Learning All Dynamics: Traffic Forecasting via Locality-Aware Spatio-Temporal Joint Transformer

doi:10.1109/TITS.2022.3197640

	Learning All Dynamics: Traffic Forecasting via Locality-Aware Spatio-Temporal Joint Transformer
	Fang, Yuchen 1; Zhao, Fang 1; Qin, Yanjun 1; Luo, Haiyong 2; Wang, Chenxing 1
	2022-08-16
发表期刊	IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
ISSN	1524-9050
页码	14
摘要	Forecasting traffic flow and speed in the urban is important for many applications, ranging from the intelligent navigation of map applications to congestion relief of city management systems. Therefore, mining the complex spatio-temporal correlations in the traffic data to accurately predict traffic is essential for the community. However, previous studies that combined the graph convolution network or self-attention mechanism with deep time series models (e.g., the recurrent neural network) can only capture spatial dependencies in each time slot and temporal dependencies in each sensor, ignoring the spatial and temporal correlations across different time slots and sensors. Besides, the state-of-the-art Transformer architecture used in previous methods is insensitive to local spatio-temporal contexts, which is hard to suit with traffic forecasting. To solve the above two issues, we propose a novel deep learning model for traffic forecasting, named Locality-aware spatio-temporal joint Transformer (Lastjormer), which elaborately designs a spatio-temporal joint attention in the Transformer architecture to capture all dynamic dependencies in the traffic data. Specifically, our model utilizes the dot-product self-attention on sensors across many time slots to extract correlations among them and introduces the linear and convolution self-attention mechanism to reduce the computation needs and incorporate local spatio-temporal information. Experiments on three real-world traffic datasets, England, METR-LA, and PEMS-BAY, demonstrate that our Lastjormer achieves state-of-the-art performances on a variety of challenging traffic forecasting benchmarks.
关键词	Forecasting Correlation Convolution Roads Transformers Predictive models Task analysis Traffic forecasting spatio-temporal joint transformer diffusion convolution network
DOI	10.1109/TITS.2022.3197640
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[61872046] ; Beijing Natural Science Foundation[4212024] ; Science and Technology Plan Project of Inner Mongolia Autonomous Region[2019GG328]
WOS研究方向	Engineering ; Transportation
WOS类目	Engineering, Civil ; Engineering, Electrical & Electronic ; Transportation Science & Technology
WOS记录号	WOS:000842745600001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：13[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/19451
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Zhao, Fang; Luo, Haiyong
作者单位	1.Beijing Univ Posts & Telecommun, Sch Comp Sci, Natl Pilot Software Engn Sch, Beijing 100876, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, Beijing Key Lab Mobile Comp & Pervas Device, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Fang, Yuchen,Zhao, Fang,Qin, Yanjun,et al. Learning All Dynamics: Traffic Forecasting via Locality-Aware Spatio-Temporal Joint Transformer[J]. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS,2022:14.
APA	Fang, Yuchen,Zhao, Fang,Qin, Yanjun,Luo, Haiyong,&Wang, Chenxing.(2022).Learning All Dynamics: Traffic Forecasting via Locality-Aware Spatio-Temporal Joint Transformer.IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS,14.
MLA	Fang, Yuchen,et al."Learning All Dynamics: Traffic Forecasting via Locality-Aware Spatio-Temporal Joint Transformer".IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS (2022):14.