TGHE: Template-based Graph Homomorphic Encryption for Privacy-Preserving GNN Inference in Edge-Cloud Systems

2026-06-25

AI Summary

This work addresses the inefficiency of existing homomorphic encryption–based graph neural network (GNN) inference, which scales poorly with global graph size and thus struggles to support privacy-preserving computation on million-scale financial graphs. The authors propose an ego-centric encrypted GNN framework that leverages a newly identified phenomenon: local computation trees in transaction graphs converge to a limited set of structural templates. By normalizing these isomorphic subgraphs and packing them into shared CKKS ciphertexts, the method enables fully SIMD-parallelized inference. Combined with approximate template fitting and topological collapse strategies, the approach effectively captures long-tail structural variations. Evaluated on the DGraphFin dataset with 3.7 million nodes, the proposed method achieves a 66.9× speedup over per-node baselines while incurring less than 0.002 AUC degradation.

Abstract

Existing homomorphic encryption (HE)-based GNN systems adopt a graph-centric paradigm that couples per-query cost to global graph size, limiting evaluations to at most ~20k nodes and making them incompatible with dynamic, large-scale financial graphs. We propose TGHE (Template-based Graph Homomorphic Encryption), an ego-centric framework that resolves this by exploiting a template phenomenon: local computation trees in transaction graphs converge into a small set of structural shapes. TGHE canonicalizes ego-graphs at the edge and packs structurally identical trees into shared CKKS ciphertexts for SIMD-parallel encrypted inference, with two long-tail optimizers (Approximate Template Fitting and Topology Collapse) ensuring full SIMD coverage. On DGraphFin (3.7M nodes, 4.3M edges), TGHE-Collapse achieves a 66.9x speedup over the sequential encrypted baseline with less than 0.002 AUC loss.

Problem

• homomorphic encryption
• graph neural networks
• privacy-preserving inference
• large-scale graphs
• edge-cloud systems

Innovation

• Graph Homomorphic Encryption
• Ego-centric GNN
• Structural Template
• SIMD Parallelism
• Privacy-Preserving Inference