Abstract
A great challenge for brain-computer interface (BCI) systems is their deployment in clinical settings or at home, where a BCI system can be used with limited calibration sessions. BCI should be ideally self-trained and take advantage of unlabeled data. When performing a task, the EEG signals change over time, hence the recorded signals have non-stationary properties. It is necessary to provide machine-learning approaches that can deal with self-training and/or use semi-supervised learning methods for signal classification. A key problem in graph-based semi-supervised learning is determining the characteristics of the affinity matrix that defines the relationships between examples, including the size of the neighborhood of each example. In this paper, we propose two approaches for building the affinity matrix using the distance between examples and the number of neighbors, with a limited number of hyper-parameters, making it easy to reuse. We also compare the Euclidean distance and Riemannian geometry distances to construct the affinity matrix. We assess the classification performance with motor imagery data with two classes from a publicly available dataset of 14 participants. The results show the interest of the proposed semi-supervised approaches with the use of distances to define the neighborhood using Riemannian geometry-based distances with an average accuracy of 73.75%.
This study was supported by the NIH-R15 NS118581 project.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Boashash, B., Azemi, G., Ali Khan, N.: Principles of time-frequency feature extraction for change detection in non-stationary signals: Applications to newborn eeg abnormality detection. Pattern Recogn. 48(3), 616–627 (2015)
Cecotti, H.: Active graph based semi-supervised learning using image matching: Application to handwritten digit recognition. Pattern Recogn. Lett. 73, 76–82 (2016)
Chapelle, O., Schölkopf, B., Zien, A. (eds.): Adaptive computation and machine learning series. MIT Press, Cambridge, Massachusetts (2010)
Cincotti, F., et al.: Non-invasive brain-computer interface system: Towards its application as assistive technology. Brain Res. Bull. 75(6), 796–803 (2008)
Klonowski, W.: From conformons to human brains: an informal overview of nonlinear dynamics and its applications in biomedicine. Nonlinear Biomed. Phys. 1(1), 5–5 (2007)
Li, Y., Wong, K.M., deBruin, H.: Eeg signal classification based on a riemannian distance measure. In: 2009 IEEE Toronto International Conference Science and Technology for Humanity (TIC-STH), pp. 268–273 (2009). https://doi.org/10.1109/TIC-STH.2009.5444491
Nicolas-Alonso, L.F., Gomez-Gil, J.: Brain computer interfaces, a review. Sensors 12(2), 1211–1279 (2012). https://doi.org/10.3390/s120201211, https://www.mdpi.com/1424-8220/12/2/1211
Padfield, N., Zabalza, J., Zhao, H., Masero, V., Ren, J.: Eeg-based brain-computer interfaces using motor-imagery: Techniques and challenges. Sensors 19, 1423 (2019). https://doi.org/10.3390/s19061423
Raza, H., Cecotti, H., Li, Y., Prasad, G.: Adaptive learning with covariate shift-detection for motor imagery based brain-computer interface. Soft. Comput. 20(8), 3085–3096 (2016)
Raza, H., Rathee, D., Zhou, S.M., Cecotti, H., Prasad, G.: Covariate shift estimation based adaptive ensemble learning for handling non-stationarity in motor imagery related EEG-based brain-computer interface. Neurocomputing (2018)
Song, Z., Yang, X., Xu, Z., King, I.: Graph-based semi-supervised learning: A comprehensive review. IEEE Trans. Neural Netw. Learn. Syst., 1–21 (2022)
Steyrl, D.: Two class motor imagery (002–2014) (2020). http://bnci-horizon-2020.eu/database/data-sets
Varoquaux, G., Raamana, P.R., Engemann, D.A., Hoyos-Idrobo, A., Schwartz, Y., Thirion, B.: Assessing and tuning brain decoders: Cross-validation, caveats, and guidelines. NeuroImage (Orlando, Fla.) 145(Pt B), 166–179 (2017)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Smrkovsky, E., Cecotti, H. (2023). Graph-Based Semi-supervised Learning Using Riemannian Geometry Distance for Motor Imagery Classification. In: Rodríguez-González, A.Y., Pérez-Espinosa, H., Martínez-Trinidad, J.F., Carrasco-Ochoa, J.A., Olvera-López, J.A. (eds) Pattern Recognition. MCPR 2023. Lecture Notes in Computer Science, vol 13902. Springer, Cham. https://doi.org/10.1007/978-3-031-33783-3_30
Download citation
DOI: https://doi.org/10.1007/978-3-031-33783-3_30
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-33782-6
Online ISBN: 978-3-031-33783-3
eBook Packages: Computer ScienceComputer Science (R0)