Abstract
Predictive spatio-temporal analytics aims to analyze and model the data with both spatial and temporal attributes for future forecasting. Among various models proposed for predictive spatio-temporal analytics, the recurrent neural network (RNN) has been widely adopted. However, the training of RNN models becomes slow when the number of spatial locations is large. Moreover, the structure of RNN is unable to dynamically adapt to incorporate new covariates or to predict the target variables with varying spatial dimensions. In this paper, we propose a novel method, named Mesoscale Anisotropically-Connected Learning (MACL), to address the aforementioned limitations in RNN. For efficient training, we group the dataset into clusters (which refers to the mesoscale) along the spatial dimension according to the spatial adjacency and develop individual prediction module for each cluster. Then we design an anisotropic information exchange mechanism (i.e., the information exchange is not symmetric), to allow the prediction modules leveraging state information from nearby clusters for enhancing the prediction accuracy. Furthermore, for timely adaptation, we develop a local updating strategy for adapting the learning model to incorporate new covariates and the target variables with varying spatial dimensions. Experimental results on a real-world prediction task demonstrate that our method can be trained faster and more accurate than existing methods. Moreover, our method is flexible to incorporate new covariates and target variables of varying spatial dimensions, without sacrificing the prediction accuracy.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Available at https://github.com/lucktroy/DeepST/tree/master/data.
References
Belletti, F., Beutel, A., Jain, S., Chi, E.: Factorized recurrent neural architectures for longer range dependence. In: International Conference on Artificial Intelligence and Statistics, pp. 1522–1530 (2018)
Gao, F., Wu, L., Zhao, L., Qin, T., Cheng, X., Liu, T.Y.: Efficient sequence learning with group recurrent networks. In: Proceedings of the 2018 Conference of the North American Association for Computational Linguistics, vol. 1, pp. 799–808 (2018)
Gilmer, J., Schoenholz, S.S., Riley, P.F., Vinyals, O., Dahl, G.E.: Neural message passing for quantum chemistry. In: Proceedings of the 34th ICML, pp. 1263–1272 (2017)
Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997)
Hou, B.J., Zhang, L., Zhou, Z.H.: Learning with feature evolvable streams. In: Advances in Neural Information Processing Systems, pp. 1417–1427 (2017)
Kuchaiev, O., Ginsburg, B.: Factorization tricks for LSTM networks. ICLR Workshop (2017)
Lei, T., Zhang, Y., Wang, S.I., Dai, H., Artzi, Y.: Simple recurrent units for highly parallelizable recurrence. In: Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 4470–4481 (2018)
Li, Y., Yu, R., Shahabi, C., Liu, Y.: Diffusion convolutional recurrent neural network: data-driven traffic forecasting. In: NIPS 2017 Time Series Workshop (2017)
Lozano, A.C., et al.: Spatial-temporal causal modeling for climate change attribution. In: Proceedings of the 15th SIGKDD, pp. 587–596. ACM (2009)
Sahoo, D., Pham, Q., Lu, J., Hoi, S.C.: Online deep learning: learning deep neural networks on the fly. In: Proceedings of the 27th International Joint Conference on Artificial Intelligence, pp. 2660–2666. AAAI Press (2018)
Tobler, W.R.: A computer movie simulating urban growth in the Detroit region. Econ. Geogr. 46(sup1), 234–240 (1970)
Wu, X., Zhu, X., Wu, G.Q., Ding, W.: Data mining with big data. IEEE Trans. Knowl. Data Eng. 26(1), 97–107 (2014)
Yao, H., Liu, Y., Wei, Y., Tang, X., Li, Z.: Learning from multiple cities: a meta-learning approach for spatial-temporal prediction. In: The World Wide Web Conference, pp. 2181–2191. ACM (2019)
Yao, H., et al.: Deep multi-view spatial-temporal network for taxi demand prediction. In: Thirty-Second AAAI Conference on Artificial Intelligence (2018)
Yu, B., Yin, H., Zhu, Z.: Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875 (2017)
Zhang, J., Zheng, Y., et al.: Predicting citywide crowd flows using deep spatio-temporal residual networks. Artif. Intell. 259, 147–166 (2018)
Zhang, Q., Zhang, P., Long, G., Ding, W., Zhang, C., Wu, X.: Online learning from trapezoidal data streams. IEEE Trans. Knowl. Data Eng. 28(10), 2709–2723 (2016)
Acknowledgement
The authors would like to thank the reviewers for their valuable comments. This work was supported by the grants from the Research Grant Council of Hong Kong SAR under Project RGC/HKBU12201619 and RGC/HKBU12201318.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Tan, Q., Liu, Y., Liu, J. (2020). Mesoscale Anisotropically-Connected Learning. In: Helic, D., Leitner, G., Stettinger, M., Felfernig, A., Raś, Z.W. (eds) Foundations of Intelligent Systems. ISMIS 2020. Lecture Notes in Computer Science(), vol 12117. Springer, Cham. https://doi.org/10.1007/978-3-030-59491-6_16
Download citation
DOI: https://doi.org/10.1007/978-3-030-59491-6_16
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-59490-9
Online ISBN: 978-3-030-59491-6
eBook Packages: Computer ScienceComputer Science (R0)