Federated learning improves hypergraph predictions with less communication
More Is Better? Exploring Latent Correspondence in Federated Hypergraph Learning
Machine LearningDistributed, Parallel, and Cluster Computing
Summary
Working with complex data structures called hypergraphs across multiple devices is hard because the connections between parts on different devices can match in many ways. The authors show it's easier to match pairs of devices but much harder to match multiple devices at once. They create a new method called FedASTRA that learns good representations of this data while only sharing model details, not raw data. Their method performs well on big datasets, efficiently balancing accuracy and communication.
What this means in practice
- •For distributed ai engineers: Improve predictive models over complex network data across multiple devices without sharing raw data, reducing communication while maintaining accuracy.
- •For social network analysts: Analyze interconnected group data spread across servers with limited data sharing, enhancing insights from multi-client hypergraph structures.
Authors
Bo Yan, Siming Zhang, Siqi Li, Hang Yu
Abstract
Federated hypergraph learning with latent cross-client hyperedge correspondence poses fundamental identifiability and computational challenges, as the same local release can correspond to multiple global hypergraph structures. We characterize the tractability of recovering these latent correspondences, showing tractable pairwise assignment but generally intractable multi-client hyperedge matching. Building on this characterization, we derive a cross-entropy decomposition that reveals when local prediction can outperform predictors fitted with additional cross-client information. Motivated by this principle, we introduce Federated Adaptive Spectral Training with Representation Alignment (FedASTRA), which learns correspondence-invariant diffusion representations through adaptive heat diffusion and masked latent prediction while communicating model parameters only. We establish its asymptotic optimality with respect to the permitted local information and prove global population-risk convergence under explicit conditions on recovery, representation, and update alignment. Experiments on eight real-world datasets demonstrate the effectiveness of FedASTRA against identifier-free federated graph and hypergraph methods. On ogbn-papers100M, FedASTRA reaches the performance-communication Pareto frontier, achieving 50.07 Accuracy and 29.73 Macro-F1 while communicating only 59.60 MiB per round.