Abstract
Over the last years, several time series classification (TSC) algorithms have been proposed both in traditional machine learning and deep learning domains which have shown remarkable enhancement over the previously published state-of-the-art methods. However, their decision-making processes generally stay as black boxes to the user. Model-agnostic (post-hoc) explainers, such as the state-of-the-art SHAP, are proposed to make the predictions of machine learning models explainable with the presence of well-designed domain mappings. In our paper, we first apply univariate classifiers on the dimensions of multivariate time series data individually. This is a straightforward technique for multivariate time series classification (MTSC). Then, we use state-of-the-art timeXplain framework to interpret the decision making process of the univariate classifiers on the multivariate time series data. With a careful choice of interpretability parameters, we demonstrate that it is possible to obtain explainability for such time series data.
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
References
Abanda, A., Mori, U., Lozano, J.A.: A review on distance based time series classification. Data Min. Knowl. Disc. 33(2), 378–412 (2019)
Babayev, R., Wiese, L.: Benchmarking classifiers on medical datasets of UEA archive. In: Proceedings of AI Health WWW 2021: International Workshop on AI in Health (2021)
Bagnall, A., et al.: The UEA multivariate time series classification archive (2018). arXiv preprint arXiv:1811.00075 (2018)
Du, M., Liu, N., Hu, X.: Techniques for interpretable machine learning. Commun. ACM 63(1), 68–77 (2019)
Giorgino, T.: Computing and visualizing dynamic time warping alignments in R: the dtw package. J. Stat. Softw. 31(7), 1–24 (2009). https://doi.org/10.18637/jss.v031.i07
Löning, M., Bagnall, A., Ganesh, S., Kazakov, V., Lines, J., Király, F.J.: sktime: a unified interface for machine learning with time series. In: Workshop on Systems for ML at NeurIPS 2019 (2019)
Lundberg, S.M., Lee, S.I.: A unified approach to interpreting model predictions. In: Advances in Neural Information Processing Systems, pp. 4765–4774 (2017)
Mujkanovic, F., Doskoč, V., Schirneck, M., Schäfer, P., Friedrich, T.: timeXplain - a framework for explaining the predictions of time series classifiers. arXiv preprint arXiv:2007.07606 (2020)
Ruiz, A.P., Flynn, M., Bagnall, A.: Benchmarking multivariate time series classification algorithms. arXiv preprint arXiv:2007.13156 (2020)
Acknowledgement
This work was supported by the Fraunhofer Internal Programs under Grant No. Attract 042-601000. We thank to the providers of timeXplain library (Felix Mujkanovic and others) for their source code and data set donors of the UEA archive (Anthony Bagnall and others) for their valuable datasets.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Babayev, R., Wiese, L. (2021). Interpreting Decision-Making Process for Multivariate Time Series Classification. In: Bellatreche, L., et al. New Trends in Database and Information Systems. ADBIS 2021. Communications in Computer and Information Science, vol 1450. Springer, Cham. https://doi.org/10.1007/978-3-030-85082-1_14
Download citation
DOI: https://doi.org/10.1007/978-3-030-85082-1_14
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-85081-4
Online ISBN: 978-3-030-85082-1
eBook Packages: Computer ScienceComputer Science (R0)