Meta-GNN: On Few-shot Node Classification in Graph Meta-learning

Authors:

• [Fan Zhou]
• Chengtai Cao
• [Kunpeng Zhang]
• [Goce Trajcevski]
• [Ting Zhong]
• [Ji Geng]

Desccription of Meta-GNN

source_code for Meta-GNN (implement of Meta-GNN): Meta-GNN: On Few-shot Node Classification in Graph Meta-learning

Environment And Dependencies

PyTorch>=1.0.0
Install other dependencies: $ pip install -r requirement.txt

Dataset

We provide the citation network datasets under meta_gnn/data/.

Dataset Partition

We have shown the details of node partition in the paper.

Usage

For Meta-GNN:

$ cd meta_gnn/
$ python sgc.py 

For Few-shot:
(1)SGC:

$ cd sgc_fewshot/
$ python citation.py 

(2)GCN:

$ cd gcn_fewshot/
$ python citation.py 

(3)Deepwalk or Node2Vec:

$ cd graph_embedding_fewshot/
$ python deepwalk.py    //    $ python node2vec.py

(4)Graphsage

$ cd graphsage_fewshot/
$ python graphsage.py --dataset cora --gpu 0

Reference

Any comments and feedback are appreciated.