Can I read Disentangled GCN Defense Against MIA on EtoBox?

Disentangled GCN Defense Against MIA by zzjjyy429 is a document available to read on EtoBox.

What is Disentangled GCN Defense Against MIA about?

The document presents DMIA, a novel disentangled-based method designed to enhance the security of Graph Convolutional Networks (GCNs) against Membership Inference Attacks (MIA) while maintaining model utility. By separating sensitive and practical information in latent representations, the method minimizes linear dependency and employs a mixed updating strategy for improved training. Experimental results demonstrate that DMIA effectively reduces MIA risks without compromising node classification performance

Author

zzjjyy429

Language

EN