We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Subgraph Federated Learning with Missing Neighbor Generation

Abstract: Graphs have been widely used in data mining and machine learning due to their unique representation of real-world objects and their interactions. As graphs are getting bigger and bigger nowadays, it is common to see their subgraphs separately collected and stored in multiple local systems. Therefore, it is natural to consider the subgraph federated learning setting, where each local system holds a small subgraph that may be biased from the distribution of the whole graph. Hence, the subgraph federated learning aims to collaboratively train a powerful and generalizable graph mining model without directly sharing their graph data. In this work, towards the novel yet realistic setting of subgraph federated learning, we propose two major techniques: (1) FedSage, which trains a GraphSage model based on FedAvg to integrate node features, link structures, and task labels on multiple local subgraphs; (2) FedSage+, which trains a missing neighbor generator along FedSage to deal with missing links across local subgraphs. Empirical results on four real-world graph datasets with synthesized subgraph federated learning settings demonstrate the effectiveness and efficiency of our proposed techniques. At the same time, consistent theoretical implications are made towards their generalization ability on the global graphs.
Comments: Accepted to NeurIPS 2021 (spotlight presentation)
Subjects: Machine Learning (cs.LG); Social and Information Networks (cs.SI)
Cite as: arXiv:2106.13430 [cs.LG]
  (or arXiv:2106.13430v3 [cs.LG] for this version)

Submission history

From: Ke Zhang [view email]
[v1] Fri, 25 Jun 2021 04:58:55 GMT (182kb,D)
[v2] Thu, 22 Jul 2021 13:37:09 GMT (374kb,D)
[v3] Mon, 18 Oct 2021 15:31:17 GMT (377kb,D)
[v4] Tue, 26 Oct 2021 19:31:23 GMT (605kb,D)
[v5] Thu, 28 Oct 2021 01:59:09 GMT (606kb,D)
[v6] Sat, 6 Nov 2021 13:48:57 GMT (606kb,D)

Link back to: arXiv, form interface, contact.