A Unified Benchmark of Federated Learning with Kolmogorov-Arnold Networks for Medical Imaging

2025-04-28

AI Summary

This work addresses the joint optimization of privacy preservation and model performance in federated learning (FL) for medical imaging. We establish the first unified FL benchmark tailored to medical imaging and systematically evaluate Kolmogorov–Arnold Networks (KANs) for blood cell classification. Under six mainstream FL algorithms—including FedAvg and FedProx—and realistic non-IID data partitions, we compare KANs against multilayer perceptrons (MLPs), conduct hyperparameter sensitivity analysis, and perform ablation studies. Results demonstrate that shallow, narrow KAN architectures significantly outperform MLPs in non-IID settings, achieving higher accuracy and superior robustness. Joint optimization of grid size and network width markedly improves convergence speed and generalization. This study establishes KANs as a novel paradigm for privacy-preserving medical AI, advancing the adoption of interpretable, lightweight, and FL-native neural networks in clinical AI applications.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Security and privacy of machine learning and AI applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Federated Learning (FL) enables model training across decentralized devices without sharing raw data, thereby preserving privacy in sensitive domains like healthcare. In this paper, we evaluate Kolmogorov-Arnold Networks (KAN) architectures against traditional MLP across six state-of-the-art FL algorithms on a blood cell classification dataset. Notably, our experiments demonstrate that KAN can effectively replace MLP in federated environments, achieving superior performance with simpler architectures. Furthermore, we analyze the impact of key hyperparameters-grid size and network architecture-on KAN performance under varying degrees of Non-IID data distribution. Additionally, our ablation studies reveal that optimizing KAN width while maintaining minimal depth yields the best performance in federated settings. As a result, these findings establish KAN as a promising alternative for privacy-preserving medical imaging applications in distributed healthcare. To the best of our knowledge, this is the first comprehensive benchmark of KAN in FL settings for medical imaging task.

Problem

• Evaluating KAN vs MLP in federated learning for medical imaging
• Analyzing hyperparameters' impact on KAN performance in Non-IID data
• Establishing KAN as a privacy-preserving alternative in healthcare FL

Innovation

• KAN replaces MLP in federated learning
• Optimizing KAN width enhances FL performance
• First FL benchmark of KAN for medical imaging