深度神经网络在城市交通预测中的应用  

Application of Deep Neural Network in Urban Traffic Prediction

作　　者：龙望晨[1] 王索 罗定福[1] 刘红[1] LONG Wang-chen;WANG Suo;LUO Ding-fu;LIU Hong(School of Computer and Information Engineering,Guangdong Songshan Polytechnic College,Shaoguan 512126,China)

机构地区：[1]广东松山职业技术学院计算机与信息工程学院,广东韶关512126

出　　处：《电脑知识与技术》2021年第16期183-185,192,共4页Computer Knowledge and Technology

摘　　要：深度神经网络能够有效地捕捉城市区域间的时间和空间相关性,同时对于特征工程依赖较少,逐渐成为交通预测中的应用热点。首先介绍交通预测的主要内容和挑战,从捕捉时空相关性的角度归纳近年来比较流行的基于深度神经网络的交通预测方法,分析各类方法的优缺点,最后对深度神经网络在交通预测中的未来发展前景进行展望。Deep neural network can effectively capture the temporal and spatial correlation between urban regions,and at the same time,it is less dependent on feature engineering,which gradually becomes the application hotspot in traffic prediction.Firstly,it introduces the main contents and challenges of traffic prediction,summarizes the popular traffic prediction methods based on deep neural network in recent years from the perspective of capturing temporal and spatial correlation,analyzes the advantages and disadvantages of various methods,and finally forecasts the future development prospect of deep neural network in traffic prediction.

关 键 词：机器学习 深度神经网络 交通预测 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]