Open access
Jul 2026
Graph Attributed Unlearning via Propagation Suppression and Knowledge Dissipation
A graph unlearning framework specifically designed for feature-level unlearning, consisting of two main stages, which zero out the features of the unlearned nodes at each layer to block their propagation through the GNN, thereby reducing their influence on neighboring node representations.
Zhiyu Chen, Jiaquan Liang, Qi Luo et al.
· Mathematics · 0 citations