Correlation inference attacks against machine learning models

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Creţu, Ana-Maria, Guépin, Florent, de Montjoye, Yves-Alexandre
Formato: Preprint
Publicado: 2021
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866929425431396352
author Creţu, Ana-Maria
Guépin, Florent
de Montjoye, Yves-Alexandre
author_facet Creţu, Ana-Maria
Guépin, Florent
de Montjoye, Yves-Alexandre
contents Despite machine learning models being widely used today, the relationship between a model and its training dataset is not well understood. We explore correlation inference attacks, whether and when a model leaks information about the correlations between the input variables of its training dataset. We first propose a model-less attack, where an adversary exploits the spherical parametrization of correlation matrices alone to make an informed guess. Second, we propose a model-based attack, where an adversary exploits black-box model access to infer the correlations using minimal and realistic assumptions. Third, we evaluate our attacks against logistic regression and multilayer perceptron models on three tabular datasets and show the models to leak correlations. We finally show how extracted correlations can be used as building blocks for attribute inference attacks and enable weaker adversaries. Our results raise fundamental questions on what a model does and should remember from its training set.
format Preprint
id arxiv_https___arxiv_org_abs_2112_08806
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Correlation inference attacks against machine learning models
Creţu, Ana-Maria
Guépin, Florent
de Montjoye, Yves-Alexandre
Machine Learning
Cryptography and Security
Despite machine learning models being widely used today, the relationship between a model and its training dataset is not well understood. We explore correlation inference attacks, whether and when a model leaks information about the correlations between the input variables of its training dataset. We first propose a model-less attack, where an adversary exploits the spherical parametrization of correlation matrices alone to make an informed guess. Second, we propose a model-based attack, where an adversary exploits black-box model access to infer the correlations using minimal and realistic assumptions. Third, we evaluate our attacks against logistic regression and multilayer perceptron models on three tabular datasets and show the models to leak correlations. We finally show how extracted correlations can be used as building blocks for attribute inference attacks and enable weaker adversaries. Our results raise fundamental questions on what a model does and should remember from its training set.
title Correlation inference attacks against machine learning models
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2112.08806