Abstract
Multivariate time series classification (MVTSC) has significant potential for Internet of Things applications. Recently, deep learning (DL) and graph neural network (GNN) methods have been applied to MVTSC tasks. Unfortunately, DL-based methods ignore explicit inter-series correlation modeling. Most existing GNN-based methods treat MVTS data as a static graph spanning the entire temporal trajectory, which inadequately captures changes in inter-series local correlations. To address this problem, we propose the spatial-temporal attention dynamic GNN (STADGNN), which explicitly models dynamic inter-series correlations by constructing the MVTS data into a dynamic graph structure at a finer granularity. It combines discrete Fourier transform (DFT) and discrete wavelet transform (DWT), which extract the global and local features of MVTS data in an end-to-end framework. In dynamic graph learning, spatial-temporal attention mechanisms are employed to simultaneously capture changes in inter-series local correlations and intra-series temporal dependencies without relying on predefined priors. Experimental results on 25 UEA datasets indicate that the STADGNN outperforms existing DL-based and GNN-based baseline models in MVTSC tasks.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Data availability
Data openly available in a public repository. https://www.timeseriesclassification.com
Code availability
The available source code has been provided in the article.
Materials availability
Not applicable.
References
Chandel A, Naruka MS (2023) Performance enhancement of smart grid integration using a novel intellectual multi-objective control technique. Soft Comput 27:12419–12437
Doddaiah R, Parvatharaju PS, Rundensteiner E, Hartvigsen T (2024) Explaining deep multi-class time series classifiers. Knowl Inf Syst
Li H, Yang S, Song Y, Luo Y, Li J, Zhou T (2022) Spatial dynamic graph convolutional network for traffic flow forecasting. Appl Intell 53:14986–14998
Wu Y, Lian C, Zeng Z, Xu B, Su Y (2023) An aggregated convolutional transformer based on slices and channels for multivariate time series classification. IEEE Trans Emerg Top Comput Intell 7:768–779
Zhou Q, Wang Z, Huang L (2023) Elm-kl-lstm: a robust and general incremental learning method for efficient classification of time series data. PeerJ Comput Sci 9
Mei J, Liu M, Yf Wang, Gao H (2016) Learning a mahalanobis distance-based dynamic time warping measure for multivariate time series classification. IEEE Trans Cyberne 46:1363–1374
Seto S, Zhang W, Zhou Y (2015) Multivariate time series classification using dynamic time warping template selection for human activity recognition. IEEE Symp Ser Comput Intell 1399–1406
Zhang HB, Wang PL, Zhang MM, Wang W (2023) Shapelet based two-step time series positive and unlabeled learning. J Comput Sci Technol 38:1387–1402
Chen J, Wan Y (2023) Localized shapelets selection for interpretable time series classification. Applied Intelligence 53:17985–18001
Cai B, Huang GL, Yang S, Xiang Y, Chi CH (2023) Se-shapelets: Semi-supervised clustering of time series using representative shapelets. Expert Syst Appl 240:122584
Dempster A, Schmidt DF, Webb GI (2022) Hydra: competing convolutional kernels for fast and accurate time series classification. Data Min Knowl Disc 37:1779–1805
Wang T, Liu Z, Zhang T, Hussain SF, Waqas MM, Li Y (2022) Adaptive feature fusion for time series classification. Knowl Based Syst 243:108459
Yu Y, Si X, Hu C, Jx Zhang (2019) A review of recurrent neural networks: Lstm cells and network architectures. Neural Comput 31:1235–1270
Zhang X, Zhong C, Zhang J, Wang T, Ng WWY (2023) Robust recurrent neural networks for time series forecasting. Neurocomputing 526:143–157
Ahmed S, Nielsen IE, Tripathi A, Siddiqui S, Ramachandran RP, Rasool G (2022) Transformers in time-series analysis: A tutorial. Circuits, Systems, and Signal Processing 42:7433–7466
Khan A, Rauf Z, Sohail A, Rehman A, Asif H, Asif A, Farooq U (2023) A survey of the vision transformers and their cnn-transformer based variants. Artif Intell Rev 56:2917–2970
Zhu Y, Luo S, Huang D, Zheng W, Su F, Hou B (2023) Drcnn: decomposing residual convolutional neural networks for time series forecasting. Sci Rep 13
Yang Y, Lu J (2022) Foreformer: an enhanced transformer-based framework for multivariate time series forecasting. Appl Intell 53:12521–12540
Xiao Z, Xu X, Zhang H, Szczerbicki E (2021) A new multi-process collaborative architecture for time series classification. Knowl Based Syst 220:106934
Guo S, Lin Y, Wan H, Li X, Cong G (2021) Learning dynamics and heterogeneity of spatial-temporal graph data for traffic forecasting. IEEE Trans Knowl Data Eng 34:5415–5428
Huang B, Dou H, Luo Y, Li J, Wang J, Zhou T (2023) Adaptive spatiotemporal transformer graph network for traffic flow forecasting by iot loop detectors. IEEE Int Things J 10:1642–1653
Jin G, Li F, Zhang J, Wang MF, Huang J (2022) Automated dilated spatio-temporal synchronous graph modeling for traffic prediction. IEEE Trans Intell Trans Syst 24:8820–8830
Jin M, Zheng Y, Li Y, Chen S, Yang B, Pan S (2022) Multivariate time series forecasting with dynamic graph neural odes. IEEE Trans Knowl Data Eng 35:9168–9180
Cini A, Zambon D, Alippi C (2022) Sparse graph learning from spatiotemporal time series. J Mach Learn Res 24:242–124236
Chen L, Chen D, Shang Z, Wu B, Zheng C, Wen B, Zhang W (2022) Multi-scale adaptive graph neural network for multivariate time series forecasting. IEEE Trans Knowl Data Eng 35:10748–10761
Zheng Y, Koh HY, Jin M, Chi L, Phan KT, Pan S, Chen YPP, Xiang W (2023) Correlation-aware spatial-temporal graph learning for multivariate time-series anomaly detection. IEEE Trans Neural Netw Learn Syst PP
Zhang K, Wen Q, Zhang C, Cai R, Jin M, Liu Y, Zhang J, Liang Y, Pang G, Song D, Pan S (2023) Self-supervised learning for time series analysis: Taxonomy, progress, and prospects. IEEE Trans Patt Anal Mach Intell PP
Lyu X, Chen Y, Chen J, Liu X, Gao J (2022) Spatio-temporal based architecture topology search for multivariate time series prediction. IEEE International Conference on Big Data (Big Data) 2022:1304–1309
Li G, Choi B, Xu J, Bhowmick SS, Chun KP, Wong GLH (2021) Efficient shapelet discovery for time series classification (extended abstract). 2021 IEEE 37th International Conference on Data Engineering (ICDE), 2336–2337
Yang J, Jing S, Huang G (2022) Accurate and fast time series classification based on compressed random shapelet forest. Appl Intell 53:5240–5258
Ji C, Zhao C, Liu S, Yang C, Pan L, Wu L, Meng X (2019) A fast shapelet selection algorithm for time series classification. Comput. Netw 148:231–240
Yuan J, Shi M, Wang Z, Liu H, Li J (2022) Random pairwise shapelets forest: an effective classifier for time series. Knowl Inf Syst 64:143–174
Li F, Feng J, Yan H, Jin G, Jin D, Li Y (2021) Dynamic graph convolutional recurrent network for traffic prediction: Benchmark and solution. ACM Trans Knowl Discov Data 17:1–21
Ruiz AP, Flynn M, Large J, Middlehurst M, Bagnall A (2020) The great multivariate time series classification bake off: a review and experimental evaluation of recent algorithmic advances. J Data Min Knowl Discov 35:401–449
Zheng Z, Zhou M, Chen Y, Huo M, Sun L, Zhao S, Chen D (2022) A fused method of machine learning and dynamic time warping for road anomalies detection. IEEE Trans Intell Transp Syst 23:827–839
Belkhouja T, Yan Y, Doppa JR (2022) Dynamic time warping based adversarial framework for time-series domain. IEEE Trans Patt Anal Mach Intell 45:7353–7366
Teng Y, Wang G, He C, Wu Y, Li C (2023) Optimization of dynamic time warping algorithm for abnormal signal detection. Int J Data Sci Anal 1–13
Peng B, Ding Y, Xia Q, Yang Y (2023) Recurrent neural networks integrate multiple graph operators for spatial time series prediction. Appl Intell 53:26067–26078
Jordan I, Sokól PA, Park IM (2019) Gated recurrent units viewed through the lens of continuous time dynamical systems. Front Comput Neurosci 15
Liu X, Luo Y, Ye Y, Lu J (2017) Mc-dcnn: Dilated convolutional neural network for computing stereo matching cost. In: International Conference on Neural Information Processing. https://api.semanticscholar.org/CorpusID:28982234
Liu J, Liu Y, Li D, Wang H, Huang X, Song L (2023) Dsdcla: driving style detection via hybrid cnn-lstm with multi-level attention fusion. Appl Intell 1–18
Liu F, Zhou X, Cao J, Wang Z, Wang T, Wang H, Zhang Y (2022) Anomaly detection in quasi-periodic time series based on automatic data segmentation and attentional lstm-cnn. IEEE Trans Knowl Data Eng 34:2626–2640
Tang W, Long G, Liu L, Zhou T, Blumenstein M, Jiang J (2020) Omni-scale cnns: a simple and effective kernel size configuration for time series classification. In: International Conference on Learning Representations. https://api.semanticscholar.org/CorpusID:249848305
Khan SH, Naseer M, Hayat M, Zamir SW, Khan FS, Shah M (2021) Transformers in vision: A survey. ACM Comput Surv (CSUR) 54:1–41
Wang W, Yao L, Chen L, Lin B, Cai D, He X, Liu W (2021) Crossformer: A versatile vision transformer hinging on cross-scale attention. In: International Conference on Learning Representations. https://api.semanticscholar.org/CorpusID:238531695
Gupta A, Maurya MK, Goyal N, Chaurasiya VK (2023) Istgcn: Integrated spatio-temporal modeling for traffic prediction using traffic graph convolution network. Appl Intell 53:29153–29168
Wu H, Xu J, Wang J, Long M (2021) Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting. In: Neural Information Processing Systems. https://api.semanticscholar.org/CorpusID:235623791
Wang Y, Zhu J, Kang R (2023) Destformer: A transformer based on explicit seasonal–trend decomposition for long-term series forecasting. Appl Sci
Wu Z, Pan S, Long G, Jiang J, Chang X, Zhang C (2020) Connecting the dots: Multivariate time series forecasting with graph neural networks. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining
Spadon G, Qiao Brandoli B, Matwin S, Rodrigues-Jr JF, Sun J (2020) Pay attention to evolution: Time series forecasting with deep graph-evolution learning. IEEE Trans Patt Anal Mach Intell 44:5368–5384
Wang X, Wang Y, Peng J, Zhang Z (2023) Multivariate long sequence time-series forecasting using dynamic graph learning. J Ambient Intell Human Comput 14:7679–7693
Deng J, Chen X, Jiang R, Yin D, Yang Y, Song X, Tsang IWH (2023) Disentangling structured components: Towards adaptive, interpretable and scalable time series forecasting. IEEE Trans Knowl Data Eng
Wang Y, Xu Y, Yang J, Wu M, Li X, Xie L, Chen Z (2020) Fully-connected spatial-temporal graph for multivariate time series data. In: AAAI Conference on Artificial Intelligence. https://api.semanticscholar.org/CorpusID:261682449
Kuo PC, Chou YT, Li KY, Chang WT, Huang YN, Chen CS (2024) Gnn-lstm-based fusion model for structural dynamic responses prediction. Eng Struct
Yang X, Li H, Feng X, Jin Z () Variable-wise generative adversarial transformer in multivariate time series anomaly detection. Appl Intell 53:28745–28767
Zhang Z, Zhao Y, Liu J, Wang S, Tao R, Xin R, Zhang J (2018) A general deep learning framework for network reconstruction and dynamics learning. Appl Netw Sci 4
Maaten L, Hinton GE (2008) Visualizing data using t-sne. J Mach Learn Res 9:2579–2605
Wu Y, Lian C, Zeng Z, Xu B, Su Y (2023) An aggregated convolutional transformer based on slices and channels for multivariate time series classification. IEEE Trans Emerg Top Comput Intell 7:768–779
Acknowledgements
The work was supported by the State Grid Corporation Science and Technology Project: Application research of a new generation of artificial intelligence in the security analysis and decision-making of future power grid (5100-202099522A-0-0-00).
Funding
Not applicable.
Author information
Authors and Affiliations
Contributions
Qiong Zuo contributed to the conception of the study; Lipeng Qian performed the experiment and wrote the manuscript; Haiguang Liu contributed significantly to analysis; Hong Zhu helped perform the analysis with constructive discussions.
Corresponding author
Ethics declarations
Conflict of interest/Competing interests
The authors have no competing interests to declare that are relevant to the content of this article.
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Qian, L., Zuo, Q., Liu, H. et al. Multivariate time series classification based on spatial-temporal attention dynamic graph neural network. Appl Intell 55, 115 (2025). https://doi.org/10.1007/s10489-024-06014-8
Accepted:
Published:
DOI: https://doi.org/10.1007/s10489-024-06014-8