MMFGU: Multimodal Federated Graph Unlearning

2026-07-30

AI Summary

This work addresses the challenge of fine-grained unlearning in multimodal federated graph learning, where existing methods struggle to precisely remove specific data instances—such as a single image or text snippet—without compromising retained content or leaving recoverable traces. To this end, we propose MMFGU, the first framework enabling fine-grained multimodal graph unlearning. MMFGU employs a target-specific representation disentanglement mechanism to uniformly model heterogeneous deletion requests as target carriers, allowing precise erasure of designated information while preserving anchor semantics. It further integrates a lightweight probing module with a prototype-response signaling mechanism to detect residual propagation pathways and selectively purify client models. Experiments demonstrate that MMFGU completely removes targeted data while maintaining model utility, achieving a 41.5× speedup over full retraining.

Abstract

Multimodal federated graph learning enables clients to collaboratively train graph models over structural, textual, and visual signals without sharing private local data. However, the presence of heterogeneous multimodal content also makes unlearning requests more frequent and fine-grained: users may delete accounts or interactions, remove a particular image or text while retaining the associated entity, or revoke the learned correspondence between retained modalities or graph attributes. Existing federated graph unlearning mainly handles entity/relation or client removal and cannot directly satisfy these multimodal requests. They introduce three challenges: removing only the requested information without damaging retained content, preventing the target from being recovered through remaining modalities or graph neighborhoods, and stopping related traces on other clients from re-entering the global model after aggregation. To address them, we propose \textsc{\textbf{MMFGU}}, a multimodal federated graph unlearning framework built around target-specific representation decoupling. \textsc{MMFGU} maps heterogeneous requests into unified target carriers, decouples requested representations while anchoring retained semantics, exposes and repairs propagated residuals with lightweight probes, and selectively purges affected clients through compact prototype and response signals. Experiments show that \textsc{MMFGU} effectively removes requested information, preserves retained graph utility, and achieves a $\boldsymbol{41.5\times}$ speedup over full retraining.

Problem

• multimodal
• federated graph learning
• unlearning
• fine-grained deletion
• privacy preservation

Innovation

• Multimodal Federated Learning
• Graph Unlearning
• Representation Decoupling
• Privacy Preservation
• Federated Graph Neural Networks