Causality-inspired Federated Learning for Dynamic Spatio-Temporal Graphs

2026-03-31

AI Summary

Existing federated graph learning methods suffer from representation entanglement and negative transfer in dynamic spatio-temporal graphs due to their neglect of spatial-temporal heterogeneity and client-specific characteristics, which undermines model generalization. To address this, this work proposes the SC-FSGL framework, which leverages a causality-inspired representation intervention mechanism to disentangle transferable causal knowledge from client-specific noise. Specifically, a conditional separation module is introduced to simulate soft interventions for disentangling causal factors, and a causal codebook combined with contrastive learning is designed to align causal prototypes across clients, thereby enabling consistent and effective knowledge sharing. Extensive experiments on five heterogeneous spatio-temporal graph datasets demonstrate that the proposed method significantly outperforms state-of-the-art approaches, substantially enhancing both generalization performance and robustness.

Abstract

Federated Graph Learning (FGL) has emerged as a powerful paradigm for decentralized training of graph neural networks while preserving data privacy. However, existing FGL methods are predominantly designed for static graphs and rely on parameter averaging or distribution alignment, which implicitly assume that all features are equally transferable across clients, overlooking both the spatial and temporal heterogeneity and the presence of client-specific knowledge in real-world graphs. In this work, we identify that such assumptions create a vicious cycle of spurious representation entanglement, client-specific interference, and negative transfer, degrading generalization performance in Federated Learning over Dynamic Spatio-Temporal Graphs (FSTG). To address this issue, we propose a novel causality-inspired framework named SC-FSGL, which explicitly decouples transferable causal knowledge from client-specific noise through representation-level interventions. Specifically, we introduce a Conditional Separation Module that simulates soft interventions through client conditioned masks, enabling the disentanglement of invariant spatio-temporal causal factors from spurious signals and mitigating representation entanglement caused by client heterogeneity. In addition, we propose a Causal Codebook that clusters causal prototypes and aligns local representations via contrastive learning, promoting cross-client consistency and facilitating knowledge sharing across diverse spatio-temporal patterns. Experiments on five diverse heterogeneity Spatio-Temporal Graph (STG) datasets show that SC-FSGL outperforms state-of-the-art methods.

Problem

• Federated Learning
• Dynamic Spatio-Temporal Graphs
• Heterogeneity
• Negative Transfer
• Representation Entanglement

Innovation

• Causality-inspired Learning
• Federated Spatio-Temporal Graph Learning
• Representation Disentanglement
• Conditional Separation Module
• Causal Codebook