How to use edge features in Graph Neural Networks (and PyTorch Geometric)

In this video I talk about edge weights, edge types and edge features and how to include them in Graph Neural Networks. :)

▬▬ Papers ▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Edge types:
- Modeling Relational Data with Graph Convolutional Network (https://arxiv.org/pdf/1703.06103.pdf)
- GNN-FiLM: Graph Neural Networks with Feature-wise Linear Modulation (https://arxiv.org/pdf/1906.12192.pdf)

Multidim. edge features:
- Neural Message Passing for Quantum Chemistry (https://arxiv.org/pdf/1704.01212.pdf)
- Principal Neighbourhood Aggregation for Graph Nets (https://arxiv.org/pdf/2004.05718.pdf)
- Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties (https://journals.aps.org/prl/abstract...)

Edge feature embeddings:
- NENN: Incorporate Node and Edge Features in Graph Neural Networks (http://proceedings.mlr.press/v129/yan...)
- Exploiting Edge Features in Graph Neural Networks (https://arxiv.org/pdf/1809.02709.pdf)

▬▬ Timestamps ▬▬▬▬▬▬▬▬▬▬▬
00:00 Introduction
05:10 Edge weights
06:10 Edge types / relations
09:21 Multidim. edge features
12:04 Edge feature embeddings
13:52 Pytorch Geometric

▬▬ Support me if you like 🌟
►Link to this channel: https://bit.ly/3zEqL1W
►Support me on Patreon: https://bit.ly/2Wed242
►Buy me a coffee on Ko-Fi: https://bit.ly/3kJYEdl