Work

On Stealing Graph Neural Network Models

Year: 2025

Type: preprint

Abstract: Current graph neural network (GNN) model-stealing methods rely heavily on queries to the victim model, assuming no hard query limits. However, in reality, the number of allowed queries can be severely limited. In this paper, we demonstrate how an adversary can extract a GNN with very limited interactions with the model. Our approach first enables the adversary to obtain the model backbone without making direct queries to the victim model and then to strategically utilize a fixed query limit to extract the most informative data. The experiments on eight real-world datasets demonstrate the effectiveness of the attack, even under a very restricted query limit and under defense against model extraction in place. Our findings underscore the need for robust defenses against GNN model extraction threats.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2511.07170

Authors: Podhajski, Marcin, Dubiński, Jan, Boenisch, Franziska, Dziedzic, Adam, Pręgowska, Agnieszka, Michalak, Tomasz P.

Institutions: Institute of Fundamental Technological Research, Polish Academy of Sciences, Helmholtz Center for Information Security, University of Warsaw

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funders: Deutsche Forschungsgemeinschaft, Helmholtz Artificial Intelligence Cooperation Unit

OpenAlex ID: W4416237206

Cites:

Cited by:

Related to: