AutoSTG+: An automatic framework to discover the optimal network for spatio-temporal graph prediction

被引:9
|
作者
Ke, Songyu [1 ,2 ,3 ]
Pan, Zheyi [2 ,3 ]
He, Tianfu [2 ,3 ]
Liang, Yuxuan [2 ,3 ,5 ]
Zhang, Junbo [2 ,3 ,4 ]
Zheng, Yu [2 ,3 ,4 ]
机构
[1] Shanghai Jiao Tong Univ, Dept Comp Sci & Engn, Shanghai, Peoples R China
[2] JD Technol, JD iCity, Beijing, Peoples R China
[3] JD Intelligent Cities Res, Beijing, Peoples R China
[4] Southwest Jiaotong Univ, Sch Comp & Artificial Intelligence, Chengdu, Peoples R China
[5] Natl Univ Singapore, Singapore, Singapore
基金
中国国家自然科学基金; 国家重点研发计划;
关键词
Neural architecture search; Spatio-temporal graph; Meta; -learning; Urban computing; Spatio-temporal data mining;
D O I
10.1016/j.artint.2023.103899
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Spatio-temporal graphs (STGs) are important structures to describe urban sensory data, e.g., traffic speed and air quality. Predicting over spatio-temporal graphs enables many essential applications in intelligent cities, such as traffic management and environment analysis. Recently, many deep learning models have been proposed for spatio-temporal graph prediction and achieved significant results. However, manually designing neural networks requires rich domain knowledge and heavy expert efforts, making it impractical for real-world deployments. Therefore, we study automated neural architecture search for spatio-temporal graphs, which meets three challenges: 1) how to define search space for capturing complex spatio-temporal correlations; 2) how to jointly model the explicit and implicit relationships between nodes of an STG; and 3) how to learn network weight parameters related to meta graphs of STGs. To tackle these challenges, we propose a novel neural architecture search framework, entitled AutoSTG(+), for automated spatio-temporal graph prediction. In our AutoSTG(+), spatial graph convolution and temporal convolution operations are adopted in the search space of AutoSTG(+) to capture complex spatio-temporal correlations. Besides, we propose to employ the meta-learning technique to learn the adjacency matrices of spatial graph convolution layers and kernels of temporal convolution layers from the meta knowledge of meta graphs. And specifically, such meta-knowledge is learned by graph meta-knowledge learners, which iteratively aggregate knowledge on the attributed graphs and the similarity graphs. Finally, extensive experiments have been conducted on multiple real-world datasets to demonstrate that AutoSTG(+) can find effective network architectures and achieve up to about 20% relative improvements compared to human-designed networks (c) 2023 Elsevier B.V. All rights reserved.
引用
收藏
页数:19
相关论文
共 50 条
  • [21] Meteorology -Assisted Spatio-Temporal Graph Network for Uncivilized Urban Fvent Prediction
    Luo, Yang
    Gu, Zehao
    Zhou, Shiyang
    Xiong, Yun
    Gao, Xiaofeng
    23RD IEEE INTERNATIONAL CONFERENCE ON DATA MINING, ICDM 2023, 2023, : 468 - 477
  • [22] Robust Traffic Prediction Using Probabilistic Spatio-Temporal Graph Convolutional Network
    Karim, Atkia Akila
    Nower, Naushin
    ENGINEERING APPLICATIONS OF NEURAL NETWORKS, EANN 2024, 2024, 2141 : 259 - 273
  • [23] A Spatio-Temporal Graph Transformer Network for Multi-Pedestrain Trajectory Prediction
    Zhu, Jingfei
    Lian, Zhichao
    Jiang, Zhukai
    2022 IEEE INTL CONF ON DEPENDABLE, AUTONOMIC AND SECURE COMPUTING, INTL CONF ON PERVASIVE INTELLIGENCE AND COMPUTING, INTL CONF ON CLOUD AND BIG DATA COMPUTING, INTL CONF ON CYBER SCIENCE AND TECHNOLOGY CONGRESS (DASC/PICOM/CBDCOM/CYBERSCITECH), 2022, : 909 - 913
  • [24] A novel dynamic spatio-temporal graph convolutional network for windspeed interval prediction
    Chen, Zhengganzhe
    Zhang, Bin
    Du, Chenglong
    Meng, Wei
    Meng, Anbo
    ENERGY, 2024, 294
  • [25] STAGP: Spatio-Temporal Adaptive Graph Pooling Network for Pedestrian Trajectory Prediction
    Liu, Zhening
    He, Li
    Yuan, Liang
    Lv, Kai
    Zhong, Runhao
    Chen, Yaohua
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2024, 9 (03) : 2001 - 2007
  • [26] Multivariate spatio-temporal modeling of drought prediction using graph neural network
    Yu, Jiaxin
    Ma, Tinghuai
    Jia, Li
    Rong, Huan
    Su, Yuming
    Wahab, Mohamed Magdy Abdel
    JOURNAL OF HYDROINFORMATICS, 2024, 26 (01) : 107 - 124
  • [27] Spatio-Temporal Attention Graph Neural Network for Remaining Useful Life Prediction
    Huang, Zhixin
    He, Yujiang
    Sick, Bernhard
    2023 INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND COMPUTATIONAL INTELLIGENCE, CSCI 2023, 2023, : 99 - 105
  • [28] Bayesian Spatio-Temporal Graph Convolutional Network for Railway Train Delay Prediction
    Li, Jianmin
    Xu, Xinyue
    Ding, Xin
    Liu, Jun
    Ran, Bin
    IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, 2024, 25 (07) : 8193 - 8208
  • [29] Attribute prediction of spatio-temporal graph nodes based on weighted graph diffusion convolution network
    Ding, Linlin
    Yu, Haiyou
    Zhu, Chenli
    Ma, Ji
    Zhao, Yue
    WORLD WIDE WEB-INTERNET AND WEB INFORMATION SYSTEMS, 2023, 26 (05): : 3655 - 3690
  • [30] Attribute prediction of spatio-temporal graph nodes based on weighted graph diffusion convolution network
    Linlin Ding
    Haiyou Yu
    Chenli Zhu
    Ji Ma
    Yue Zhao
    World Wide Web, 2023, 26 : 3655 - 3690