基于区块链和联邦学习的运单准点预测算法  

A blockchain and federated learning based waybill on-time prediction algorithm

作　　者：叶进[1] 王柏棋 李晓欢 蒋祖平 YE Jin;WANG Baiqi;LI Xiaohuan;JIANG Zuping(School of Computer,Electronics and Information,Guangxi University,Nanning 530004,China;Guangxi Research Institute of Integrated Transportation BigData,Nanning 530025,China;Guangxi Guiwu Intelligent Technology Co.,Ltd.,Nanning 530299,China)

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004 [2]广西综合交通大数据研究院,广西南宁530025 [3]广西桂物智慧科技有限公司,广西南宁530299

出　　处：《广西大学学报（自然科学版）》2024年第4期847-853,共7页Journal of Guangxi University（Natural Science Edition）

基　　金：广西重点研发计划项目(桂科AB21196059)。

摘　　要：为了克服平台之间数据分布差异性带来的训练结果不理想,提出了一种带客户端漂移控制的动量加速联邦学习算法,利用采集的运单数据对改进算法进行验证。实验结果表明:在不同的数据分布情况下,改进的联邦学习算法相较传统的联邦平均算法(FedAvg)在性能方面得到了提高,其中收敛速度最高提升36%,F_(1)值最高提升5.7%。In order to overcome the unsatisfactory training results caused by differences in data distribution between platforms,a momentum-accelerated federated learning algorithm with client drift control was proposed.The improved algorithm was verified by the collected waybill data.Experiment results show that under different data distributions,the improved federated learning algorithm is improved in performance compared with the traditional federal average algorithm(FedAvg),among which the convergence speed is increased by up to 36%,and the F_(1) score is increased by up to 5.7%.

关 键 词：网络货运平台 准点预测 区块链 联邦学习 动量策略 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]