TY - GEN
T1 - A Reinforcement Learning-Based Client Selection Strategy for Efficient Federated Learning
AU - Jiang, Mingzhi
AU - Song, Ziqi
AU - Cui, Chen
AU - Wang, Shen
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.
PY - 2026
Y1 - 2026
N2 - With growing concerns over data privacy, federated learning (FL) enables collaborative model training without exposing raw data. However, FL faces persistent challenges in efficient client selection and effective resource utilization, especially under heterogeneous and Non-IID conditions. This paper presents a reinforcement learning-based client selection strategy leveraging a Dueling Double Deep Q-Network (DD-DQN) to enable adaptive and informed decision-making. Additionally, a novel aggregation mechanism is proposed, which assigns dynamic weights based on normalized Q-values, local training loss, and client data volume to improve aggregation accuracy. Experiments conducted on CIFAR-10, CIFAR-100, and MNIST under Non-IID partitions demonstrate that the proposed method improves global model accuracy by approximately 1.0% on average compared to FedAvg and FLASH-RL, while also achieving more stable convergence and higher system efficiency. These results validate the effectiveness of the proposed approach in enhancing federated learning performance under challenging data distributions.
AB - With growing concerns over data privacy, federated learning (FL) enables collaborative model training without exposing raw data. However, FL faces persistent challenges in efficient client selection and effective resource utilization, especially under heterogeneous and Non-IID conditions. This paper presents a reinforcement learning-based client selection strategy leveraging a Dueling Double Deep Q-Network (DD-DQN) to enable adaptive and informed decision-making. Additionally, a novel aggregation mechanism is proposed, which assigns dynamic weights based on normalized Q-values, local training loss, and client data volume to improve aggregation accuracy. Experiments conducted on CIFAR-10, CIFAR-100, and MNIST under Non-IID partitions demonstrate that the proposed method improves global model accuracy by approximately 1.0% on average compared to FedAvg and FLASH-RL, while also achieving more stable convergence and higher system efficiency. These results validate the effectiveness of the proposed approach in enhancing federated learning performance under challenging data distributions.
KW - Client Selection
KW - Federated Learning
KW - Non-IID Data
KW - Reinforcement Learning
UR - https://www.scopus.com/pages/publications/105028903626
U2 - 10.1007/978-3-032-13048-8_16
DO - 10.1007/978-3-032-13048-8_16
M3 - 会议稿件
AN - SCOPUS:105028903626
SN - 9783032130471
T3 - Smart Innovation, Systems and Technologies
SP - 202
EP - 215
BT - Advances in Intelligent Data Analysis and Applications - Proceedings of the 9th Euro-China Conference on Intelligent Data Analysis and Applications
A2 - Snášel, Václav
A2 - Kong, Lingping
A2 - Chu, Chu-Chuan
A2 - Pan, Jeng-Shyang
A2 - Esposito, Anna
PB - Springer Science and Business Media Deutschland GmbH
T2 - 9th Euro-China Conference on Intelligent Data Analysis and Applications, ECC 2025
Y2 - 21 July 2025 through 23 July 2025
ER -