Skip to content
Review Open access

Review on Aggregation Frameworks For High-Performance Federated Learning

Sep 2026 · Adolescência e Saúde · 0 citations · 15 references

TL;DR

This work establishes the research direction toward an adaptive weighted aggregation framework in which client contributions are dynamically adjusted using multiple performance-related indicators, expected to improve global accuracy, convergence stability, fairness, minority-class representation, and robustness to low-quality updates, while retaining the privacy-preserving characteristics of FL.

Abstract

Federated Learning (FL) enables collaborative model training across distributed clients while keeping their original data within local environments, thereby reducing the need for centralized data collection. Despite this advantage, the effectiveness of FL strongly depends on model aggregation, which determines how locally trained updates contribute to the global model. Conventional aggregation methods, particularly uniform or data-size-based averaging, may become ineffective when clients exhibit non-IID data distributions, class imbalance, unequal participation, device heterogeneity, client drift, variable update quality, and unreliable behaviour. This paper presents a systematic investigation of model aggregation strategies and organizes existing approaches into synchronous, asynchronous, hierarchical, adaptive, personalized, and robust aggregation frameworks. Their operating principles, advantages, limitations, and suitability for heterogeneous federated environments are comparatively examined. Particular emphasis is placed on aggregation mechanisms that account for local model performance, training loss, convergence behaviour, data contribution, statistical heterogeneity, and update quality when determining the influence of participating clients. The analysis reveals that fixed aggregation rules cannot adequately accommodate dynamically changing client and data characteristics. Accordingly, this work establishes the research direction toward an adaptive weighted aggregation framework in which client contributions are dynamically adjusted using multiple performance-related indicators. Such an approach is expected to improve global accuracy, convergence stability, fairness, minority-class representation, and robustness to low-quality updates, while retaining the privacy-preserving characteristics of FL. The study provides a foundation for developing reliable and scalable aggregation mechanisms for distributed applications including healthcare, intelligent transportation, smart cities, industrial systems, IoT, and next-generation communication networks.

Read PDF

Similar papers

Exploring aggregation strategies for federated learning in national statistics 1

A weighted Federated Averaging aggregation scheme which aims to highlight the potential benefits of positively weighing clients who contain more homogeneous data and the preliminary results exhibit that when clients with more homogeneous data are weighed higher, the global model can achieve better performance, but that...

B. Santos, Julian Templeton, Rafik Chemli Statistics Canada et al. · 0 citations
Oct 2026

FedGPA: a multi-level aggregation framework for federated learning with collaborative optimization of generalization and personalization

A novel framework is introduced, FedGPA, which integrates the global model, clustered models, individual optimal models, and individual local models through an aggregation mechanism leveraging the hierarchical structure of deep neural networks, and incorporates an adaptive weight matrix that dynamically adjusts feature...

Zi-Gang Chen, Hai-Chong Jiang, De-Yang Xie et al. · 0 citations
Open access Sep 2026

Adaptive Local Step Tuning for federated learning under data heterogeneity

Federated Learning (FL) enables collaborative model training across decentralized clients while preserving data privacy. However, non-IID data often induces client drift and unstable convergence due to mismatched local training dynamics across clients. Standard FL methods typically utilize fixed local training budgets,...

Loc Thanh Huynh · 0 citations
#machine learning Preprint Sep 2026

Similarity-Aware Personalized Federated Learning in Heterogeneous Environments

This work proposes SAPE-FL (Similarity-Aware Personalized Federated Learning), a novel personalization framework that anchors each client's model to both the global model and a similarity-weighted peer averaged model that mitigates negative transfer and enhances robustness in heterogeneous settings.

V. ArunKumarA, Sunil Gupta, N. Dang et al. · 0 citations
Open access Aug 2026

A Meta-Learning-Based Aggregation Strategy for Heterogeneous Federated Learning Scenarios

It is demonstrated that prediction-space meta-learning constitutes a practical alternative for federated aggregation without requiring shared parameter initialization, and among the different forms of heterogeneity in federated learning, this work focuses on statistical and model heterogeneity.

Sergio Pérez-Picazo, Hiram Galeana-Zapién, Edwin Aldana-Bobadilla · 0 citations
#federated learning Open access Nov 2026

FLEX: Flexible Federated Learning With Asynchronous Peer-to-Peer Communication and Adaptive Aggregation

In this paper, we introduce FLEX, a flexible federated learning (FL) scheme that combines peer-to-peer communication with asynchronous client updates, enabling clients to operate independently. We further propose a novel aggregation algorithm inspired by gradient-like updates, governed by two key parameters: (i) a simi...

C. Sad, D. Masouros, George Retsinas et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.