Differential privacy of cross-attention with provable guarantee

Y Liang, Z Shi, Z Song, Y Zhou - arXiv preprint arXiv:2407.14717, 2024 - arxiv.org
Cross-attention has become a fundamental module nowadays in many important artificial
intelligence applications, eg, retrieval-augmented generation (RAG), system prompt, guided …

A unifying framework for differentially private sums under continual observation

M Henzinger, J Upadhyay, S Upadhyay - … of the 2024 Annual ACM-SIAM …, 2024 - SIAM
We study the problem of maintaining a differentially private decaying sum under continual
observation. We give a unifying framework and an efficient algorithm for this problem for any …

Differential privacy mechanisms in neural tangent kernel regression

J Gu, Y Liang, Z Sha, Z Shi, Z Song - arXiv preprint arXiv:2407.13621, 2024 - arxiv.org
Training data privacy is a fundamental problem in modern Artificial Intelligence (AI)
applications, such as face recognition, recommendation systems, language generation, and …

Differentially Private Hierarchical Heavy Hitters

A Biswas, G Cormode, Y Kanza, D Srivastava… - Proceedings of the ACM …, 2024 - dl.acm.org
The task of finding Hierarchical Heavy Hitters (HHH) was introduced by Cormode et al.[12]
as a generalisation of the heavy hitter problem. While finding HHH in data streams has been …

Concurrent shuffle differential privacy under continual observation

J Tenenbaum, H Kaplan, Y Mansour… - International …, 2023 - proceedings.mlr.press
We introduce the concurrent shuffle model of differential privacy. In this model we have
multiple concurrent shufflers permuting messages from different, possibly overlapping …

Differentially Private Substring and Document Counting

G Bernardini, P Bille, IL Gørtz, TA Steiner - arXiv preprint arXiv:2412.13813, 2024 - arxiv.org
Differential privacy is the gold standard for privacy in data analysis. In many data analysis
applications, the data is a database of documents. For databases consisting of many …