Capsule-ConvKAN: A Hybrid Neural Approach to Medical Image Classification

TL;DR

This paper introduces Capsule-ConvKAN, a hybrid neural network architecture combining capsule networks and Kolmogorov-Arnold networks to enhance feature representation and classification accuracy in biomedical image analysis.

Contribution

The paper proposes a novel hybrid neural network architecture, Capsule-ConvKAN, integrating dynamic routing and spatial hierarchy with flexible function approximation for improved medical image classification.

Findings

Capsule-ConvKAN achieved 91.21% accuracy on histopathological data.

The hybrid model outperformed traditional CNNs and individual architectures.

Capsule-ConvKAN effectively captures spatial patterns and manages complex features.

Abstract

This study conducts a comprehensive comparison of four neural network architectures: Convolutional Neural Network, Capsule Network, Convolutional Kolmogorov-Arnold Network, and the newly proposed Capsule-Convolutional Kolmogorov-Arnold Network. The proposed Capsule-ConvKAN architecture combines the dynamic routing and spatial hierarchy capabilities of Capsule Network with the flexible and interpretable function approximation of Convolutional Kolmogorov-Arnold Networks. This novel hybrid model was developed to improve feature representation and classification accuracy, particularly in challenging real-world biomedical image data. The architectures were evaluated on a histopathological image dataset, where Capsule-ConvKAN achieved the highest classification performance with an accuracy of 91.21%. The results demonstrate the potential of the newly introduced Capsule-ConvKAN in capturing spatial patterns, managing complex features, and addressing the limitations of traditional convolutional models in medical image classification.