计算机科学 ›› 2025, Vol. 52 ›› Issue (10): 70-78.doi: 10.11896/jsjkx.241000088
刘钰婷, 顾晶晶, 周强
LIU Yuting, GU Jingjing, ZHOU Qiang
摘要： 城市出行流量预测是智慧城市研究中的重要课题,为城市规划和资源优化提供了关键的数据支持。近年来,基于图神经网络的城市流量预测模型在提升预测精度上取得了显著进展。然而,大多数现有研究都假设训练数据和测试数据来自相同的分布,忽视了现实世界中城市流量分布动态变化的复杂性,导致模型在面对分布偏移时表现不佳。为了解决这一问题,提出一种基于结构因果模型的城市出行流量预测方法,旨在应对分布偏移带来的模型泛化挑战。该方法首先利用结构因果模型揭示环境因素作为混淆变量对流量预测的影响效应,并设计共享分布估计器以学习环境信息的先验分布,进而引入后门调整方法,结合变分推断有效消除环境因素引起的混淆影响。该方法能够公平地考虑不同环境信息,提升流量预测的准确性与鲁棒性。在两个真实世界数据集上的实验结果表明,所提方法在应对分布偏移时具有较高的预测精度和鲁棒性。与6种主流基线模型相比,预测性能提升了2.26%~9.18%。
中图分类号:
| [1]GUO S,LIN Y,GONG L,et al.Self-supervised spatial-temporal bottleneck attentive network for efficient long-term traffic forecasting[C]//2023 IEEE 39th International Conference on Data Engineering (ICDE).IEEE,2023:1585-1596. [2]HAN J,ZHANG W,LIU H,et al.BigST:Linear ComplexitySpatio-Temporal Graph Neural Network for Traffic Forecasting on Large-Scale Road Networks[C]//Proceedings of the VLDB Endowment.2024:1081-1090. [3]PENG H,DU B,LIU M,et al.Dynamic graph convolutional network for long-term traffic flow prediction with reinforcement learning[J].Information Sciences,2021,578:401-416. [4]ZHENG C,FAN X,WANG C,et al.Gman:A graph multi-at-tention network for traffic prediction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:1234-1241. [5]ZHENG Z,GU J,ZHOU Q,et al.Prediction in Long-term Evolution:Exploiting the Interaction Between Urban Crowd Flow Variation and POI Transition Patterns[C]//2023 IEEE International Conference on Data Mining (ICDM).IEEE,2023:1559-1564. [6]GU J,ZHOU Q,YANG J,et al.Exploiting interpretablepat-terns for flow prediction in dockless bike sharing systems[J].IEEE Transactions on Knowledge and Data Engineering,2020,34(2):640-652. [7]YU B,YIN H,ZHU Z.Spatio-temporal graph convolutional networks:a deep learning framework for traffic forecasting[C]//Proceedings of the 27th International Joint Conference on Artificial Intelligence.2018:3634-3640. [8]SONG C,LIN Y,GUO S,et al.Spatial-temporal synchronous graph convolutional networks:A new framework for spatial-temporal network data forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:914-921. [9]WU Z,PAN S,LONG G,et al.Graph wavenet for deep spatial-temporal graph modeling[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence.2019:1907-1913. [10]BAI L,YAO L,LI C,et al.Adaptive graph convolutional recurrent network for traffic forecasting[J].Advances in Neural Information Processing Systems,2020,33:17804-17815. [11]GUO S,LIN Y,FENG N,et al.Attention based spatial-temporal graph convolutional networks for traffic flow forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2019:922-929. [12]MAGLIACANE S,VAN OMMEN T,CLAASSEN T,et al.Domain adaptation by using causal inference to predict invariant conditional distributions[C]//Advances in Neural Information Processing Systems.2018. [13]ZHANG S,YAO D,ZHAO Z,et al.Causerec:Counterfactualuser sequence synthesis for sequential recommendation[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.2021:367-377. [14]ROBERTSM E,STEWART B M,NIELSEN R A.Adjusting for confounding with text matching[J].American Journal of Political Science,2020,64(4):887-903. [15]ZHANG D,ZHANG H,TANG J,et al.Causal intervention for weakly-supervised semantic segmentation[J].Advances in Neural Information Processing Systems,2020,33:655-666. [16]NIU Y,TANG K,ZHANG H,et al.Counterfactual vqa:Acause-effect look at language bias[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2021:12700-12710. [17]GE C,SONG S,HUANG G.Causal intervention for human trajectory prediction with cross attention mechanism[C]//Procee-dings of the AAAI Conference on Artificial Intelligence.2023:658-666. [18]PEARL J.Models,reasoning and inference[M].Cambridge:Cambridge University Press,2000,19(2):3. [19]PEARL J.Causality[M].Cambridge University Press,2009. [20]BLEID M,KUCUKELBIR A,MCAULIFFE J D.Variational inference:A review for statisticians[J].Journal of the American Statistical Association,2017,112(518):859-877. [21]DIAOM Z,BALASUBRAMANIAN K,CHEWI S,et al.For-ward-backward Gaussian variational inference via JKO in the Bures-Wasserstein space[C]//International Conference on Machine Learning.PMLR,2023:7960-7991. [22]RUDNERT G J,CHEN Z,TEH Y W,et al.Tractable function-space variational inference in bayesian neural networks[J].Advances in Neural Information Processing Systems,2022,35:22686-22698. [23]BURDA Y,GROSSE R,SALAKHUTDINOV R.Importanceweighted autoencoders[J].arXiv:1509.00519,2015. [24]HOFFMANM D,JOHNSON M J.Elbo surgery:yet anotherway to carve up the variational evidence lower bound[C]//Workshop in Advances in Approximate Bayesian Inference,NIPS.2016. [25]YANG C,WU Q,WEN Q,et al.Towards out-of-distribution sequential event prediction:A causal treatment[J].Advances in neural information processing systems,2022,35:22656-22670. [26]ZHOU Z,HUANG Q,YANG K,et al.Maintaining the status quo:Capturing invariant relations for ood spatiotemporal learning[C]//Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.2023:3603-3614. |
|
||
首页