Revealing the influence of participant failures on model quality in cross-silo Federated Learning

2026-03-26

AI Summary

This study addresses the lack of systematic investigation into how participant failures impact model quality in cross-silo federated learning. Through large-scale controlled experiments, it comprehensively examines the effects of missing nodes on the effectiveness, stability, and reproducibility of global models across image, tabular, and time-series data under non-IID settings, diverse model architectures, and varying participant availability patterns. The findings reveal that data skew is a dominant factor, often leading to overly optimistic performance estimates and significantly distorting the apparent influence of other variables. Moreover, the work uncovers distinct patterns in model robustness and generalization under different failure scenarios, offering theoretical foundations for designing more reliable federated learning systems.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

Federated Learning (FL) is a paradigm for training machine learning (ML) models in collaborative settings while preserving participants' privacy by keeping raw data local. A key requirement for the use of FL in production is reliability, as insufficient reliability can compromise the validity, stability, and reproducibility of learning outcomes. FL inherently operates as a distributed system and is therefore susceptible to crash failures, network partitioning, and other fault scenarios. Despite this, the impact of such failures on FL outcomes has not yet been studied systematically. In this paper, we address this gap by investigating the impact of missing participants in FL. To this end, we conduct extensive experiments on image, tabular, and time-series data and analyze how the absence of participants affects model performance, taking into account influencing factors such as data skewness, different availability patterns, and model architectures. Furthermore, we examine scenario-specific aspects, including the utility of the global model for missing participants. Our experiments provide detailed insights into the effects of various influencing factors. In particular, we show that data skewness has a strong impact, often leading to overly optimistic model evaluations and, in some cases, even altering the effects of other influencing factors.

Problem

• Federated Learning
• participant failures
• model quality
• cross-silo
• reliability

Innovation

• participant failure
• cross-silo federated learning
• data skewness
• model reliability
• missing participants