English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Graph unlearning has emerged as a pivotal method to delete information from apre-trained graph neural network (GNN). One may delete nodes, a class of nodes,edges, or a class of edges. An unlearning method enables the GNN model tocomply with data protection regulations (i.e., the right to be forgotten),adapt to evolving data distributions, and reduce the GPU-hours carbon footprintby avoiding repetitive retraining. Existing partitioning and aggregation-basedmethods have limitations due to their poor handling of local graph dependenciesand additional overhead costs. More recently, GNNDelete offered amodel-agnostic approach that alleviates some of these issues. Our work takes anovel approach to address these challenges in graph unlearning throughknowledge distillation, as it distills to delete in GNN (D2DGN). It is amodel-agnostic distillation framework where the complete graph knowledge isdivided and marked for retention and deletion. It performs distillation withresponse-based soft targets and feature-based node embedding while minimizingKL divergence. The unlearned model effectively removes the influence of deletedgraph elements while preserving knowledge about the retained graph elements.D2DGN surpasses the performance of existing methods when evaluated on variousreal-world graph datasets by up to $43.1\%$ (AUC) in edge and node unlearningtasks. Other notable advantages include better efficiency, better performancein removing target elements, preservation of performance for the retainedelements, and zero overhead costs. Notably, our D2DGN surpasses thestate-of-the-art GNNDelete in AUC by $2.4\%$, improves membership inferenceratio by $+1.3$, requires $10.2\times10^6$ fewer FLOPs per forward pass and upto $\mathbf{3.2}\times$ faster.

Distill to Delete: Unlearning in Graph Networks with Knowledge Distillation