The invention discloses a traffic flow prediction method for personalized federal learning based on a prior attention mechanism, and the method comprises the steps: 1, a central server initializes and deploys a client model, and a client carries out the encoder model training and the attention mechanism calculation of time correlation, and uploads a traffic flow transmission state; 2, the central server receives the traffic flow transmission state, performs prior attention mechanism calculation of spatial correlation, and issues the processed traffic flow transmission state; 3, the client receives the processed traffic flow transmission state, performs decoder model training, and predicts a real label; 4, the client uploads the trained encoder parameters and decoder parameters to a central server, and the central server performs attention mechanism aggregation on the parameters and issues the parameters; and 5, the local server receives the aggregated parameters and brings the aggregated parameters into the model for traffic flow prediction. According to the invention, the accuracy of traffic flow prediction can be improved, and the data privacy of each client can be guaranteed.

    本发明公开了一种基于先验注意力机制的个性化联邦学习的交通流量预测方法,包括:1中央服务器初始化并部署客户端模型,客户端进行编码器模型训练以及时间相关性的注意力机制计算,并上传交通流量传递状态;2中央服务器接收流量传递状态,并进行空间相关性的先验注意力机制计算,并下发处理后的交通流量传递状态;3客户端接收处理后的交通流量传递状态并进行解码器模型训练,预测真实标签;4客户端将训练好的编码器参数和解码器参数上传至中央服务器,中央服务器对参数进行注意力机制的聚合并下发参数;5地方服务器接收聚合后的参数带入模型用于交通流量预测。本发明不仅能提高交通流量预测的准确性,还能保障每个客户端的数据隐私。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Personalized federated learning traffic flow prediction method based on prior attention mechanism


    Weitere Titelangaben:

    基于先验注意力机制的个性化联邦学习的交通流量预测方法


    Beteiligte:
    CHAI YIDONG (Autor:in) / HUANG AMEI (Autor:in) / LIU YI (Autor:in) / LIU LONGSHUN (Autor:in) / PAN YUE (Autor:in) / CHEN YUYING (Autor:in) / LIU HAOXIN (Autor:in) / ZHOU LUJUAN (Autor:in) / ZHOU ANQI (Autor:in) / YANG XINGGE (Autor:in)

    Erscheinungsdatum :

    2024-04-30


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G06Q Datenverarbeitungssysteme oder -verfahren, besonders angepasst an verwaltungstechnische, geschäftliche, finanzielle oder betriebswirtschaftliche Zwecke, sowie an geschäftsbezogene Überwachungs- oder Voraussagezwecke , DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Traffic flow prediction method based on attention mechanism

    SUN LIJUN / LIU MINGZHI / LIU GUANFENG et al. | Europäisches Patentamt | 2022

    Freier Zugriff

    FedGRU: Privacy-preserving Traffic Flow Prediction via Federated Learning

    Liu, Yi / Zhang, Shuyu / Zhang, Chenhan et al. | IEEE | 2020


    Traffic flow prediction method and system based on federated learning, and medium

    LU MINGMING / HE WENYONG | Europäisches Patentamt | 2023

    Freier Zugriff

    Driver Monitoring-Based Lane-Change Prediction: A Personalized Federated Learning Framework

    Du, Runjia / Han, Kyungtae / Gupta, Rohit et al. | IEEE | 2023


    Traffic flow prediction method based on space-time attention mechanism

    WANG JIAYING / YANG HENG / SHAN JING et al. | Europäisches Patentamt | 2024

    Freier Zugriff