New REMARK framework enhances GNN ownership verification

Researchers have developed REMARK, a novel framework for verifying the ownership of Graph Neural Networks (GNNs). This method addresses limitations in existing watermark and fingerprint-based techniques by generating in-distribution watermark graphs that minimize performance degradation. REMARK extracts robust fingerprints from output differences, removing the need for surrogate models to be trained on watermark-containing datasets or to expose specific output levels. Experiments show REMARK achieves state-of-the-art accuracy and robustness in GNN ownership verification while preserving model utility. AI

IMPACT This framework could help protect intellectual property and prevent unauthorized use of costly GNN models.

RANK_REASON The cluster contains a research paper detailing a new framework for GNN ownership verification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• graph neural networks
• REMARK

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →