Differentially private optimization on large model at small cost

Z Bu, YX Wang, S Zha… - … Conference on Machine …, 2023 - proceedings.mlr.press
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 …

Llm-pbe: Assessing data privacy in large language models

Q Li, J Hong, C Xie, J Tan, R Xin, J Hou, X Yin… - arXiv preprint arXiv …, 2024 - arxiv.org
Large Language Models (LLMs) have become integral to numerous domains, significantly
advancing applications in data management, mining, and analysis. Their profound …

Explaining the model, protecting your data: Revealing and mitigating the data privacy risks of post-hoc model explanations via membership inference

C Huang, M Pawelczyk, H Lakkaraju - arXiv preprint arXiv:2407.17663, 2024 - arxiv.org
Predictive machine learning models are becoming increasingly deployed in high-stakes
contexts involving sensitive personal data; in these contexts, there is a trade-off between …

ExpShield: Safeguarding Web Text from Unauthorized Crawling and Language Modeling Exploitation

R Liu, T Tran, T Wang, H Hu, S Wang… - arXiv preprint arXiv …, 2024 - arxiv.org
As large language models (LLMs) increasingly depend on web-scraped datasets, concerns
over unauthorized use of copyrighted or personal content for training have intensified …