FedTAD: Topology-aware Data-free Knowledge Distillation for Subgraph Federated Learning

arXiv:2404.14061v1 Announce Type: new
Abstract: Subgraph federated learning (subgraph-FL) is a new distributed paradigm that facilitates the collaborative training of graph neural networks (GNNs) by multi-client subgraphs. Unfortunately, a significant challenge of subgraph-FL arises from subgraph heterogeneity, which stems from node and topology variation, causing the impaired performance of the global GNN. Despite various studies, they have not yet thoroughly investigated the impact mechanism of subgraph heterogeneity. To this end, we decouple node and topology variation, revealing that they correspond …

abstract arxiv challenge client collaborative cs.ai cs.db cs.lg cs.si data distillation distributed federated learning free global gnns graph graph neural networks knowledge networks neural networks node paradigm performance topology training type variation