Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing the Convergence of Federated Learning Aggregation Strategies with Limited Data

Judith Sáinz-Pardo Díaz, Álvaro López García

TL;DR

The paper tackles privacy constraints in medical imaging by evaluating federated learning aggregation strategies for brain MRI classification with limited data, and by introducing FedAvgOpt as an optimized aggregation method. FedAvgOpt computes aggregated weights by solving a nonlinear problem to optimally weight client contributions, expressed as $w_{FedAvgOpt}^{r}= rac{1}{ extstyle extstyle extstyle extstyle extstyle extstyle extstyle} abla_i n_i w_i^{r} oldsymbol{\alpha}^{r}$ (with $oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol}}}}}}}}}}}$) solved via Nelder-Mead; when $oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{oldsymbol{ oldsymbol{ oldsymbol{ oldsymbol{ oldsymbol{ oldsymbol{ oldsymbol{ oldsymbol{ }}}}}}}}}}}}}}=1_n$, it reduces to FedAvg. Across four CNN backbones (VGG16, InceptionV3, ResNet-50 V2, DenseNet121) deployed in a 4-class brain MRI task with distributed data, FedAvgOpt consistently outperformed FedAvg, FedAvgM, FedMedian, FedOpt, and FedYogi, delivering faster convergence and higher accuracy. This demonstrates a practical, privacy-preserving approach for collaborative medical imaging model development across data centers. The work suggests substantial impact for multi-institution collaboration under stringent data governance.

Abstract

The development of deep learning techniques is a leading field applied to cases in which medical data is used, particularly in cases of image diagnosis. This type of data has privacy and legal restrictions that in many cases prevent it from being processed from central servers. However, in this area collaboration between different research centers, in order to create models as robust as possible, trained with the largest quantity and diversity of data available, is a critical point to be taken into account. In this sense, the application of privacy aware distributed architectures, such as federated learning arises. When applying this type of architecture, the server aggregates the different local models trained with the data of each data owner to build a global model. This point is critical and therefore it is fundamental to analyze different ways of aggregation according to the use case, taking into account the distribution of the clients, the characteristics of the model, etc. In this paper we propose a novel aggregation strategy and we apply it to a use case of cerebral magnetic resonance image classification. In this use case the aggregation function proposed manages to improve the convergence obtained over the rounds of the federated learning process in relation to different aggregation strategies classically implemented and applied.

Enhancing the Convergence of Federated Learning Aggregation Strategies with Limited Data

TL;DR

The paper tackles privacy constraints in medical imaging by evaluating federated learning aggregation strategies for brain MRI classification with limited data, and by introducing FedAvgOpt as an optimized aggregation method. FedAvgOpt computes aggregated weights by solving a nonlinear problem to optimally weight client contributions, expressed as (with ) solved via Nelder-Mead; when , it reduces to FedAvg. Across four CNN backbones (VGG16, InceptionV3, ResNet-50 V2, DenseNet121) deployed in a 4-class brain MRI task with distributed data, FedAvgOpt consistently outperformed FedAvg, FedAvgM, FedMedian, FedOpt, and FedYogi, delivering faster convergence and higher accuracy. This demonstrates a practical, privacy-preserving approach for collaborative medical imaging model development across data centers. The work suggests substantial impact for multi-institution collaboration under stringent data governance.

Abstract

The development of deep learning techniques is a leading field applied to cases in which medical data is used, particularly in cases of image diagnosis. This type of data has privacy and legal restrictions that in many cases prevent it from being processed from central servers. However, in this area collaboration between different research centers, in order to create models as robust as possible, trained with the largest quantity and diversity of data available, is a critical point to be taken into account. In this sense, the application of privacy aware distributed architectures, such as federated learning arises. When applying this type of architecture, the server aggregates the different local models trained with the data of each data owner to build a global model. This point is critical and therefore it is fundamental to analyze different ways of aggregation according to the use case, taking into account the distribution of the clients, the characteristics of the model, etc. In this paper we propose a novel aggregation strategy and we apply it to a use case of cerebral magnetic resonance image classification. In this use case the aggregation function proposed manages to improve the convergence obtained over the rounds of the federated learning process in relation to different aggregation strategies classically implemented and applied.

Paper Structure

This paper contains 14 sections, 4 equations, 4 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Federated Learning: aggregation strategies
  3. Federated Average (FedAvg)
  4. Federated Average with Momentum (FedAvgM)
  5. Federated Median (FedMedian)
  6. Adaptive Federated Optimization (FedOpt and FedYogi)
  7. State of the Art
  8. Federated Average Optimized: FedAvgOpt
  9. Methodology
  10. Data under study
  11. Models analyzed
  12. Federated Learning configuration
  13. Results
  14. Conclusions and future work

Figures (4)

  • Figure 1: Federated learning architecture applied to the current use case.
  • Figure 2: Distribution of the classes in the four clients.
  • Figure 3: Example of the available images for each of the four categories. Extracted from the data of the first client.
  • Figure 4: Evolution of the aggregated accuracy in the client's test set in each round of the FL architecture with each base model.