Machines à Vecteurs de Support Quantiques Appliquées à la Classification des Signaux EEG en Géométrie Riemannienne
Anton Andreev, Grégoire Cattan · HAL (Le Centre pour la Communication Scientifique Directe) · 2023
In this work, we explored the performance of a quantum-enhanced support vector machine versus linear discriminant analysis, for the classification of electroencephalography (EEG) recordings. The data were prepared and vectorized using Riemannian Geometry, a ubiquitous method for EEG analysis. The results demonstrate that quantum classification achieved a good performance, although lower than the one achieved with LDA. We conclude that quantum computation does not provide an advantage as compared to classical computation for the classification of well-separable data. Further studies needs to investigate if quantum computation could offer an advantage in situations where classical classification fails.