Method for Adaptive EEG signal processing using reinforcement learning and System Using the same

摘要

The present invention relates to an adaptive EEG analysis method using deep reinforcement learning, and an apparatus thereof and, more specifically, to a method and apparatus for pre-processing an EEG signal, encoding the same to a neural network, and selecting an EEG signal classifier previously learned from an attention module learned through deep reinforcement learning to classify the encoded EEG signal. According to the present invention, the method for classifying EEG signals achieves high EEG signal classification accuracy, and even when the status of a subject or a data module learning a classifier or the subject affecting the EEG is different, selects the classifier suitable for the subject in advance, thereby grouping and processing separately tested data at the same time. In addition, the optimal classifier for each subject context information is selected, thereby performing fast EEG analysis near real-time in comparison to an existing ensemble system without waiting for the results of all classifiers.