FedIA: A Plug-and-Play Importance-Aware Gradient Pruning Aggregation Method for Domain-Robust Federated Graph Learning on Node Classification

2025-09-17

AI Summary

Federated Graph Learning (FGL) suffers from unstable aggregation and inefficient convergence under domain shifts (e.g., Twitch Gamers, multilingual Wikipedia) due to high gradient noise across clients. To address this, we propose a lightweight “project-then-aggregate” denoising framework: the server performs top-ρ coordinate pruning based on gradient importance and incorporates influence-regularized momentum to suppress anomalous updates—all without increasing uplink communication overhead or imposing measurable additional server memory cost. Crucially, we pioneer the integration of projection as a *pre-aggregation* gradient preprocessing step, enabling a plug-and-play, two-stage denoising pipeline. We theoretically establish that our method preserves the optimal convergence rate of 𝒪(σ²/√T). Extensive experiments on heterogeneous graph benchmarks demonstrate consistent superiority over nine strong baselines, achieving smoother convergence and higher node classification accuracy.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Federated Graph Learning (FGL) under domain skew -- as observed on platforms such as emph{Twitch Gamers} and multilingual emph{Wikipedia} networks -- drives client models toward incompatible representations, rendering naive aggregation both unstable and ineffective. We find that the culprit is not the weighting scheme but the emph{noisy gradient signal}: empirical analysis of baseline methods suggests that a vast majority of gradient dimensions can be dominated by domain-specific variance. We therefore shift focus from "aggregation-first" to a emph{projection-first} strategy that denoises client updates emph{before} they are combined. The proposed FedIA framework realises this underline{I}mportance-underline{A}ware idea through a two-stage, plug-and-play pipeline: (i) a server-side top-$ρ$ mask keeps only the most informative about 5% of coordinates, and (ii) a lightweight influence-regularised momentum weight suppresses outlier clients. FedIA adds emph{no extra uplink traffic and only negligible server memory}, making it readily deployable. On both homogeneous (Twitch Gamers) and heterogeneous (Wikipedia) graphs, it yields smoother, more stable convergence and higher final accuracy than nine strong baselines. A convergence sketch further shows that dynamic projection maintains the optimal $mathcal{O}(σ^{2}/sqrt{T})$ rate.

Problem

• Addresses unstable aggregation in federated graph learning with domain skew
• Solves noisy gradient signals dominating domain-specific variance in client updates
• Improves convergence stability and accuracy for node classification tasks

Innovation

• Plug-and-play importance-aware gradient pruning aggregation
• Server-side top-ρ mask keeps most informative coordinates
• Lightweight influence-regularised momentum weight suppresses outliers