SoK: Data Reconstruction Attacks Against Machine Learning Models: Definition, Metrics, and Benchmark

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Wen, Rui, Liu, Yiyong, Backes, Michael, Zhang, Yang
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915333849219072
author Wen, Rui
Liu, Yiyong
Backes, Michael
Zhang, Yang
author_facet Wen, Rui
Liu, Yiyong
Backes, Michael
Zhang, Yang
contents Data reconstruction attacks, which aim to recover the training dataset of a target model with limited access, have gained increasing attention in recent years. However, there is currently no consensus on a formal definition of data reconstruction attacks or appropriate evaluation metrics for measuring their quality. This lack of rigorous definitions and universal metrics has hindered further advancement in this field. In this paper, we address this issue in the vision domain by proposing a unified attack taxonomy and formal definitions of data reconstruction attacks. We first propose a set of quantitative evaluation metrics that consider important criteria such as quantifiability, consistency, precision, and diversity. Additionally, we leverage large language models (LLMs) as a substitute for human judgment, enabling visual evaluation with an emphasis on high-quality reconstructions. Using our proposed taxonomy and metrics, we present a unified framework for systematically evaluating the strengths and limitations of existing attacks and establishing a benchmark for future research. Empirical results, primarily from a memorization perspective, not only validate the effectiveness of our metrics but also offer valuable insights for designing new attacks.
format Preprint
id arxiv_https___arxiv_org_abs_2506_07888
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SoK: Data Reconstruction Attacks Against Machine Learning Models: Definition, Metrics, and Benchmark
Wen, Rui
Liu, Yiyong
Backes, Michael
Zhang, Yang
Cryptography and Security
Machine Learning
Data reconstruction attacks, which aim to recover the training dataset of a target model with limited access, have gained increasing attention in recent years. However, there is currently no consensus on a formal definition of data reconstruction attacks or appropriate evaluation metrics for measuring their quality. This lack of rigorous definitions and universal metrics has hindered further advancement in this field. In this paper, we address this issue in the vision domain by proposing a unified attack taxonomy and formal definitions of data reconstruction attacks. We first propose a set of quantitative evaluation metrics that consider important criteria such as quantifiability, consistency, precision, and diversity. Additionally, we leverage large language models (LLMs) as a substitute for human judgment, enabling visual evaluation with an emphasis on high-quality reconstructions. Using our proposed taxonomy and metrics, we present a unified framework for systematically evaluating the strengths and limitations of existing attacks and establishing a benchmark for future research. Empirical results, primarily from a memorization perspective, not only validate the effectiveness of our metrics but also offer valuable insights for designing new attacks.
title SoK: Data Reconstruction Attacks Against Machine Learning Models: Definition, Metrics, and Benchmark
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2506.07888