GrOVe: Ownership Verification of Graph Neural Networks using Embeddings
Asim Waheed, Vasisht Duddu, N. Asokan
IEEE Symposium on Security and Privacy 2024 · Day 2 · Continental Ballroom 5
In an era where Graph Neural Networks (GNNs) are becoming indispensable for modeling complex real-world relationships in social networks, recommendation systems, and scientific applications, the intellectual property of these sophisticated models faces significant threats. This talk introduces **GrOVe** (Ownership Verification of Graph Neural Networks using Embeddings), a novel framework designed to demonstrate and verify the ownership of GNNs in the face of increasingly sophisticated model extraction attacks. Presented by Asim Waheed, alongside co-authors Vasisht Duddu and N. Asokan, the research addresses a critical vulnerability in the deployment of proprietary GNN models.
AI review
This research delivers a robust, practical solution to a critical problem: GNN intellectual property theft via model extraction. Leveraging embedding similarity as a fingerprint, GrOVe achieves a critical 0% false negative rate against known extraction attacks, making it a highly effective defensive mechanism. This isn't just theory; it's a well-engineered countermeasure with verifiable results.