GraphP-FL: Personalized Federated Graph Learning via Dynamic Structure Awareness and Fisher Information Elastic Alignment

Haoyu Chen, Zening Zhao, Jinsong Wang, Kai Shi, Zongpu Wei, Jianhao Li
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:18517-18536, 2026.

Abstract

Federated Graph Learning (FGL) enables distributed clients to collaboratively train graph neural networks while strictly preserving data privacy. However, existing FGL methods implicitly assume the reliability of local graph structures and lack elastic awareness of parameter importance during model aggregation, leading to representation degradation under topological noise and catastrophic forgetting caused by model drift. To address these challenges, we propose GraphP-FL, a general personalized FGL framework. (1) Specifically, we design a self-supervised dynamic topology reconstruction mechanism on the client side. This mechanism mines implicit dependencies to adaptively rectify noisy topologies, effectively suppressing topological noise propagation and capturing precise structural relationships for high-quality representations. (2) Additionally, we introduce a Fisher-based Elastic Parameter Alignment (FRPA) algorithm. FRPA imposes anisotropic regularization constraints in the parameter space to precisely quantify parameter importance, enabling the model to strictly preserve critical local knowledge while flexibly aligning with the global model, thus effectively overcoming catastrophic forgetting. Extensive experiments on seven benchmarks (including biochemical molecules, social networks, and large-scale encrypted traffic) demonstrate that GraphP-FL significantly outperforms state-of-the-art methods, improving accuracy by up to 8.6% while exhibiting superior generalization and robustness.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26gu, title = {{G}raph{P}-{FL}: Personalized Federated Graph Learning via Dynamic Structure Awareness and {F}isher Information Elastic Alignment}, author = {Chen, Haoyu and Zhao, Zening and Wang, Jinsong and Shi, Kai and Wei, Zongpu and Li, Jianhao}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {18517--18536}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chen26gu/chen26gu.pdf}, url = {https://proceedings.mlr.press/v306/chen26gu.html}, abstract = {Federated Graph Learning (FGL) enables distributed clients to collaboratively train graph neural networks while strictly preserving data privacy. However, existing FGL methods implicitly assume the reliability of local graph structures and lack elastic awareness of parameter importance during model aggregation, leading to representation degradation under topological noise and catastrophic forgetting caused by model drift. To address these challenges, we propose GraphP-FL, a general personalized FGL framework. (1) Specifically, we design a self-supervised dynamic topology reconstruction mechanism on the client side. This mechanism mines implicit dependencies to adaptively rectify noisy topologies, effectively suppressing topological noise propagation and capturing precise structural relationships for high-quality representations. (2) Additionally, we introduce a Fisher-based Elastic Parameter Alignment (FRPA) algorithm. FRPA imposes anisotropic regularization constraints in the parameter space to precisely quantify parameter importance, enabling the model to strictly preserve critical local knowledge while flexibly aligning with the global model, thus effectively overcoming catastrophic forgetting. Extensive experiments on seven benchmarks (including biochemical molecules, social networks, and large-scale encrypted traffic) demonstrate that GraphP-FL significantly outperforms state-of-the-art methods, improving accuracy by up to 8.6% while exhibiting superior generalization and robustness.} }
Endnote
%0 Conference Paper %T GraphP-FL: Personalized Federated Graph Learning via Dynamic Structure Awareness and Fisher Information Elastic Alignment %A Haoyu Chen %A Zening Zhao %A Jinsong Wang %A Kai Shi %A Zongpu Wei %A Jianhao Li %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chen26gu %I PMLR %P 18517--18536 %U https://proceedings.mlr.press/v306/chen26gu.html %V 306 %X Federated Graph Learning (FGL) enables distributed clients to collaboratively train graph neural networks while strictly preserving data privacy. However, existing FGL methods implicitly assume the reliability of local graph structures and lack elastic awareness of parameter importance during model aggregation, leading to representation degradation under topological noise and catastrophic forgetting caused by model drift. To address these challenges, we propose GraphP-FL, a general personalized FGL framework. (1) Specifically, we design a self-supervised dynamic topology reconstruction mechanism on the client side. This mechanism mines implicit dependencies to adaptively rectify noisy topologies, effectively suppressing topological noise propagation and capturing precise structural relationships for high-quality representations. (2) Additionally, we introduce a Fisher-based Elastic Parameter Alignment (FRPA) algorithm. FRPA imposes anisotropic regularization constraints in the parameter space to precisely quantify parameter importance, enabling the model to strictly preserve critical local knowledge while flexibly aligning with the global model, thus effectively overcoming catastrophic forgetting. Extensive experiments on seven benchmarks (including biochemical molecules, social networks, and large-scale encrypted traffic) demonstrate that GraphP-FL significantly outperforms state-of-the-art methods, improving accuracy by up to 8.6% while exhibiting superior generalization and robustness.
APA
Chen, H., Zhao, Z., Wang, J., Shi, K., Wei, Z. & Li, J.. (2026). GraphP-FL: Personalized Federated Graph Learning via Dynamic Structure Awareness and Fisher Information Elastic Alignment. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:18517-18536 Available from https://proceedings.mlr.press/v306/chen26gu.html.

Related Material