Author(s): Michael Shapiro MD MSc

Originally published on Towards AI.

A Practical Guide to Data Preparation and Training Graph Neural Networks on Knowledge Graphs

Surprisingly, there aren’t many good resources on how to obtain embeddings for nodes in a Knowledge Graph. The challenge lies in the unique nature of KGs in the context of graph learning — they’re large, heterogeneous graphs with multiple edge types and no node features. An abomination, if you ask anyone working in graph learning. (If you’re not a Medium Member, read it for free here)

This article outlines the process of training a Graph Neural Network (GNN) on a Knowledge Graph, focusing on the use of PyTorch and PyTorch Geometric. It discusses the setup, data preparation, and challenges encountered, ultimately guiding readers through the necessary steps, including environment setup, model creation, and training, supplemented with coding examples and explanations. Furthermore, it highlights key considerations like input management, architecture optimization, and training methodologies to improve overall model efficiency.

Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI