FedBCD:Communication-Efficient Accelerated Block Coordinate Gradient Descent for Federated Learning
#FedBCD #communication-efficient #block coordinate gradient descent #federated learning #distributed optimization
📌 Key Takeaways
- FedBCD is a new federated learning algorithm designed to reduce communication costs.
- It uses accelerated block coordinate gradient descent to improve efficiency.
- The method aims to speed up convergence while minimizing data transmission.
- FedBCD addresses challenges in distributed machine learning environments.
📖 Full Retelling
arXiv:2603.05116v1 Announce Type: cross
Abstract: Although Federated Learning has been widely studied in recent years, there are still high overhead expenses in each communication round for large-scale models such as Vision Transformer. To lower the communication complexity, we propose a novel Federated Block Coordinate Gradient Descent (FedBCGD) method for communication efficiency. The proposed method splits model parameters into several blocks, including a shared block and enables uploading a
🏷️ Themes
Federated Learning, Optimization Algorithms
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Machine Learning arXiv:2603.05116 [Submitted on 5 Mar 2026] Title: FedBCD:Communication-Efficient Accelerated Block Coordinate Gradient Descent for Federated Learning Authors: Junkang Liu , Fanhua Shang , Yuanyuan Liu , Hongying Liu , Yuangang Li , YunXiang Gong View a PDF of the paper titled FedBCD:Communication-Efficient Accelerated Block Coordinate Gradient Descent for Federated Learning, by Junkang Liu and 5 other authors View PDF Abstract: Although Federated Learning has been widely studied in recent years, there are still high overhead expenses in each communication round for large-scale models such as Vision Transformer. To lower the communication complexity, we propose a novel Federated Block Coordinate Gradient Descent method for communication efficiency. The proposed method splits model parameters into several blocks, including a shared block and enables uploading a specific parameter block by each client, which can significantly reduce communication overhead. Moreover, we also develop an accelerated FedBCGD algorithm (called FedBCGD+) with client drift control and stochastic variance reduction. To the best of our knowledge, this paper is the first work on parameter block communication for training large-scale deep models. We also provide the convergence analysis for the proposed algorithms. Our theoretical results show that the communication complexities of our algorithms are a factor $1/N$ lower than those of existing methods, where $N$ is the number of parameter blocks, and they enjoy much faster convergence than their counterparts. Empirical results indicate the superiority of the proposed algorithms compared to state-of-the-art algorithms. The code is available at this https URL . Subjects: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI) Cite as: arXiv:2603.05116 [cs.LG] (or arXiv:2603.05116v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2603.05116 Focus to learn more arXiv-issued DOI via DataCite (pending...
Read full article at source