Identification of Switches by Machine Learning of EEG Data When Listening to Switch Sounds

The purpose of this study was to investigate the possibility of identifying switches based on electro-encephalography (EEG) data when listening to switch sounds, i.e., sounds emitted when switches are pressed. We collected data using a 32-channel recorder when participants heard 24 switch sounds. The EEG data acquired were learned and classified using EEGNet, which is a neural network. High accuracy was obtained when cross-validation was performed by exchanging data evoked by sounds repeated multiple times for a specific participant and a specific switch sound.