ICA가 청각 상상 뇌파 분류에 미치는 영향

초록

While conventional preprocessing methods for Electroencephalogram(EEG) signals can significantly impact signal analysis and classification performance, the influence of Independent Component Analysis(ICA) on auditory stimulus and auditory imagery data, as well as its evaluation using deep learning, has not been thoroughly investigated. This study analyzed the effects of ICA application on artifact removal in actual speech and imagined speech data for auditory stimuli using Event-Related Potential(ERP), a key characteristic of EEG signals, and evaluated the impact of ICA on model performance through deep learning models. The findings suggest that ICA may be effective for artifact removal in auditory stimulus and auditory imagery data. However, it has limitations in maintaining the temporal consistency of ERP signals, improving deep learning model performance, and enhancing the extraction of useful features.

키워드