Title:  Scalable and Verifiable Foundations for Graph Learning

Date: Thursday, April 23, 2026

Time: 10:00 AM – 12:00 PM (Eastern Time)

Location (Virtual): https://gatech.zoom.us/j/95898773125

Mingyu Guan

Ph.D. Student

School of Computer Science

Georgia Institute of Technology

Committee Members

Dr. Taesoo Kim (Advisor), School of Cybersecurity and Privacy, Georgia Institute of Technology

Dr. Anand Iyer (Co-advisor), School of Computer Science, Georgia Institute of Technology

Dr. Ada Gavrilovska, School of Computer Science, Georgia Institute of Technology

Dr. Kexin Rong, School of Computer Science, Georgia Institute of Technology

Dr. Jay Stokes, Microsoft Research

Abstract

Graph neural networks (GNNs) are increasingly deployed in high-stakes domains such as fraud detection, traffic prediction, and social network analysis. However, real-world graphs are heterogeneous, dynamic, and privacy-critical, posing challenges in structural complexity, scalability, and verifiability across the model lifecycle. This thesis argues that graph learning problems contain inherent structural regularities that, when explicitly leveraged in model and system design, enable scalable and verifiable graph learning across critical stages of the model lifecycle. First, we present HetTree, a scalable heterogeneous GNN that exploits the natural tree hierarchy among metapaths by constructing a semantic tree representation and introducing a subtree attention mechanism to capture hierarchical relationships with low computation and memory overhead. Second, we present ReD, a system for scalable dynamic GNN training that leverages independence across snapshot sequences to enable sequence-parallel training without cross-machine communication, supported by sequence-first mini-batching and a two-level cache store. Third, we present TAITEE, a system that establishes verifiable training provenance by integrating training recording with Confidential Computing, recording comprehensive provenance via standard training APIs and generating cryptographically signed training certificates from Trusted Execution Environments with minimal overhead. Together, the three contributions span the critical graph learning lifecycle: what to compute, how to compute it efficiently, and whether the training was conducted as claimed — providing scalable and verifiable foundations for deploying graph learning models in practice.