Black-Box Privacy Attacks on Shared Representations in Multitask Learning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Abascal, John, Berrios, Nicolás, Oprea, Alina, Ullman, Jonathan, Smith, Adam, Jagielski, Matthew
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913903750938624
author Abascal, John
Berrios, Nicolás
Oprea, Alina
Ullman, Jonathan
Smith, Adam
Jagielski, Matthew
author_facet Abascal, John
Berrios, Nicolás
Oprea, Alina
Ullman, Jonathan
Smith, Adam
Jagielski, Matthew
contents Multitask learning (MTL) has emerged as a powerful paradigm that leverages similarities among multiple learning tasks, each with insufficient samples to train a standalone model, to solve them simultaneously while minimizing data sharing across users and organizations. MTL typically accomplishes this goal by learning a shared representation that captures common structure among the tasks by embedding data from all tasks into a common feature space. Despite being designed to be the smallest unit of shared information necessary to effectively learn patterns across multiple tasks, these shared representations can inadvertently leak sensitive information about the particular tasks they were trained on. In this work, we investigate what information is revealed by the shared representations through the lens of inference attacks. Towards this, we propose a novel, black-box task-inference threat model where the adversary, given the embedding vectors produced by querying the shared representation on samples from a particular task, aims to determine whether that task was present when training the shared representation. We develop efficient, purely black-box attacks on machine learning models that exploit the dependencies between embeddings from the same task without requiring shadow models or labeled reference data. We evaluate our attacks across vision and language domains for multiple use cases of MTL and demonstrate that even with access only to fresh task samples rather than training data, a black-box adversary can successfully infer a task's inclusion in training. To complement our experiments, we provide theoretical analysis of a simplified learning setting and show a strict separation between adversaries with training samples and fresh samples from the target task's distribution.
format Preprint
id arxiv_https___arxiv_org_abs_2506_16460
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Black-Box Privacy Attacks on Shared Representations in Multitask Learning
Abascal, John
Berrios, Nicolás
Oprea, Alina
Ullman, Jonathan
Smith, Adam
Jagielski, Matthew
Machine Learning
Cryptography and Security
Multitask learning (MTL) has emerged as a powerful paradigm that leverages similarities among multiple learning tasks, each with insufficient samples to train a standalone model, to solve them simultaneously while minimizing data sharing across users and organizations. MTL typically accomplishes this goal by learning a shared representation that captures common structure among the tasks by embedding data from all tasks into a common feature space. Despite being designed to be the smallest unit of shared information necessary to effectively learn patterns across multiple tasks, these shared representations can inadvertently leak sensitive information about the particular tasks they were trained on. In this work, we investigate what information is revealed by the shared representations through the lens of inference attacks. Towards this, we propose a novel, black-box task-inference threat model where the adversary, given the embedding vectors produced by querying the shared representation on samples from a particular task, aims to determine whether that task was present when training the shared representation. We develop efficient, purely black-box attacks on machine learning models that exploit the dependencies between embeddings from the same task without requiring shadow models or labeled reference data. We evaluate our attacks across vision and language domains for multiple use cases of MTL and demonstrate that even with access only to fresh task samples rather than training data, a black-box adversary can successfully infer a task's inclusion in training. To complement our experiments, we provide theoretical analysis of a simplified learning setting and show a strict separation between adversaries with training samples and fresh samples from the target task's distribution.
title Black-Box Privacy Attacks on Shared Representations in Multitask Learning
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2506.16460