Scheduling and Communication Schemes for Decentralized Federated Learning
This work addresses connectivity problems in federated learning for networks with arbitrary topology, offering incremental improvements in scheduling for decentralized implementations.
The paper tackled the scalability and connectivity issues in federated learning by introducing a decentralized model with stochastic gradient descent and proposing three scheduling policies for client-server communications, resulting in improved convergence speed and final model performance as shown in experiments.
Federated learning (FL) is a distributed machine learning paradigm in which a large number of clients coordinate with a central server to learn a model without sharing their own training data. One central server is not enough, due to problems of connectivity with clients. In this paper, a decentralized federated learning (DFL) model with the stochastic gradient descent (SGD) algorithm has been introduced, as a more scalable approach to improve the learning performance in a network of agents with arbitrary topology. Three scheduling policies for DFL have been proposed for communications between the clients and the parallel servers, and the convergence, accuracy, and loss have been tested in a totally decentralized mplementation of SGD. The experimental results show that the proposed scheduling polices have an impact both on the speed of convergence and in the final global model.