KAC: Kolmogorov-Arnold Classifier for Continual Learning

PyTorch code for the CVPR 2025 Highliaght paper:
KAC: Kolmogorov-Arnold Classifier for Continual Learning
Yusong Hu, Zichen Liang, Fei Yang, Qibin Hou, Xialei Liu and Ming-ming Cheng
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025 Highlight
[paper]

Setup and Datasets Preparation

Please following the setup steps in CODA-Prompt

Setup

• Install anaconda: https://www.anaconda.com/distribution/
• set up conda environment w/ python 3.8, ex: conda create --name coda python=3.8
• conda activate coda
• sh install_requirements.sh
• NOTE: this framework was tested using torch == 2.0.0 but should work for previous versions

Datasets

• Create a folder data/
• CUB200: retrive from: https://www.kaggle.com/datasets/cyizhuo/cub-200-2011-by-classes-folder
• ImageNet-R: retrieve from: https://github.com/hendrycks/imagenet-r
• DomainNet: retrieve from: http://ai.bu.edu/M3SDA/

Training

All commands should be run under the project root directory. The scripts are set up for 4 GPUs but can be modified for your hardware.

sh experiments/cub.sh
sh experiments/imagenet-r.sh
sh experiments/domainnet.sh

The methods ( L2P, Dual-Prompt and CODA-Prompt ) with a linear classifier and with KAC will be evaluated.

Results

Results will be saved in a folder named outputs/. To get the final average accuracy, retrieve the final number in the file outputs/**/results-acc/global.yaml

Citation

If you found our work useful for your research, please cite our work:

@inproceedings{hu2025kac,
title={Kac: Kolmogorov-arnold classifier for continual learning},
author={Hu, Yusong and Liang, Zichen and Yang, Fei and Hou, Qibin and Liu, Xialei and Cheng, Ming-Ming},
booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference},
pages={15297--15307},
year={2025}
}

Thanks

The code is developed based on CODA-Prompt, and the implementation of KAC follows Fast-KAN.