A ML-based Approach to Enhance Metrological Performance of Wearable Brain-Computer Interfaces

In this paper, the adoption of Machine Learning (ML) classifiers is addressed to improve the performance of highly wearable, single-channel instrumentation for Brain-Computer Interfaces (BCIs). The proposed BCI is based on the classification of Steady-State Visually Evoked Potentials (SSVEPs). In this setup, Augmented Reality Smart Glasses are used to generate and display the flickering stimuli for the SSVEP elicitation. An experimental campaign was conducted on 20 adult volunteers. Successively, a Leave-One-Subject-Out Cross Validation was performed to validate the proposed algorithm. The obtained experimental results demonstrate that suitable ML-based processing strategies outperform the state-of-the-art techniques in terms of classification accuracy. Furthermore, it was also shown that the adoption of an inter-subjective model successfully led to a decrease in the 3-σ uncertainty: this can facilitate future developments of ready-to-use systems.