Robust Motor Imagery Tasks Classification Approach Using Bayesian Neural Network

• Published in: Sensors (Basel, Switzerland) Vol. 23; no. 2; p. 703
• Main Authors: Milanés-Hermosilla, Daily, Trujillo-Codorniú, Rafael, Lamar-Carbonell, Saddid, Sagaró-Zamora, Roberto, Tamayo-Pacheco, Jorge Jadid, Villarejo-Mayor, John Jairo, Delisle-Rodriguez, Denis
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 08.01.2023; MDPI
• Subjects: Algorithms; Bayes Theorem; Bayesian Neural Network; Brain research; Brain-Computer Interfaces; brain–computer interface; Classification; classification with reject option; Datasets; Deep learning; Diabetic retinopathy; Electroencephalography; Electroencephalography - methods; Imagery, Psychotherapy; Imagination; Neural networks; Neural Networks, Computer; Physiology; Random variables; uncertainty estimation; variational inference; uncertainty estimation; brain–computer interface; classification with reject option; variational inference; Bayesian Neural Network
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s23020703

The development of Brain–Computer Interfaces based on Motor Imagery (MI) tasks is a relevant research topic worldwide. The design of accurate and reliable BCI systems remains a challenge, mainly in terms of increasing performance and usability. Classifiers based on Bayesian Neural Networks are proposed in this work by using the variational inference, aiming to analyze the uncertainty during the MI prediction. An adaptive threshold scheme is proposed here for MI classification with a reject option, and its performance on both datasets 2a and 2b from BCI Competition IV is compared with other approaches based on thresholds. The results using subject-specific and non-subject-specific training strategies are encouraging. From the uncertainty analysis, considerations for reducing computational cost are proposed for future work.