Evaluating Federated Kolmogorov-Arnold Networks on Non-IID Data
Abstract
Federated Kolmogorov-Arnold Networks (F-KANs) have already been proposed, but their assessment is at an initial stage. We present a comparison between KANs (using B-splines and Radial Basis Functions as activation functions) and Multi-Layer Perceptrons (MLPs) with a similar number of parameters for 100 rounds of federated learning in the MNIST classification task using non-IID partitions with 100 clients. After 15 trials for each model, we show that the best accuracies achieved by MLPs can be achieved by Spline-KANs in half of the time (in rounds), with just a moderate increase in computing time.
Cite
Text
Sasse and de Farias. "Evaluating Federated Kolmogorov-Arnold Networks on Non-IID Data." NeurIPS 2024 Workshops: LXAI, 2024.Markdown
[Sasse and de Farias. "Evaluating Federated Kolmogorov-Arnold Networks on Non-IID Data." NeurIPS 2024 Workshops: LXAI, 2024.](https://mlanthology.org/neuripsw/2024/sasse2024neuripsw-evaluating/)BibTeX
@inproceedings{sasse2024neuripsw-evaluating,
title = {{Evaluating Federated Kolmogorov-Arnold Networks on Non-IID Data}},
author = {Sasse, Arthur M. and de Farias, Claudio Miceli},
booktitle = {NeurIPS 2024 Workshops: LXAI},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/sasse2024neuripsw-evaluating/}
}