Implementation of Kolmogorov-Arnold Network (KAN) for Graphs. Any communications, collaborations, issues, PRs are welcomed. The contributors will be listed here. Please contact [email protected]. If you find this repository useful to your research or work, it is really appreciate to star this repository. ❤️
Table of Contents
KAN4Graph is implemented with Python3.8.16 and 1 NVIDIA Tesla V100 SXM2 16 GB
Python package information is summarized in requirements.txt:
- torch==1.7.1
- tqdm==4.59.0
- numpy==1.19.2
- munkres==1.1.4
- scikit_learn==1.2.0
| Dataset | Type | # Nodes | # Feature Dimensions | # Edges | # Classes |
|---|---|---|---|---|---|
| BAT | Attribute Graph | 131 | 81 | 1038 | 4 |
| UAT | Attribute Graph | 1,190 | 239 | 13,599 | 4 |
| EAT | Attribute Graph | 399 | 203 | 5,994 | 4 |
still updating...
clone this repository and change directory to KAN4Graph
git clone https://github.com/yueliu1999/KAN4Graph.git
cd ./KAN4Graph
run codes
python train.py
| Dataset | Metric | Score |
|---|---|---|
| BAT | ACC | 77.86 |
| NMI | 54.48 | |
| ARI | 52.33 | |
| F1 | 77.34 | |
| UAT | ACC | 57.05 |
| NMI | 25.49 | |
| ARI | 24.97 | |
| F1 | 55.80 | |
| EAT | ACC | 57.87 |
| NMI | 34.16 | |
| ARI | 27.52 | |
| F1 | 58.09 |
still updating...
Our code are partly based on the following GitHub repository. Thanks for their awesome works.
- pykan: the official implement of KAN.
- fast-kan: the implement of KAN (fast version).
- Awesome Deep Graph Clustering: a collection of deep graph clustering (papers, codes, and datasets).
- SCGC: the official implement of Simple Contrastive Graph Clustering (SCGC) model.
If you find this repository helpful, please cite our paper.
@article{liuyue_SCGC,
title={Simple contrastive graph clustering},
author={Liu, Yue and Yang, Xihong and Zhou, Sihang and Liu, Xinwang},
journal={IEEE Transactions on Neural Networks and Learning Systems},
year={2023},
publisher={IEEE}
}
@misc{KAN4Graph,
author = {Yue Liu},
title = {KAN4Graph},
year = {2024},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/yueliu1999/KAN4Graph}},
}