Alignahead

This repo covers the implementation of the following IJCNN 2022 paper: Alignahead: Online Cross-Layer Knowledge Extraction on Graph Neural Networks

Installation

This repo was tested with Python 3.7, PyTorch 1.7.0, CUDA 10.1 and dgl 0.4.2.

Running

Train multiple student models

python main.py --model_num 2 --strategy alignahead --gpu 0 --model_name GAT --a 1

where the flags are explained as:

• --model_num: the number of student models.
• --strategy: the strategy of alignment,[alignahead,OC].
• --model_name: the structure of student models.
• --a: the hyper-parameter of alignment loss.

Citation

If you find this repository useful, please consider citing the following paper:

@article{guo2022alignahead,
  title={Alignahead: Online Cross-Layer Knowledge Extraction on Graph Neural Networks},
  author={Guo, Jiongyu and Chen, Defang and Wang, Can},
  journal={International Joint Conference on Neural Networks (IJCNN)},
  year={2022}
}