milkigit / graphml-tutorials

Tutorials on Machine Learning for Graphs

Tutorials for Machine Learning on Graphs

Contributors

• Payal Chandak ([email protected])
• Haoxin Li ([email protected])
• Min Jean Cho ([email protected])
• Pavlin Policar ([email protected])
• Marinka Zitnik ([email protected])

Overview

Graph machine learning provides a powerful toolbox to learn representations from any arbitrary graph structure and use learned representations for a variety of downstream tasks. These tutorials aim to:

1. Introduce the concept of graph neural networks (GNNs).
2. Discuss the theoretical motivation behind different GNN architectures.
3. Provide implementations of these architectures.
4. Apply the architectures to key prediction problems on interconnected data in science and medicine.
5. Provide end-to-end real-world examples of graph machine learning.

Requirements

Recent versions of NumPy, PyTorch, PyTorch Geometric and Jupyter are required. All the required packages can be installed using the following command:

Installation

1. git clone https://github.com/mims-harvard/graphml-tutorials.git
2. cd graphml-tutorials
3. conda env create -f environment.yml
4. conda activate graphml_venv

Contributing

Pull requests are welcome.

License

MIT