On Stealing Graph Neural Network Models

TL;DR

This paper presents a novel method for extracting graph neural network models with minimal queries, demonstrating effectiveness even under strict query limits and existing defenses, highlighting the need for stronger protections.

Contribution

The paper introduces a two-step attack that first recovers the GNN backbone without queries and then maximizes information gain within limited queries.

Findings

Effective extraction under strict query limits

Successful attack against existing defenses

Applicable to multiple real-world datasets

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.