Get Started
Research questionHow can GNN ownership be verified without out-of-distribution watermarks or watermark-trained surrogates?Ownership tests for GNNs can reduce protected-model utility when watermark graphs differ from the training data. They may also rely on surrogate models trained with the watermark or on particular output levels.
AI
Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.A Robust Watermark-based Fingerprint Framework for GNNs Ownership VerificationThe source describes REMARK, which generates in-distribution watermark graphs and extracts fingerprints from output differences across real-world datasets and GNN architectures. Its evidence is experimental and does not specify deployment or query-access requirements.research paper · Sep 4, 2026
Related questions
How can synthetic images be traced to a target generator without modifying it or embedding watermarks?How can watermarks in generated speech remain detectable after neural audio transformations?How can zero-shot graph anomaly detectors remain reliable on unlabeled targets under domain shift?How can watermarking provide trustworthy provenance for LLM text at scale despite transformations and accumulating false positives?
Home
Topics
Search
Library