PuSH - Publication Server of Helmholtz Zentrum München

Export: TXT Text RIS Endnote (RIS) BIB BibTeX
Bani-Harouni, D.* ; Mueller, T.T.* ; Rueckert, D.* ; Kaissis, G.

Gradient Self-alignment in Private Deep Learning.

In: (26th International Conference on Medical Image Computing and Computer-Assisted Intervention , MICCAI 2023, 8 - 12 October 2023, Vancouver, CANADA). Berlin [u.a.]: Springer, 2023. 89-97 (Lect. Notes Comput. Sc. ; 14393)
DOI
Open Access Green as soon as Postprint is submitted to ZB.
Differential Privacy (DP) has become a gold-standard to preserve privacy in deep learning. Intuitively speaking, DP ensures that the output of a model is approximately invariant to the inclusion or exclusion of a single individual’s data from the training set. There is, however, a trade-off between privacy and utility. DP models tend to perform worse than non-DP models trained on the same data. This is caused by the clipping of per-sample gradients and the addition of noise required for DP guarantees causing an obfuscation of the individual data point’s contribution. In this work, we propose a method to reduce this discrepancy by improving the alignment between the per-sample gradients of each individual training sample with its non-DP gradient by increasing their cosine similarity. Optimizing the alignment in only a relevant subset of gradient dimensions, further improves the performance. We evaluate our method on CIFAR-10 and a pediatric pneumonia chest x-ray dataset.
Altmetric
Tags
Annotations
Special Publikation
Hide on homepage

Edit extra information
Edit own tags
Private
Edit own annotation
Private
Hide on publication lists
on hompage
Mark as special
publikation
Publication type Article: Conference contribution
Keywords Differential Privacy ; Gradient Alignment ; Private Learning
Language english
Publication Year 2023
HGF-reported in Year 2023
ISSN (print) / ISBN 0302-9743
e-ISSN 1611-3349
Conference Title 26th International Conference on Medical Image Computing and Computer-Assisted Intervention , MICCAI 2023
Conference Date 8 - 12 October 2023
Conference Location Vancouver, CANADA
Journal Lecture Notes in Computer Science
Quellenangaben Volume: 14393, Issue: , Pages: 89-97 Article Number: , Supplement: ,
Publisher Springer
Publishing Place Berlin [u.a.]
Institute(s) Institute for Machine Learning in Biomed Imaging (IML)
POF-Topic(s) 30205 - Bioengineering and Digital Health
Research field(s) Enabling and Novel Technologies
PSP Element(s) G-507100-001
DOI 10.1007/978-3-031-47401-9_9
WOS ID 001160722800009
Scopus ID 85180624612
Erfassungsdatum 2024-01-19
[➜Report correction]