BuffGraph is an innovative model to adjust message passing of each edge in the graph, meticulously crafted to address the heterophily challenge of imbalanced node classification within graph data.

• Code Overview
• Requirements
• Running Code
• License

This code repository consists of several folders:

• model: BuffGraph code.
• ablation study: Ablation study code of BuffGraph: deduct heterophily loss; deduct original graph; deduct updating message passing.
• parameters sensitivity: Finetuning mixup parameter alpha and heterophily loss parameter lambda code.
• scalability study: Scalability study code on Coauthor-Physics.

We list main requirements of this repository below. For full requirements, please refer to the provided environment.yml file

• dgl==1.1.0
• torch==2.0.0+cu117
• torch-cluster==1.6.1+pt20cu117
• torch-geometric==2.3.0
• torch-scatter==2.1.1+pt20cu117
• torch-sparse==0.6.17+pt20cu117
• torch-spline-conv==1.2.2+pt20cu117
• torchaudio==2.0.1+cu117
• torchmetrics==0.11.4
• torchvision==0.15.1+cu117
• transformers==4.30.2

For example, if you want to run BuffGraph code on the dataset Amazon-Computers, go to the model folder first, then run the following commands:

``` 
python pre_train.py --dataset="amazoncomputers"
python process_edges.py --dataset="amazoncomputers"
python buffgraph.py --dataset="amazoncomputers"
```

This dataset is under license CC BY-SA 4.0.