![[2601.22669] Beyond Fixed Rounds: Data-Free Early Stopping for Practical Federated Learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2601.22669] Beyond Fixed Rounds: Data-Free Early Stopping for Practical Federated Learning
Summary
This paper introduces a data-free early stopping framework for federated learning, enhancing efficiency and privacy by eliminating the need for validation data during training.
Why It Matters
Federated learning is crucial for privacy-preserving machine learning, yet its reliance on fixed rounds and validation data limits practical applications. This research offers a novel solution to optimize performance while maintaining data privacy, making federated learning more accessible and efficient.
Key Takeaways
- Proposes a data-free early stopping method for federated learning.
- Reduces computational costs and privacy risks associated with validation data.
- Demonstrates performance improvements of over 12% compared to traditional methods.
Computer Science > Machine Learning arXiv:2601.22669 (cs) [Submitted on 30 Jan 2026 (v1), last revised 26 Feb 2026 (this version, v2)] Title:Beyond Fixed Rounds: Data-Free Early Stopping for Practical Federated Learning Authors:Youngjoon Lee, Hyukjoon Lee, Seungrok Jung, Andy Luo, Jinu Gong, Yang Cao, Joonhyuk Kang View a PDF of the paper titled Beyond Fixed Rounds: Data-Free Early Stopping for Practical Federated Learning, by Youngjoon Lee and 6 other authors View PDF HTML (experimental) Abstract:Federated Learning (FL) facilitates decentralized collaborative learning without transmitting raw data. However, reliance on fixed global rounds or validation data for hyperparameter tuning hinders practical deployment by incurring high computational costs and privacy risks. To address this, we propose a data-free early stopping framework that determines the optimal stopping point by monitoring the task vector's growth rate using solely server-side parameters. The numerical results on skin lesion/blood cell classification demonstrate that our approach is comparable to validation-based early stopping across various state-of-the-art FL methods. In particular, the proposed framework requires an average of 45/12 (skin lesion/blood cell) additional rounds to achieve over 12.3%/8.9% higher performance than early stopping based on validation data. To the best of our knowledge, this is the first work to propose an data-free early stopping framework for FL methods. Comments: Subjects: Mac...
