Deep Learning on Real-World Graphs

TL;DR

This paper introduces new models to enhance Graph Neural Networks' scalability, temporal handling, directionality, and robustness, making them more applicable to real-world, industrial-scale graph data.

Contribution

It presents a series of models addressing key real-world challenges in GNNs, bridging the gap between academic benchmarks and industrial applications.

Findings

SIGN enables scalable graph learning on large datasets.

TGN effectively models temporal dynamics in evolving graphs.

Dir-GNN handles directed and heterophilic networks.

Abstract

Graph Neural Networks (GNNs) have become a central tool for learning on graph-structured data, yet their applicability to real-world systems remains limited by key challenges such as scalability, temporality, directionality, data incompleteness, and structural uncertainty. This thesis introduces a series of models addressing these limitations: SIGN for scalable graph learning, TGN for temporal graphs, Dir-GNN for directed and heterophilic networks, Feature Propagation (FP) for learning with missing node features, and NuGget for game-theoretic structural inference. Together, these contributions bridge the gap between academic benchmarks and industrial-scale graphs, enabling the use of GNNs in domains such as social and recommender systems.