A meta-learning based framework for few-shot learning on graph classification. For more details, please refer to our paper "Adaptive-Step Graph Meta-Learner for Few-Shot Graph Classification" that was published on CIKM2020.

• python 3.6
• pytorch 1.3.0
• torch-cluster 1.4.5
• torch-geometric 1.3.2
• torch-scatter 1.4.0
• torch-sparse 0.4.3

In experiments, we use TRIANGLES (click to download) with the partition rules of Jatin Chauhan's paper. Extract the downloaded file and put the files in ./data/TRIANGLES. For Graph-R52 and COIL-DEL can also be downloaded now.

If you need origin dataset of TRIANGLES, you can download it from here

To train the AS-MAML framework with GraghSAGE and SAGPool on TRIANGLES dataset, please run:

python main.py

To test the trained model, please run the following code with specified model path:

python test.py --model_dir *