Asynchronous Aggregation

Asynchronous aggregation in federated learning aims to improve training efficiency and scalability by allowing clients to update a shared model independently and at varying times, unlike the slower synchronous approach. Current research focuses on developing robust algorithms, such as buffered asynchronous methods and those incorporating incentive mechanisms or quantized communication, to mitigate the challenges of stale updates and client heterogeneity. These advancements are significant because they enable faster and more efficient training of machine learning models across distributed devices while addressing privacy concerns inherent in federated learning, leading to broader applicability in diverse settings.

Papers

June 5, 2024

Buffered Asynchronous Secure Aggregation for Cross-Device Federated Learning
Kun Wang, Yi-Rui Yang, Wu-Jun Li
Secure Aggregation Cross Device Federated Learning Asynchronous Aggregation

June 1, 2024

FedAST: Federated Asynchronous Simultaneous Training
Baris Askin, Pranay Sharma, Carlee Joe-Wong, Gauri Joshi
Asynchronous Federated Learning Simultaneous Learning Simultaneous Training Asynchronous Aggregation

May 11, 2024

Robust Model Aggregation for Heterogeneous Federated Learning: Analysis and Optimizations
Yumeng Shao, Jun Li, Long Shi, Kang Wei, Ming Ding, Qianmu Li, Zengxiang Li, Wen Chen, Shi Jin
General Analysis Federated Learning Heterogeneous Federated Learning Efficient Optimization Model Aggregation Asynchronous Aggregation

October 10, 2023

Asynchronous Federated Learning with Incentive Mechanism Based on Contract Theory
Danni Yang, Yun Ji, Zhoubin Kou, Xiaoxiong Zhong, Sheng Zhang
Asynchronous Federated Learning Incentive Mechanism Contract Theory Asynchronous Aggregation

August 1, 2023

Asynchronous Federated Learning with Bidirectional Quantized Communications and Buffered Aggregation
Tomas Ortega, Hamid Jafarkhani
Asynchronous Federated Learning Quantization Error Asynchronous Aggregation

May 6, 2023

Semi-Asynchronous Federated Edge Learning Mechanism via Over-the-air Computation
Zhoubin Kou, Yun Ji, Xiaoxiong Zhong, Sheng Zhang
Edge Learning Local SGD Asynchronous Aggregation

October 3, 2022

Unbounded Gradients in Federated Learning with Buffered Asynchronous Aggregation
Mohammad Taha Toghani, César A. Uribe
Potential Scalability Convergence Rate Secure Aggregation Unbounded Gradient Asynchronous Aggregation

June 21, 2022

A General Theory for Federated Optimization with Asynchronous and Heterogeneous Clients Updates
Yann Fraboni, Richard Vidal, Laetitia Kameni, Marco Lorenzi
Stochastic Gradient Descent FedAvg Algorithm Federated Optimization Heterogeneous Client Weight Aggregation FedAvg Converges Asynchronous Aggregation

Asynchronous Aggregation

Papers

Buffered Asynchronous Secure Aggregation for Cross-Device Federated Learning

FedAST: Federated Asynchronous Simultaneous Training

Robust Model Aggregation for Heterogeneous Federated Learning: Analysis and Optimizations

Asynchronous Federated Learning with Incentive Mechanism Based on Contract Theory

Asynchronous Federated Learning with Bidirectional Quantized Communications and Buffered Aggregation

Semi-Asynchronous Federated Edge Learning Mechanism via Over-the-air Computation

Unbounded Gradients in Federated Learning with Buffered Asynchronous Aggregation

A General Theory for Federated Optimization with Asynchronous and Heterogeneous Clients Updates