of training examples. While most DP research has focused on improving a model's privacy-utility tradeoff, some find that DP can be unfair to or biased against underrepresented groups. In this work, we show the impact of DP on bias in LLMs through empirical analysis. Differentially private training can increase the model bias against protected groups w.r.t AUC-based bias metrics. DP makes it more difficult for the model to differentiate between the positive and negative examples from the protected groups and other groups in the rest of the population. Our results also show that the impact of DP on bias is not only affected by the privacy protection level but also the underlying distribution of the dataset.

📄 PDF Abstract BibTeX arXiv:2410.18749

Code (1)

khairulislam/dp-on-nlp-bias 공식 구현 pytorch

Similar Papers 제목 키워드 기반

How Does Differential Privacy Affect Social Bias in LLMs? A Systematic Evaluation

2026-05-11 · Eduardo Tenorio, Karuna Bhaila, Xintao Wu arxiv

Large language models (LLMs) trained on web-scale corpora can memorize sensitive training data, posing significant privacy risks. Differential privacy (DP) has emerged as a principled framework that limits the influence …

Question Answering

Removing Disparate Impact of Differentially Private Stochastic Gradient Descent on Model Accuracy

2020-03-08 · Depeng Xu, Wei Du, Xintao Wu

When we enforce differential privacy in machine learning, the utility-privacy trade-off is different w.r.t. each group. Gradient clipping and random noise addition disproportionately affect underrepresented and complex c…

The Impact of Differential Privacy on Group Disparity Mitigation

2022-03-05 · NAACL (PrivateNLP) 2022 7 · Victor Petrén Bach Hansen, Atula Tejaswi Neerkaje, Ramit Sawhney, Lucie Flek 외

The performance cost of differential privacy has, for some applications, been shown to be higher for minority groups; fairness, conversely, has been shown to disproportionally compromise the privacy of members of such gr…

Fairness

The Impact of Differential Privacy on Recommendation Accuracy and Popularity Bias

2024-01-08 · Peter Müllner, Elisabeth Lex, Markus Schedl, Dominik Kowald

Collaborative filtering-based recommender systems leverage vast amounts of behavioral user data, which poses severe privacy risks. Thus, often, random noise is added to the data to ensure Differential Privacy (DP). Howev…

Collaborative FilteringRecommendation Systems

Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining

2022-12-13 · Florian Tramèr, Gautam Kamath, Nicholas Carlini

The performance of differentially private machine learning can be boosted significantly by leveraging the transfer learning capabilities of non-private models pretrained on large public datasets. We critically review thi…

PositionPrivacy PreservingTransfer Learning