Generalizable Hyperparameter Optimization for Federated Learning on Non-IID Cancer Images

2026-01-19

AI Summary

This work addresses the challenge of limited model generalization and sensitivity to hyperparameter selection in federated learning with non-IID cancer histopathology images. The authors propose a cross-dataset hyperparameter transfer strategy: first, Bayesian optimization is applied under a centralized setting to identify optimal hyperparameters separately for ovarian and colorectal cancer datasets; then, a simple yet effective aggregation heuristic—averaging learning rates and selecting the mode for optimizer and batch size—is used to construct a universal configuration, which is subsequently transferred to the federated learning scenario. Experimental results demonstrate that this approach significantly improves classification performance under non-IID conditions, confirming the efficacy and practicality of cross-dataset hyperparameter transfer in federated medical image analysis.

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
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Deep learning for cancer histopathology training conflicts with privacy constraints in clinical settings. Federated Learning (FL) mitigates this by keeping data local; however, its performance depends on hyperparameter choices under non-independent and identically distributed (non-IID) client datasets. This paper examined whether hyperparameters optimized on one cancer imaging dataset generalized across non-IID federated scenarios. We considered binary histopathology tasks for ovarian and colorectal cancers. We perform centralized Bayesian hyperparameter optimization and transfer dataset-specific optima to the non-IID FL setup. The main contribution of this study is the introduction of a simple cross-dataset aggregation heuristic by combining configurations by averaging the learning rates and considering the modal optimizers and batch sizes. This combined configuration achieves a competitive classification performance.

Problem

• Federated Learning
• Hyperparameter Optimization
• Non-IID
• Cancer Histopathology
• Generalization

Innovation

• Federated Learning
• Hyperparameter Optimization
• Non-IID Data
• Cross-dataset Generalization
• Cancer Histopathology