Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DA-PFL: Dynamic Affinity Aggregation for Personalized Federated Learning

Xu Yang, Jiyuan Feng, Songyue Guo, Ye Wang, Ye Ding, Binxing Fang, Qing Liao

TL;DR

A novel Dynamic Affinity-based Personalized Federated Learning model (DA-PFL) is proposed to alleviate the class imbalanced problem during federated learning and build an affinity metric from a complementary perspective to guide which clients should be aggregated.

Abstract

Personalized federated learning becomes a hot research topic that can learn a personalized learning model for each client. Existing personalized federated learning models prefer to aggregate similar clients with similar data distribution to improve the performance of learning models. However, similaritybased personalized federated learning methods may exacerbate the class imbalanced problem. In this paper, we propose a novel Dynamic Affinity-based Personalized Federated Learning model (DA-PFL) to alleviate the class imbalanced problem during federated learning. Specifically, we build an affinity metric from a complementary perspective to guide which clients should be aggregated. Then we design a dynamic aggregation strategy to dynamically aggregate clients based on the affinity metric in each round to reduce the class imbalanced risk. Extensive experiments show that the proposed DA-PFL model can significantly improve the accuracy of each client in three real-world datasets with state-of-the-art comparison methods.

DA-PFL: Dynamic Affinity Aggregation for Personalized Federated Learning

TL;DR

A novel Dynamic Affinity-based Personalized Federated Learning model (DA-PFL) is proposed to alleviate the class imbalanced problem during federated learning and build an affinity metric from a complementary perspective to guide which clients should be aggregated.

Abstract

Personalized federated learning becomes a hot research topic that can learn a personalized learning model for each client. Existing personalized federated learning models prefer to aggregate similar clients with similar data distribution to improve the performance of learning models. However, similaritybased personalized federated learning methods may exacerbate the class imbalanced problem. In this paper, we propose a novel Dynamic Affinity-based Personalized Federated Learning model (DA-PFL) to alleviate the class imbalanced problem during federated learning. Specifically, we build an affinity metric from a complementary perspective to guide which clients should be aggregated. Then we design a dynamic aggregation strategy to dynamically aggregate clients based on the affinity metric in each round to reduce the class imbalanced risk. Extensive experiments show that the proposed DA-PFL model can significantly improve the accuracy of each client in three real-world datasets with state-of-the-art comparison methods.
Paper Structure (23 sections, 10 equations, 4 figures, 6 tables, 1 algorithm)

This paper contains 23 sections, 10 equations, 4 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Regularization-based PFL
  4. Architecture-based PFL
  5. Similarity-based PFL
  6. DA-PFL Algorithm
  7. Problem Definition
  8. Affinity Metric
  9. Dynamic Affinity-based Aggregation
  10. Experiments Setting
  11. Datasets
  12. Class Imbalance Settings
  13. Comparison Methods
  14. Implementation Details
  15. Results Analysis
  16. ...and 8 more sections

Figures (4)

  • Figure 1: The relationship between accuracy and weighting metrics, including the affinity of DA-PFL and the similarity of FedAMP.
  • Figure 2: Illustration of the DA-PFL. The workflow includes 6 steps: ① clients send the statistical information and local models; ②, ③ the server calculates overlapping-class vectors and affinity for each client; ④, ⑤ the server calculates dynamic aggregated weights based on affinity and local models and aggregates affinity-based aggregation models for each client by dynamic aggregated weights; ⑥ clients download the affinity-based aggregation models and update their local models.
  • Figure 3: Different imbalance distribution of class.
  • Figure 4: Accuracy of different communication rounds on test data set in model training.