DropKAN: Dropout Kolmogorov–Arnold Networks

We propose DropKAN (Dropout Kolmogorov—Arnold Networks), a regularization method that introduces dropout masks at the edge level within Kolmogorov—Arnold Networks (KANs) layers, randomly masking a subset of activation outputs in the computation graph. Forward pass analysis reve...

Full description

Saved in:
Bibliographic Details
Main Author: Mohammed Ghaith Altarabichi
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11121822/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849233784344412160
author Mohammed Ghaith Altarabichi
author_facet Mohammed Ghaith Altarabichi
author_sort Mohammed Ghaith Altarabichi
collection DOAJ
description We propose DropKAN (Dropout Kolmogorov&#x2014;Arnold Networks), a regularization method that introduces dropout masks at the edge level within Kolmogorov&#x2014;Arnold Networks (KANs) layers, randomly masking a subset of activation outputs in the computation graph. Forward pass analysis reveals that DropKAN, when combined with scaling, accurately preserves the expected output signal magnitude in line with theoretical expectations. In contrast, conventional neuron-level Dropout&#x2014;with its scaling&#x2014;distorts signal propagation due to the nonlinear nature of KANs activations. Through extensive experiments on several classical benchmark datasets, DropKAN consistently achieves superior test accuracy compared to both unregularized KANs and KANs employing standard Dropout. Sensitivity analysis across dropout rates reveals DropKAN&#x2019;s robustness, effectively mitigating underfitting at high dropout levels by preserving partial neuron activity via edge-level masking. Additionally, DropKAN demonstrates enhanced sample efficiency under limited training data conditions, outperforming baselines on multiple large-scale datasets. In computer vision benchmarks (MNIST, Fashion MNIST, EMNIST, CIFAR-10), DropKAN further validates its regularization efficacy by consistently improving generalization over standard KAN and Dropout configurations. These results establish DropKAN as a principled and practical regularization technique for KANs architectures. Our implementation of DropKAN is available at (<uri>https://github.com/ghaith81/dropkan</uri>).
format Article
id doaj-art-6d723ea8d65d4b9184204586490e89a1
institution Kabale University
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-6d723ea8d65d4b9184204586490e89a12025-08-20T04:03:25ZengIEEEIEEE Access2169-35362025-01-011314160914161810.1109/ACCESS.2025.359755411121822DropKAN: Dropout Kolmogorov&#x2013;Arnold NetworksMohammed Ghaith Altarabichi0https://orcid.org/0000-0002-6040-2269Center for Applied Intelligent Systems Research, Halmstad University, Halmstad, SwedenWe propose DropKAN (Dropout Kolmogorov&#x2014;Arnold Networks), a regularization method that introduces dropout masks at the edge level within Kolmogorov&#x2014;Arnold Networks (KANs) layers, randomly masking a subset of activation outputs in the computation graph. Forward pass analysis reveals that DropKAN, when combined with scaling, accurately preserves the expected output signal magnitude in line with theoretical expectations. In contrast, conventional neuron-level Dropout&#x2014;with its scaling&#x2014;distorts signal propagation due to the nonlinear nature of KANs activations. Through extensive experiments on several classical benchmark datasets, DropKAN consistently achieves superior test accuracy compared to both unregularized KANs and KANs employing standard Dropout. Sensitivity analysis across dropout rates reveals DropKAN&#x2019;s robustness, effectively mitigating underfitting at high dropout levels by preserving partial neuron activity via edge-level masking. Additionally, DropKAN demonstrates enhanced sample efficiency under limited training data conditions, outperforming baselines on multiple large-scale datasets. In computer vision benchmarks (MNIST, Fashion MNIST, EMNIST, CIFAR-10), DropKAN further validates its regularization efficacy by consistently improving generalization over standard KAN and Dropout configurations. These results establish DropKAN as a principled and practical regularization technique for KANs architectures. Our implementation of DropKAN is available at (<uri>https://github.com/ghaith81/dropkan</uri>).https://ieeexplore.ieee.org/document/11121822/Dropout Kolmogorov-Arnold networksKolmogorov-Arnold networksdropoutregularization in neural networks
spellingShingle Mohammed Ghaith Altarabichi
DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
IEEE Access
Dropout Kolmogorov-Arnold networks
Kolmogorov-Arnold networks
dropout
regularization in neural networks
title DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
title_full DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
title_fullStr DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
title_full_unstemmed DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
title_short DropKAN: Dropout Kolmogorov&#x2013;Arnold Networks
title_sort dropkan dropout kolmogorov x2013 arnold networks
topic Dropout Kolmogorov-Arnold networks
Kolmogorov-Arnold networks
dropout
regularization in neural networks
url https://ieeexplore.ieee.org/document/11121822/
work_keys_str_mv AT mohammedghaithaltarabichi dropkandropoutkolmogorovx2013arnoldnetworks