Researchers have developed GraphToxin, a novel attack capable of reconstructing entire unlearned graphs from graph neural networks (GNNs). This method goes beyond previous membership inference attacks by not only recovering deleted individual data but also sensitive information of their neighbors. GraphToxin utilizes a curvature matching module for fine-grained guidance and has demonstrated effectiveness in both white-box and black-box scenarios, highlighting the limitations of current graph unlearning verification standards. AI
IMPACT This research highlights significant vulnerabilities in current graph unlearning techniques, potentially impacting data privacy and security in AI systems that handle graph data.
RANK_REASON The cluster contains a research paper detailing a new attack method against graph unlearning. [lever_c_demoted from research: ic=1 ai=1.0]
- graph neural networks
- GraphToxin
- Graph Unlearning with Efficient Partial Retraining
- Membership Inference Attacks
- right to be forgotten
- Ying Song
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
