Firsov Nikita, Myasnikov Evgeny, Lobanov Valeriy, Khabibullin Roman, Kazanskiy Nikolay, Khonina Svetlana, Butt Muhammad A, Nikonorov Artem
Samara National Research University, Samara 443086, Russia.
Adyghe State University, Maykop 385000, Russia.
Sensors (Basel). 2024 Nov 30;24(23):7683. doi: 10.3390/s24237683.
In traditional neural network designs, a multilayer perceptron (MLP) is typically employed as a classification block following the feature extraction stage. However, the Kolmogorov-Arnold Network (KAN) presents a promising alternative to MLP, offering the potential to enhance prediction accuracy. In this paper, we studied KAN-based networks for pixel-wise classification of hyperspectral images. Initially, we compared baseline MLP and KAN networks with varying numbers of neurons in their hidden layers. Subsequently, we replaced the linear, convolutional, and attention layers of traditional neural networks with their KAN-based counterparts. Specifically, six cutting-edge neural networks were modified, including 1D (1DCNN), 2D (2DCNN), and 3D convolutional networks (two different 3DCNNs, NM3DCNN), as well as transformer (SSFTT). Experiments conducted using seven publicly available hyperspectral datasets demonstrated a substantial improvement in classification accuracy across all the networks. The best classification quality was achieved using a KAN-based transformer architecture.
在传统的神经网络设计中，多层感知器（MLP）通常在特征提取阶段之后用作分类模块。然而，柯尔莫哥洛夫 - 阿诺德网络（KAN）为MLP提供了一个有前景的替代方案，具有提高预测准确性的潜力。在本文中，我们研究了基于KAN的网络用于高光谱图像的逐像素分类。首先，我们比较了隐藏层中神经元数量不同的基线MLP和KAN网络。随后，我们用基于KAN的对应层替换了传统神经网络的线性、卷积和注意力层。具体而言，对六个前沿神经网络进行了修改，包括一维（1DCNN）、二维（2DCNN）和三维卷积网络（两种不同的3DCNN，NM3DCNN）以及变压器（SSFTT）。使用七个公开可用的高光谱数据集进行的实验表明，所有网络的分类准确率都有显著提高。使用基于KAN的变压器架构实现了最佳分类质量。
