From Coarse to Fine: Enable Comprehensive Graph Self-supervised Learning with Multi-granular Semantic Ensemble

This is the official implementaiton of the paper From Coarse to Fine: Enable Comprehensive Graph Self-supervised Learning with Multi-granular Semantic Ensemble published in ICML 2024. Our paper can be found at Openreivew Link.

Dependency

The packages we use include,

- python >= 3.8.5
- Pytorch >= 1.10.1
- PyTorch Geometric == 2.0.4;
- PyGCL == 0.1.2
- Numpy >= 1.20.1
- pandas 
- scikit-learn >= 0.24.1
- wandb
- tqdm
- rdkit

Please also install faiss if you want to run the hierarchical K-means clustering.

Instruction

Dataset Preparation:

We follow the instruction of Pre-training GNNs to prepare the datasets in chemistry and biology domain. All the datasets can be downloaded from the project page.

Pre-training:

First, we need to create a folder for storing pre-triained teacher model checkpoint in the root direction of this project.

mkdir pretrain

Then, given the specfic pre-trained teacher model, like GraphCL, we need to download its pre-trained model checkpoint and save it in the foler. After that, we can run the following command to train the student models.

python src/models/MGSE/pretrain.py --teacher_model graphcl

Fine-tuning

Finally, we can run the following command to fine-tune the student models on the downstream datasets (e.g., bbbp) and do the evaluation.

python src/models/MGSE/finetune.py --dataset bbbp

Citation

Please cite our work if you find it useful:

@inproceedings{
wen2024from,
title={From Coarse to Fine: Enable Comprehensive Graph Self-supervised Learning with Multi-granular Semantic Ensemble},
author={Qianlong Wen and Mingxuan Ju and Zhongyu Ouyang and Chuxu Zhang and Yanfang Ye},
booktitle={Forty-first International Conference on Machine Learning},
year={2024}
}