FedPSA: Modeling Behavioral Staleness in Asynchronous Federated Learning

2026-02-16

AI Summary

In asynchronous federated learning, model staleness often degrades performance, yet existing approaches assess staleness only through coarse-grained round differences, neglecting the internal state of the model. This work proposes FedPSA, a novel framework that introduces a fine-grained staleness metric based on parameter sensitivity and incorporates a dynamic momentum queue to continuously identify the current training phase, thereby adaptively adjusting tolerance for stale updates. Experimental results demonstrate that FedPSA consistently outperforms baseline methods across multiple datasets, achieving performance gains of up to 6.37% and surpassing the current state-of-the-art by 1.93%.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Adversarial Agents
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements

Abstract

Asynchronous Federated Learning (AFL) has emerged as a significant research area in recent years. By not waiting for slower clients and executing the training process concurrently, it achieves faster training speed compared to traditional federated learning. However, due to the staleness introduced by the asynchronous process, its performance may degrade in some scenarios. Existing methods often use the round difference between the current model and the global model as the sole measure of staleness, which is coarse-grained and lacks observation of the model itself, thereby limiting the performance ceiling of asynchronous methods. In this paper, we propose FedPSA (Parameter Sensitivity-based Asynchronous Federated Learning), a more fine-grained AFL framework that leverages parameter sensitivity to measure model obsolescence and establishes a dynamic momentum queue to assess the current training phase in real time, thereby adjusting the tolerance for outdated information dynamically. Extensive experiments on multiple datasets and comparisons with various methods demonstrate the superior performance of FedPSA, achieving up to 6.37\% improvement over baseline methods and 1.93\% over the current state-of-the-art method.

Problem

• Asynchronous Federated Learning
• staleness
• model obsolescence
• parameter sensitivity
• behavioral staleness

Innovation

• asynchronous federated learning
• parameter sensitivity
• behavioral staleness
• dynamic momentum queue
• model obsolescence