Kolmogorov-Arnold networks for metal surface defect classification

TL;DR

This paper introduces Kolmogorov-Arnold Networks (KAN) for metal surface defect classification, demonstrating improved accuracy and efficiency over CNNs by leveraging spline-based function approximation.

Contribution

The paper proposes a novel application of KAN for defect classification, showing it outperforms CNNs in accuracy and convergence speed with fewer parameters.

Findings

KAN achieves higher accuracy than CNNs.

KAN requires fewer parameters and converges faster.

KAN effectively classifies various surface defects.

Abstract

This paper presents the application of Kolmogorov-Arnold Networks (KAN) in classifying metal surface defects. Specifically, steel surfaces are analyzed to detect defects such as cracks, inclusions, patches, pitted surfaces, and scratches. Drawing on the Kolmogorov-Arnold theorem, KAN provides a novel approach compared to conventional multilayer perceptrons (MLPs), facilitating more efficient function approximation by utilizing spline functions. The results show that KAN networks can achieve better accuracy than convolutional neural networks (CNNs) with fewer parameters, resulting in faster convergence and improved performance in image classification.