https://github.com/rootlu/L2P-GNN

Codes and datasets for AAAI-2021 paper "Learning to Pre-train Graph Neural Networks"
https://github.com/rootlu/L2P-GNN

aaai2021 gnn meta-learning pre-train

Last synced: 14 days ago
JSON representation

Codes and datasets for AAAI-2021 paper "Learning to Pre-train Graph Neural Networks"

• Host: GitHub
• URL: https://github.com/rootlu/L2P-GNN
• Owner: rootlu
• Created: 2020-12-03T08:07:21.000Z (over 3 years ago)
• Default Branch: main
• Last Pushed: 2021-02-27T05:56:52.000Z (over 3 years ago)
• Last Synced: 2023-03-04T12:34:41.910Z (over 1 year ago)
• Topics: aaai2021, gnn, meta-learning, pre-train
• Language: Python
• Homepage:
• Size: 19.7 MB
• Stars: 79
• Watchers: 2
• Forks: 15
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Lists

• StarryDivineSky - rootlu/L2P-GNN

README

        
# Learning to Pre-train Graph Neural Networks

This repository is the official implementation of AAAI-2021 paper [Learning to Pre-train Graph Neural Networks](https://yuanfulu.github.io/publication/AAAI-L2PGNN.pdf)

## Requirements

To install requirements:

```setup

pip install -r requirements.txt

```

## Dataset

All the necessary data files can be downloaded from the following links.

For Biology dataset, download from [Google Drive](https://drive.google.com/drive/folders/18vpBvSajIrme2xsbx8Oq8aTWIWMlSgJT?usp=sharing) and [BaiduYun](https://pan.baidu.com/s/1Yv6dN7F1jgTSz9-nU1eN-A) (Extraction code: j97n), unzip it, and put the  under `data/bio/`.

The new compilation of bibliographic graphs, i.e., PreDBLP,  download from [Google Drive](https://drive.google.com/drive/folders/18vpBvSajIrme2xsbx8Oq8aTWIWMlSgJT?usp=sharing) and [BaiduYun](https://pan.baidu.com/s/1Yv6dN7F1jgTSz9-nU1eN-A) (Extraction code: j97n), unzip it, and move the `dblp.graph` file to `data/dblp/unsupervised/processed/` and the `dblpfinetune.graph` file to `data/dblp/supervised/processed/`, respectively.

Also, to avoid the "file incomplete" errors caused by compressed files, we also upload the uncompressed dblp dataset at [BaiduYun](https://pan.baidu.com/s/1Yv6dN7F1jgTSz9-nU1eN-A) (Extraction code: j97n).

## Training

To pre-train L2P-GNN on Biology dataset w.r.t. GIN model, run this command:

```shell

python main.py --dataset DATASET  --gnn_type GNN_MODEL --model_file PRE_TRAINED_MODEL_NAME --device 1

```

The pre-trained models are saved into `res/DATASET/` .

## Evaluation

To fine-tune L2P-GNN on Biology dataset, run:

```shell

python eval_bio.py --dataset DATASET  --gnn_type GNN_MODEL --emb_trained_model_file EMB_TRAINED_FILE --pre_trained_model_file GNN_TRAINED_FILE --pool_trained_model_file POOL_TRAINED_FILE --result_file RESULT_FILE --device 1

```

The results w.r.t 10 random running seeds are saved into `res/DATASET/finetune_seed(0-9)/`

## Results

To analysis results of downstream tasks, run:

```shell

python result_analysis.py  --dataset DATASET --times SEED_NUM

```

where `SEED_NUM` is the number of random seed ranging from 0 to 9, thus it is usually set to 10.

## Reproducing results in the paper

Our results in the paper can be reproduced by directly running:

```shell

python eval_bio.py --dataset bio --gnn_type gin --emb_trained_model_file co_adaptation_5_300_gin_50_emb.pth --pre_trained_model_file co_adaptation_5_300_gin_50_gnn.pth --pool_trained_model_file co_adaptation_5_300_gin_50_pool.pth --result_file co_adaptation_5_300_gin_50 --device 0

```

and

```shell

python eval_dblp.py --dataset dblp --gnn_type gin --split random --emb_trained_model_file co_adaptation_5_300_s50q30_gin_20_emb.pth --pre_trained_model_file co_adaptation_5_300_s50q30_gin_20_gnn.pth --pool_trained_model_file co_adaptation_5_300_s50q30_gin_20_pool.pth --result_file co_adaptation_5_300_s50q30_gin_20  --device 0 --dropout_ratio 0.1

```