A Teacher-Student Spatiotemporal Semi-Supervised Method for Urban Event Forecasting

ZHOU Zheng-yang; LIU Hao; WANG Kun; WANG Peng-kun; WANG Xu; WANG Yang

doi:10.12263/DZXB.20211579

您当前的位置：

首页 >

文章列表页 >

A Teacher-Student Spatiotemporal Semi-Supervised Method for Urban Event Forecasting

PAPERS | 更新时间：2025-12-11

- A Teacher-Student Spatiotemporal Semi-Supervised Method for Urban Event Forecasting
- ACTA ELECTRONICA SINICA Vol. 51, Issue 12, Pages: 3557-3571(2023)
- 作者机构：
  
  1.中国科学技术大学计算机科学与技术学院,安徽合肥 230031
  2.中国科学技术大学软件学院,安徽合肥 230041
  3.中国科学技术大学大数据学院,安徽合肥 230031
  4.中国科学技术大学苏州高等研究院,江苏苏州 215125
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(62072427);the Project of Stable Support for Youth Team in Basic Research Field, CAS(YSBR-005)
- DOI：10.12263/DZXB.20211579
  CLC： TP391;
- Received：24 November 2021，
  
  Revised：2022-12-02，
  
  Published：25 December 2023
- 稿件说明：
移动端阅览
周正阳,刘浩,王琨等.基于教师-学生时空半监督网络的城市事件预测方法[J].电子学报,2023,51(12):3557-3571.

ZHOU Zheng-yang,LIU Hao,WANG Kun,et al.A Teacher-Student Spatiotemporal Semi-Supervised Method for Urban Event Forecasting[J].ACTA ELECTRONICA SINICA,2023,51(12):3557-3571.
周正阳,刘浩,王琨等.基于教师-学生时空半监督网络的城市事件预测方法[J].电子学报,2023,51(12):3557-3571. DOI： 10.12263/DZXB.20211579.

ZHOU Zheng-yang,LIU Hao,WANG Kun,et al.A Teacher-Student Spatiotemporal Semi-Supervised Method for Urban Event Forecasting[J].ACTA ELECTRONICA SINICA,2023,51(12):3557-3571. DOI： 10.12263/DZXB.20211579.

摘要

离散时空事件预测是城市计算领域中的重要科学问题之一.现有工作主要聚焦于使用多样化的时空神经网络对城市动态特征与事件时空关联进行建模，且已经取得了一定成效，但仍然存在以下问题：首先，城市事件具有诱因多源和时空稀疏性，而这种时空稀疏性可能同时源于事件本身的稀少性和采集的不完整性，现有工作尚未能解决短期预测中的稀疏性挑战及零膨胀问题；其次，已发生事件倾向于继续向周边区域传播事件风险，但由于现有工作同质化了动态特征和事件之间的交互关联，因此其不能捕捉历史事件对未来事件风险带来的交互影响.鉴于此，为协同地利用事件标记信息和时空特征，本文提出基于教师-学生时空半监督学习框架以预测短期离散事件的时空分布.在教师网络中，为应对事件标记的稀疏性，本文在时空学习中引入半监督机制，提出基于自编码器的特征重建和时空方差异常描述引导的动态特征表示学习；在学生网络中，本文设计了特征-事件解耦的双管道学习机制，并提出时空衰减图卷积网络与长短期记忆网络来模拟事件在时空范围内发生的风险传播.此外，本文发展了时空多粒度预测机制，通过易学的粗粒度预测任务指导细粒度的高质量预测，最终实现粗-细粒度协同提名的离散时空事件预测.实验基于纽约和苏州工业园区数据集开展，本文模型能够在事件击中准确率上分别超越最好的基线模型5.46%和10.65%，充分验证了提出方法的有效性.

Abstract

Discrete spatiotemporal event forecasting is one of critical scientific problems in the field of urban computing. Existing works mostly focus on leveraging various spatiotemporal neural networks to model spatiotemporal correlations among dynamic urban features and events

and have achieved promising results. However

it still remains the following problems. First

urban events are naturally induced by multiple causes and distributed spatiotemporal sparsely

while such spatiotemporal sparsity can be induced by the inherent infrequent occurrence and its collection incompleteness. Given that

existing works cannot well address such sparsity challenge and zero-inflated issue in short-term forecasting. Second

the occurred events have the potential to raise future risks on neighboring regions. Unfortunately

off-the-shelf literatures tend to homogenize the correlations of feature-event and event-event

and fail to capture the wane-and-wax influences of historical event sequences on future events. Therefore

to cooperatively exploit the event labels and spatiotemporal features

this paper proposes a teacher-student spatiotemporal semi-supervised learning framework

addressing the challenge of short-term spatiotemporal event forecasting. In the teacher network

to tackle the sparsity challenge of event labels

this paper introduces the semi-supervised scheme into spatiotemporal learning where it designs an AutoEncoder-based feature reconstruction learning and spatiotemporal variance-based anomaly descriptor to facilitate feature representations. In the student network

this work designs a feature-event disentangled dual pipeline and proposes the spatiotemporal attenuation graph convolution network (GCN) and long-short term memory network (LSTM) to imitate the natural risk propagation along spatiotemporal domains. In addition

this paper also develops the spatiotemporal multi-granularity risk prediction task

which emphasizes the easy-to-learn coarse-grained prediction to guide the high-quality fine-grained forecasting

and finally realizes the high-risk discrete region nomination with coarse-to-fine learning mechanism. Experiments on NYC and SIP datasets illustrate that the proposed event forecasting framework outperforms the best baselines by respectively 5.46% and 10.65%

verifying the effectiveness of our work.

关键词

Keywords

references

ZHAO X Y , TANG J L . Exploring transfer learning for crime prediction [C ] // 2017 IEEE International Conference on Data Mining Workshops . Piscataway : IEEE , 2017 : 1158 - 1159 .

WANG B , LUO X Y , ZHANG F B , et al . Graph-based deep modeling and real time forecasting of sparse spatio-temporal data [EB/OL ] . ( 2018-04-02 )[ 2021-11-24 ] . https://arxiv.org/abs/1804.00684 https://arxiv.org/abs/1804.00684 .

NIKOLOPOULOS K , PUNIA S , SCHÄFERS A , et al . Forecasting and planning during a pandemic: COVID-19 growth rates, supply chain disruptions, and governmental decisions [J ] . European Journal of Operational Research , 2021 , 290 ( 1 ): 99 - 115 .

CHANG C Y , LU C W , WANG C J . A multi-step-ahead markov conditional forward model with cube perturbations for extreme weather forecasting [C ] // Proc of the 35th AAAI Conference on Artificial Intelligence . California : AAAI , 2021 : 6948 - 6955 .

ZHOU Z Y , WANG Y , XIE X K , et al . Foresee urban sparse traffic accidents: A spatiotemporal multi-granularity perspective [J ] . IEEE Transactions on Knowledge and Data Engineering , 2022 , 34 ( 8 ): 3786 - 3799 .

ZHOU Z Y , WANG Y , XIE X K , et al . RiskOracle: A minute-level citywide traffic accident forecasting framework [C ] // Proc of the 34th the Association for the Advancement of Artificial Intelligence Conference on Artificial Intelligence . California : AAAI , 2020 : 1258 - 1265 .

MOOSAVI S , SAMAVATIAN M H , PARTHASARATHY S , et al . Accident risk prediction based on heterogeneous sparse data: New dataset and insights [C ] // Proc of the 27th ACM Sigspatial International Conference on Advances in Geographic Information Systems . New York : ACM , 2019 : 33 - 42 .

CHEN L R , GUAN Q F , XIONG Y H , et al . A spatially constrained multi-autoencoder approach for multivariate geochemical anomaly recognition [J ] . Computers & Geosciences , 2019 , 125 : 43 - 54 .

黄志清 , 曲志伟 , 张吉 , 等 . 基于深度强化学习的端到端无人驾驶决策 [J ] . 电子学报 , 2020 , 48 ( 9 ): 1711 - 1719 .

HUANG Z Q , QU Z W , ZHANG J , et al . End-to-end autonomous driving decision based on deep reinforcement learning [J ] . Acta Electronica Sinica , 2020 , 48 ( 9 ): 1711 - 1719 . (in Chinese)

张啸剑 , 孟小峰 . 面向数据发布和分析的差分隐私保护 [J ] . 计算机学报 , 2014 , 37 ( 4 ): 927 - 949 .

ZHANG X J , MENG X F . Differential privacy in data publication and analysis . Chinese Journal of Computers , 2014 , 37 ( 4 ): 927 - 949 . (in Chinese)

YE Q Q , HU H B , AU M H , et al . LF-GDPR: A framework for estimating graph metrics with local differential privacy [J ] . IEEE Transactions on Knowledge and Data Engineering , 2022 , 34 ( 10 ): 4905 - 4920 .

ZHAO J W , GAO Y J , MA C Y , et al . On efficiently diversified top-k geo-social keyword query processing in road networks [J ] . Information Sciences , 2020 , 512 : 813 - 829 .

GNEITING T , RAFTERY A E . Weather forecasting with ensemble methods [J ] . Science , 2005 , 310 ( 5746 ): 248 - 249 .

HOLMSTROM M , LIU D , VO C . Machine learning applied to weather forecasting [J ] . Meteorol , 2016 , 4 : 1 - 5 .

冯宁 , 郭晟楠 , 宋超 , 等 . 面向交通流量预测的多组件时空图卷积网络 [J ] . 软件学报 , 2019 , 30 ( 3 ): 759 - 769 .

FENG N , GUO S N , SONG C , et al . Multi-component spatial-temporal graph convolution networks for traffic flow forecasting [J ] . Journal of Software , 2019 , 30 ( 3 ): 759 - 769 . (in Chinese)

杜圣东 , 李天瑞 , 杨燕 , 等 . 一种基于序列到序列时空注意力学习的交通流预测模型 [J ] . 计算机研究与发展 , 2020 , 57 ( 8 ): 1715 - 1728 .

DU S D , LI T R , YANG Y , et al . A sequence-to-sequence spatial-temporal attention learning model for urban traffic flow prediction [J ] . Research and Development of Computers , 2020 , 57 ( 8 ): 1715 - 1728 . (in Chinese)

丁治明 , 韩京宇 , 李曼 , 等 . 基于网络受限移动对象数据库的交通流统计分析模型 [J ] . 计算机研究与发展 , 2008 , 45 ( 4 ): 646 - 655 .

DING Z M , HAN J Y , LI M , et al . Network-constrained moving objects database based traffic flow statistical analysis model [J ] . Research and Development of Computers , 2008 , 45 ( 4 ): 646 - 655 . (in Chinese)

ZHANG J B , ZHENG Y , QI D K . Deep spatio-temporal residual networks for citywide crowd flows prediction [C ] // Proc of the 31st AAAI Conference on Artificial Intelligence . California : AAAI , 2017 : 1655 - 1661 .

HAO Z K , LU C Q , HUANG Z Y , et al . ASGN: An active semi-supervised graph neural network for molecular property prediction [C ] // Proc of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . New York : ACM , 2020 : 731 - 752 .

YU B , YIN H T , ZHU Z X . Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting [C ] // Proc of the 27th International Joint Conference on Artificial Intelligence . Melbourne : IJCAI , 2018 : 3634 - 3640 .

BAI L , YAO L N , KANHERE S S , et al . STG2seq: Spatial-temporal graph to sequence model for multi-step passenger demand forecasting [C ] // Proc of the 28th International Joint Conference on Artificial Intelligence . Melbourne : IJCAI , 2019 : 1981 - 1987 .

SONG C , LIN Y F , GUO S N , et al . Spatial-temporal synchronous graph convolutional networks: A new framework for spatial-temporal network data forecasting [C ] // Proc of the 34th AAAI Conference on Artificial Intelligence . California : AAAI , 2020 : 914 - 921 .

徐薇 , 黄厚宽 , 王英杰 . 一种基于数据融合和方法融合的时空综合预测算法 [J ] . 计算机研究与发展 , 2005 ( 7 ): 1255 - 1260 .

XU W , HUANG H K , WANG Y J . An integrated spatio-temporal forecasting approach based on data fusion and method fusion [J ] . Research and Development of Computers , 2005 ( 7 ): 1255 - 1260 . (in Chinese)

MA J , LI Z , JACK C P C , et al . Air quality prediction at new stations using spatially transferred bi-directional long short-term memory network [J ] . Science of The Total Environment , 2020 , 705 : 135771 .

YI X W , ZHANG J B , WANG Z Y , et al . Deep distributed fusion network for air quality prediction [C ] // Proc of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . New York : ACM , 2018 : 965 - 973 .

YUAN Z N , ZHOU X , YANG T B . Hetero-convlSTM: A deep learning approach to traffic accident prediction on heterogeneous spatio-temporal data [C ] // Proc of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . New York : ACM , 2018 : 984 - 992 .

FENG J , LIN Z Q , XIA T , et al . A sequential convolution network for population flow prediction with explicitly correlation modelling [C ] // Proc of the 29th International Joint Conference on Artificial Intelligence . Melbourne : IJCAI , 2020 : 1331 - 1337 .

SHI H Z , YAO Q M , GUO Q , et al . Predicting origin-destination flow via multi-perspective graph convolutional network [C ] // Proc of the 36th IEEE International Conference on Data Engineering . Piscataway : IEEE , 2020 : 1818 - 1821 .

XU F L , LI Y , XU S S . Attentional multi-graph convolutional network for regional economy prediction with open migration data [C ] // Proc of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . New York : ACM , 2020 : 2225 - 2233 .

ZHANG M Y , LI T , LI Y , et al . Multi-view joint graph representation learning for urban region embedding [C ] // Proc of the 29th International Joint Conference on Artificial Intelligence . Melbourne : IJCAI , 2020 : 4431 - 4437 .

FENG J , LI Y , ZHANG C , et al . Deepmove: Predicting human mobility with attentional recurrent networks [C ] // Proc of the 27th the Web Conference . New York : ACM , 2018 : 1459 - 1468 .

TENNESSEE M . Tennessee highway patrol provides predictive analytics software to all tennessee sheriffs' offices [EB/OL ] . ( 2017-10-10 )[ 2021-08-01 ] . https://www.tn.gov/safety/news/2017/10/10/ https://www.tn.gov/safety/news/2017/10/10/ .

CHANG L Y , CHEN W C . Data mining of tree-based models to analyze freeway accident frequency [J ] . Journal of Safety Research , 2005 , 36 ( 4 ): 365 - 375 .

HUANG C , ZHANG C X , DAI P , et al . Deep dynamic fusion network for traffic accident forecasting [C ] // Proc of the 28th ACM International Conference on Information and Knowledge Management . New York : ACM , 2019 : 2673 - 2681 .

REN H L , SONG Y , WANG J W , et al . A deep learning approach to the citywide traffic accident risk prediction [C ] // Proc of the 21st International Conference on Intelligent Transportation Systems . Piscataway : IEEE , 2018 : 3346 - 3351 .

CHEN C , FAN X L , ZHENG C P , et al . SDCAE: Stack denoising convolutional autoencoder model for accident risk prediction via traffic big data [C ] // Proc of the 6th International Conference on Advanced Cloud and Big Data (CBD) . Piscataway : IEEE , 2018 : 328 - 333 .

ZHOU Z Y . Attention based stack resnet for citywide traffic accident prediction [C ] // Proc of the 20th IEEE International Conference on Mobile Data Management . Piscataway : IEEE , 2019 : 369 - 370 .

BAO J , LIU P , UKKUSURI S V . A spatiotemporal deep learning approach for citywide short-term crash risk prediction with multi-source data [J ] . Accident Analysis & Prevention , 2019 , 122 : 239 - 254 .

HONG L , HONG L , ZHENG Y , et al . Detecting urban black holes based on human mobility data [C ] // Proc of the 23rd SIGSPATIAL International Conference on Advances in Geographic Information Systems . New York : ACM , 2015 : 1 - 10 .

YAO H X , TANG X F , WEI H , et al . Revisiting spatial-temporal similarity: A deep learning framework for traffic prediction [C ] // Proc of the 33rd AAAI Conference on Artificial Intelligence . California : AAAI , 2019 : 5668 - 5675 .

徐兴荣 , 刘聪 , 李婷 , 等 . 基于双向准循环神经网络和注意力机制的业务流程剩余时间预测方法 [J ] . 电子学报 , 2022 , 50 ( 8 ): 1975 - 1984 .

XU X R , LIU C , LI T , et al . Business process remaining time prediction: An approach based on bidirectional quasi recurrent neural network with attention [J ] . Acta Electronica Sinica , 2022 , 50 ( 8 ): 1975 - 1984 . (in Chinese)

许淼 , 刘宏飞 , 苏岳龙 . 考虑交通事件影响的城市道路行程时间预测 [J ] . 中国公路学报 , 2021 , 34 ( 12 ): 229 - 238 .

XU M , LIU H F , SU Y L . Urban road travel time prediction considering impact of traffic even [J ] . China Journal of Highway and Transport , 2021 , 34 ( 12 ): 229 - 238 . (in Chinese)

刘浩阳 , 林耀进 , 刘景华 , 等 . 由粗到细的分层特征选择 [J ] . 电子学报 , 2022 , 50 ( 11 ): 2778 - 2789 .

LIU H Y , LIN Y J , LIU J H , et al . Hierarchical feature selection from coarse to fine [J ] . Acta Electronica Sinica , 2022 , 50 ( 11 ): 2778 - 2789 . (in Chinese)

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

CS-ROMF:Efficient Community Search Based on Graph Combinatorial Optimization

RWK-GNN: Fraud Detection for Imbalanced Graphs with Feature Enhancement and Subkernel Decomposition

Anomaly Detection with Dual-Channel Heterogeneous Graph Neural Network Based on Hypersphere Dual Learning

An Automatic Construction Method of Indistinguishable Obfuscation for Generic Computing Circuits

Related Author

ZHOU Zhengyang

WANG Pengkun

LI Li-bo

ZHANG An-ran

WANG Xing-fen

ZHAO Yu-han

LIU Wei

YU Hao-miao

Related Institution

Hong Kong Baptist University, Hongkong

Beijing Information Science and Technology University

School of Software, Beihang University

School of Computer Science, Northwestern Polytechnical University

College of Computer Science, Chongqing University

⁰