Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unlink to Unlearn: Simplifying Edge Unlearning in GNNs

Jiajun Tan, Fei Sun, Ruichen Qiu, Du Su, Huawei Shen

TL;DR

This work tackles privacy-preserving edge unlearning in Graph Neural Networks by diagnosing over-forgetting in the GNNDelete framework and proposing a minimalist alternative, UtU. UtU eliminates the two loss terms used by GNNDelete and unlearns simply by unlinking forgotten edges, effectively blocking their message-passing influence with constant-time cost. Empirical results on four real-world datasets across multiple GNN backbones show UtU achieves performance and privacy protections close to a model retrained from scratch, while markedly reducing computational overhead and avoiding the losses that induce over-forgetting. The findings highlight that removing a small subset of edges can have limited impact on parameters and that simple structural interventions can deliver practical, scalable edge unlearning for GNNs.

Abstract

As concerns over data privacy intensify, unlearning in Graph Neural Networks (GNNs) has emerged as a prominent research frontier in academia. This concept is pivotal in enforcing the \textit{right to be forgotten}, which entails the selective removal of specific data from trained GNNs upon user request. Our research focuses on edge unlearning, a process of particular relevance to real-world applications. Current state-of-the-art approaches like GNNDelete can eliminate the influence of specific edges yet suffer from \textit{over-forgetting}, which means the unlearning process inadvertently removes excessive information beyond needed, leading to a significant performance decline for remaining edges. Our analysis identifies the loss functions of GNNDelete as the primary source of over-forgetting and also suggests that loss functions may be redundant for effective edge unlearning. Building on these insights, we simplify GNNDelete to develop \textbf{Unlink to Unlearn} (UtU), a novel method that facilitates unlearning exclusively through unlinking the forget edges from graph structure. Our extensive experiments demonstrate that UtU delivers privacy protection on par with that of a retrained model while preserving high accuracy in downstream tasks, by upholding over 97.3\% of the retrained model's privacy protection capabilities and 99.8\% of its link prediction accuracy. Meanwhile, UtU requires only constant computational demands, underscoring its advantage as a highly lightweight and practical edge unlearning solution.

Unlink to Unlearn: Simplifying Edge Unlearning in GNNs

TL;DR

This work tackles privacy-preserving edge unlearning in Graph Neural Networks by diagnosing over-forgetting in the GNNDelete framework and proposing a minimalist alternative, UtU. UtU eliminates the two loss terms used by GNNDelete and unlearns simply by unlinking forgotten edges, effectively blocking their message-passing influence with constant-time cost. Empirical results on four real-world datasets across multiple GNN backbones show UtU achieves performance and privacy protections close to a model retrained from scratch, while markedly reducing computational overhead and avoiding the losses that induce over-forgetting. The findings highlight that removing a small subset of edges can have limited impact on parameters and that simple structural interventions can deliver practical, scalable edge unlearning for GNNs.

Abstract

As concerns over data privacy intensify, unlearning in Graph Neural Networks (GNNs) has emerged as a prominent research frontier in academia. This concept is pivotal in enforcing the \textit{right to be forgotten}, which entails the selective removal of specific data from trained GNNs upon user request. Our research focuses on edge unlearning, a process of particular relevance to real-world applications. Current state-of-the-art approaches like GNNDelete can eliminate the influence of specific edges yet suffer from \textit{over-forgetting}, which means the unlearning process inadvertently removes excessive information beyond needed, leading to a significant performance decline for remaining edges. Our analysis identifies the loss functions of GNNDelete as the primary source of over-forgetting and also suggests that loss functions may be redundant for effective edge unlearning. Building on these insights, we simplify GNNDelete to develop \textbf{Unlink to Unlearn} (UtU), a novel method that facilitates unlearning exclusively through unlinking the forget edges from graph structure. Our extensive experiments demonstrate that UtU delivers privacy protection on par with that of a retrained model while preserving high accuracy in downstream tasks, by upholding over 97.3\% of the retrained model's privacy protection capabilities and 99.8\% of its link prediction accuracy. Meanwhile, UtU requires only constant computational demands, underscoring its advantage as a highly lightweight and practical edge unlearning solution.
Paper Structure (16 sections, 5 equations, 3 figures, 3 tables)

This paper contains 16 sections, 5 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Operation of GNNs
  4. Edge Unlearning on GNNs
  5. GNNDelete
  6. UtU: Unlink to Unlearn
  7. Over-forgetting in Edge Unlearning
  8. Analysis on Unlearning Target
  9. UtU: a Minimalist Approach
  10. Experiments
  11. Experimental Setup
  12. Experiment Result and Anlysis
  13. Downstream Task
  14. Unlearning Efficacy
  15. Over-forgetting evaluation
  16. ...and 1 more sections

Figures (3)

  • Figure 1: An intuitive demonstration of over-forgetting, comparing the difference of retaining edge predictions before and after forgetting. The x-axis represents the change in predicted probabilities after unlearning, and the y-axis shows the distribution of these changes across the retain set $E_r$.
  • Figure 2: Activation distance ($\downarrow$) on forget set (5.0% edges).
  • Figure 3: Comparison of over-forgetting on GAT backbone.