Ulixes: Facial Recognition Privacy with Adversarial Machine Learning

Facial recognition tools are becoming exceptionally accurate in identifying people from images. However, this comes at the cost of privacy for users of online services with photo management (e.g. social media platforms). Particularly troubling is the ability to leverage unsupervised learning to reco...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Cilloni Thomas, Wang Wei, Walter Charles, Fleming Charles
Formato:	article
Lenguaje:	EN
Publicado:	Sciendo 2022
Materias:	adversarial machine learning facial recognition privacy Ethics BJ1-1725 Electronic computers. Computer science QA75.5-76.95
Acceso en línea:	https://doaj.org/article/547ff9f600b24ae38477bf13f24ee95c
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:547ff9f600b24ae38477bf13f24ee95c
record_format	dspace
spelling	oai:doaj.org-article:547ff9f600b24ae38477bf13f24ee95c2021-12-05T14:11:10ZUlixes: Facial Recognition Privacy with Adversarial Machine Learning2299-098410.2478/popets-2022-0008https://doaj.org/article/547ff9f600b24ae38477bf13f24ee95c2022-01-01T00:00:00Zhttps://doi.org/10.2478/popets-2022-0008https://doaj.org/toc/2299-0984Facial recognition tools are becoming exceptionally accurate in identifying people from images. However, this comes at the cost of privacy for users of online services with photo management (e.g. social media platforms). Particularly troubling is the ability to leverage unsupervised learning to recognize faces even when the user has not labeled their images. In this paper we propose Ulixes, a strategy to generate visually non-invasive facial noise masks that yield adversarial examples, preventing the formation of identifiable user clusters in the embedding space of facial encoders. This is applicable even when a user is unmasked and labeled images are available online. We demonstrate the effectiveness of Ulixes by showing that various classification and clustering methods cannot reliably label the adversarial examples we generate. We also study the effects of Ulixes in various black-box settings and compare it to the current state of the art in adversarial machine learning. Finally, we challenge the effectiveness of Ulixes against adversarially trained models and show that it is robust to countermeasures.Cilloni ThomasWang WeiWalter CharlesFleming CharlesSciendoarticleadversarial machine learningfacial recognitionprivacyEthicsBJ1-1725Electronic computers. Computer scienceQA75.5-76.95ENProceedings on Privacy Enhancing Technologies, Vol 2022, Iss 1, Pp 148-165 (2022)
institution	DOAJ
collection	DOAJ
language	EN
topic	adversarial machine learning facial recognition privacy Ethics BJ1-1725 Electronic computers. Computer science QA75.5-76.95
spellingShingle	adversarial machine learning facial recognition privacy Ethics BJ1-1725 Electronic computers. Computer science QA75.5-76.95 Cilloni Thomas Wang Wei Walter Charles Fleming Charles Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
description	Facial recognition tools are becoming exceptionally accurate in identifying people from images. However, this comes at the cost of privacy for users of online services with photo management (e.g. social media platforms). Particularly troubling is the ability to leverage unsupervised learning to recognize faces even when the user has not labeled their images. In this paper we propose Ulixes, a strategy to generate visually non-invasive facial noise masks that yield adversarial examples, preventing the formation of identifiable user clusters in the embedding space of facial encoders. This is applicable even when a user is unmasked and labeled images are available online. We demonstrate the effectiveness of Ulixes by showing that various classification and clustering methods cannot reliably label the adversarial examples we generate. We also study the effects of Ulixes in various black-box settings and compare it to the current state of the art in adversarial machine learning. Finally, we challenge the effectiveness of Ulixes against adversarially trained models and show that it is robust to countermeasures.
format	article
author	Cilloni Thomas Wang Wei Walter Charles Fleming Charles
author_facet	Cilloni Thomas Wang Wei Walter Charles Fleming Charles
author_sort	Cilloni Thomas
title	Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
title_short	Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
title_full	Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
title_fullStr	Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
title_full_unstemmed	Ulixes: Facial Recognition Privacy with Adversarial Machine Learning
title_sort	ulixes: facial recognition privacy with adversarial machine learning
publisher	Sciendo
publishDate	2022
url	https://doaj.org/article/547ff9f600b24ae38477bf13f24ee95c
work_keys_str_mv	AT cillonithomas ulixesfacialrecognitionprivacywithadversarialmachinelearning AT wangwei ulixesfacialrecognitionprivacywithadversarialmachinelearning AT waltercharles ulixesfacialrecognitionprivacywithadversarialmachinelearning AT flemingcharles ulixesfacialrecognitionprivacywithadversarialmachinelearning
_version_	1718371299247521792

Ulixes: Facial Recognition Privacy with Adversarial Machine Learning

Ejemplares similares