Search for: All records

Federated learning (FL) is a collaborative machine-learning (ML) framework particularly suited for ML models requiring numerous training samples, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Random Forest, in the context of various applications, e.g., next-word prediction and eHealth. FL involves various clients participating in the training process by uploading their local models to an FL server in each global iteration. The server aggregates these models to update a global model. The traditional FL process may encounter bottlenecks, known as the straggler problem, where slower clients delay the overall training time. This paper introduces the Latency-awarE Semi-synchronous client Selection and mOdel aggregation for federated learNing (LESSON) method. LESSON allows clients to participate at different frequencies: faster clients contribute more frequently, therefore mitigating the straggler problem and expediting convergence. Moreover, LESSON provides a tunable trade-off between model accuracy and convergence rate by setting varying deadlines. Simulation results show that LESSON outperforms two baseline methods, namely FedAvg and FedCS, in terms of convergence speed and maintains higher model accuracy compared to FedCS.

 

Federated learning (FL) is a distributed machine learning technique to address the data privacy issue. Participant selection is critical to determine the latency of the training process in a heterogeneous FL architecture, where users with different hardware setups and wireless channel conditions communicate with their base station to participate in the FL training process. Many solutions have been designed to consider computational and uploading latency of different users to select suitable participants such that the straggler problem can be avoided. However, none of these solutions consider the waiting time of a participant, which refers to the latency of a participant waiting for the wireless channel to be available, and the waiting time could significantly affect the latency of the training process, especially when a huge number of participants are involved in the training process and share the wireless channel in the time-division duplexing manner to upload their local FL models. In this paper, we consider not only the computational and uploading latency but also the waiting time (which is estimated based on an M/G/1 queueing model) of a participant to select suitable participants. We formulate an optimization problem to maximize the number of selected participants, who can upload their local models before the deadline in a global iteration. The Latency awarE pARticipant selectioN (LEARN) algorithm is proposed to solve the problem and the performance of LEARN is validated via simulations.