Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FedBWO: Enhancing Communication Efficiency in Federated Learning

Vahideh Hayyolalam, Öznur Özkasap

TL;DR

Federated Learning on resource-constrained devices is hindered by high communication costs from exchanging full model weights. FedBWO integrates the Black Widow Optimization to select the best client score and update the global model by transmitting only small performance scores, while applying BWO to refine the winning client’s weights. The approach demonstrates improved global accuracy and substantial communication savings over FedAvg and several metaheuristic FL baselines on CIFAR-10, highlighting a practical path toward scalable, bandwidth-efficient FL. However, FedBWO incurs higher execution time, suggesting a trade-off between accuracy/communication efficiency and latency that warrants further optimization for real-time use cases.

Abstract

Federated Learning (FL) is a distributed Machine Learning (ML) setup, where a shared model is collaboratively trained by various clients using their local datasets while keeping the data private. Considering resource-constrained devices, FL clients often suffer from restricted transmission capacity. Aiming to enhance the system performance, the communication between clients and server needs to be diminished. Current FL strategies transmit a tremendous amount of data (model weights) within the FL process, which needs a high communication bandwidth. Considering resource constraints, increasing the number of clients and, consequently, the amount of data (model weights) can lead to a bottleneck. In this paper, we introduce the Federated Black Widow Optimization (FedBWO) technique to decrease the amount of transmitted data by transmitting only a performance score rather than the local model weights from clients. FedBWO employs the BWO algorithm to improve local model updates. The conducted experiments prove that FedBWO remarkably improves the performance of the global model and the communication efficiency of the overall system. According to the experimental outcomes, FedBWO enhances the global model accuracy by an average of 21% over FedAvg, and 12% over FedGWO. Furthermore, FedBWO dramatically decreases the communication cost compared to other methods.

FedBWO: Enhancing Communication Efficiency in Federated Learning

TL;DR

Federated Learning on resource-constrained devices is hindered by high communication costs from exchanging full model weights. FedBWO integrates the Black Widow Optimization to select the best client score and update the global model by transmitting only small performance scores, while applying BWO to refine the winning client’s weights. The approach demonstrates improved global accuracy and substantial communication savings over FedAvg and several metaheuristic FL baselines on CIFAR-10, highlighting a practical path toward scalable, bandwidth-efficient FL. However, FedBWO incurs higher execution time, suggesting a trade-off between accuracy/communication efficiency and latency that warrants further optimization for real-time use cases.

Abstract

Federated Learning (FL) is a distributed Machine Learning (ML) setup, where a shared model is collaboratively trained by various clients using their local datasets while keeping the data private. Considering resource-constrained devices, FL clients often suffer from restricted transmission capacity. Aiming to enhance the system performance, the communication between clients and server needs to be diminished. Current FL strategies transmit a tremendous amount of data (model weights) within the FL process, which needs a high communication bandwidth. Considering resource constraints, increasing the number of clients and, consequently, the amount of data (model weights) can lead to a bottleneck. In this paper, we introduce the Federated Black Widow Optimization (FedBWO) technique to decrease the amount of transmitted data by transmitting only a performance score rather than the local model weights from clients. FedBWO employs the BWO algorithm to improve local model updates. The conducted experiments prove that FedBWO remarkably improves the performance of the global model and the communication efficiency of the overall system. According to the experimental outcomes, FedBWO enhances the global model accuracy by an average of 21% over FedAvg, and 12% over FedGWO. Furthermore, FedBWO dramatically decreases the communication cost compared to other methods.
Paper Structure (16 sections, 4 equations, 7 figures, 3 algorithms)

This paper contains 16 sections, 4 equations, 7 figures, 3 algorithms.

Table of Contents

  1. Introduction
  2. Preliminaries and Related Work
  3. Federated Learning
  4. Black Widow Optimization
  5. Related Work
  6. FedBWO Methodology
  7. Motivation
  8. Proposed Method
  9. The Solution Representation
  10. Experimental Evaluation
  11. Experimental Setup
  12. Accuracy Evaluation
  13. Loss Evaluation
  14. Communication Cost Evaluation
  15. Execution Time Evaluation
  16. ...and 1 more sections

Figures (7)

  • Figure 1: General Overview of Federated Learning
  • Figure 2: The weighted aggregation method (like FedAvg) calculates the average of the weight value obtained from the K clients and then transmits the updated weight ($W_{t + 1}$) back to the client.
  • Figure 3: The procedure of updating the weights in the FedBWO algorithm (The server requests the client with the highest score to serve as the global model after collecting scores from all clients.
  • Figure 4: Accuracy comparison.
  • Figure 5: Loss comparison.
  • ...and 2 more figures