By Xiaozhi Deng, Dong Huang, Ding-Hua Chen, Chang-Dong Wang and Jian-Huang Lai.
This is a Pytorch implementation of the paper. ()
The representation encoder of the proposed SACC is ResNet34.
Dataset | NMI | ACC | ARI |
---|---|---|---|
CIFAR-10 | 76.5 | 85.1 | 72.4 |
CIFAR-100 | 44.8 | 44.3 | 28.2 |
STL-10 | 69.1 | 75.9 | 62.6 |
ImageNet-10 | 87.7 | 90.5 | 84.3 |
ImageNet-dogs | 45.5 | 43.7 | 28.5 |
- python>=3.7
- pytorch>=1.6.0
- torchvision>=0.8.1
- munkres>=1.1.4
- numpy>=1.19.2
- opencv-python>=4.4.0.46
- pyyaml>=5.3.1
- scikit-learn>=0.23.2
- cudatoolkit>=11.0
There is a configuration file "config/config.yaml", where one can edit both the training and test options.
If you find SACC useful in your research, please consider citing:
@article{deng2023strongly,
title={Strongly augmented contrastive clustering},
author={Deng, Xiaozhi and Huang, Dong and Chen, Ding-Hua and Wang, Chang-Dong and Lai, Jian-Huang},
journal={Pattern Recognition},
volume={139},
pages={109470},
year={2023},
publisher={Elsevier}
}