测试技术学报2025，Vol.39Issue(2)：230-237,8.DOI:10.62756/csjs.1671-7449.2025028
分布式联邦学习结合联盟博弈在物联网中的应用
Application of Distributed Federated Learning Combined with Coalition Game in the Internet of Things
摘要
Abstract
As a new distributed machine learning paradigm,federated learning can effectively protect data privacy in Internet of Things(IoT)applications,but it still faces challenges such as low model update effi-ciency and poor real-time performance.To address these issues,a new model combining distributed feder-ated learning and coalition games is proposed,where clients collaborate in federated learning by weighing the benefits and costs of forming a coalition.Given the limited resources of devices in the IoT,a leader is selected within the coalition to coordinate the training process.To guide clients in adaptively forming coali-tions while ensuring the accuracy and efficiency of model updates,a distributed coalition formation algo-rithm was designed.Through continuous execution of coalition mergers and splits,the ultimate coalition partition is achieved to maximize the utility of cooperative devices.To ensure a fair distribution of coali-tion costs,a cost allocation mechanism is proposed to maintain the stability of the algorithm's results.Finally,experimental comparisons with other strategies validate the effectiveness of the proposed model.关键词
联邦学习/模型更新效率/联盟博弈/选择领导者/分布式联盟形成Key words
federated learning/model update efficiency/coalition game/leader selection/distributed coalition formation分类
计算机与自动化引用本文复制引用
祁中富,张志才..分布式联邦学习结合联盟博弈在物联网中的应用[J].测试技术学报,2025,39(2):230-237,8.基金项目
山西省基础研究计划自然科学研究面上资助项目(202103021224024) （202103021224024）
山西省基础研究计划青年科学研究资助项目(202103021223021) （202103021223021）
山西省重点研发计划资助项目(202202020101004) （202202020101004）
