About these tutorials

By Federico Reyes Gómez, Weihua Hu, and Jure Leskovec

Homepage: https://medium.com/stanford-cs224w

These Graph Machine Learning tutorials and case studies are a culmination of many months of work by the students of CS224W, Stanford University’s course on Machine Learning with Graphs, with a focus on the exciting field of Graph Neural Networks!

Graph Neural Networks (GNNs), are a new kind of deep learning model architecture able to reason over a variety of tasks and domains by leveraging the underlying structure of a problem in the form of a Graph. Given entities (nodes) and relations between them (edges), we can represent our dataset in this form, allowing us to learn representations of these entities and relations in a way that is useful for any sort of prediction task.

This year, the students worked on some incredible tutorials and case studies meant for anyone who is curious about how to utilize this powerful new tool in their research or applications.

These tutorials leverage PyG (PyTorch Geometric), the most powerful and easy to use library for machine learning on structured data. PyG allows you to define and train GNNs in minutes! All tutorials also link to a Google Colab with the code in the tutorial for you to follow along with as you read it!

The PyG package (source)

If you’d like to learn more about the class at Stanford, you can visit cs224w.stanford.edu and if you’re interested in diving deeper into GNNs, the whole course has been made available for free on YouTube!

Please let us know if there’s any more content you’d like to see and make sure to spread the word!

Have an application in mind?

Check out our tutorials by application

• Methodological Issues
• Recommendations
• Biology & Chemistry
• Finance
• Physics & Simulation
• General Methods
• Knowledge Graphs
• Traffic Prediction

Other Examples:

Tasks (See articles by task)

• Node-Level Classification & Regression
• Link Classification & Regression
• Graph-Level Classification & Regression
• Graph Auto-Encoding and Self-Supervision
• GNN Explainability

Model Architectures (Examples Linked)

• GraphSAGE
• Graph Attention Network (GAT)
• Graph Isomorphism Network (GIN)
• Spatio-Temporal Graph Attention Network (ST-GAT)
• GNNExplainer & SubgraphX
• LightGCN
• Variational Graph AutoEncoder (VGAE)

Miscellaneous Examples(See tutorials by application)

• Los Angeles Traffic Prediction
• Visual Question Answering
• Session-Based Purchase Recommendations
• Spotify Next Song Prediction
• Financial Network Fraud Detection
• Simulating Real-World Physics
• Biomedical Interaction Predictions between Drugs, Diseases, Proteins, and Genes
• Many many more!

Resources:

• PyTorch Geometric: https://pytorch-geometric.readthedocs.io/en/latest/
• CS224W Course Website: http://cs224w.stanford.edu/
• CS224W Free Course Videos: https://www.youtube.com/playlist?list=PL-Y8zK4dwCrQyASidb2mjj_itW2-YYx6-
• Homepage: https://medium.com/stanford-cs224w