HMT: An EEG Signal Classification Method Based on CNN Architecture

Electroencephalogram (EEG) is a non-invasive measurement method commonly used to measure human brain electrical activity,It detects weak electrical signals generated by neurons in the cerebral cortex by placing electrodes on the scalp. These signals reflect the bioelectricity activity of the brain when thinking, perception, emotion, behavior and other activities.It has been used to measure the electrical activity of the human brain. The EEG signal can reflect the changes in the brain waves of the human brain, it is widely used to help doctors diagnose neurological diseases such as motor neurons, Parkinson's disease, and spasms. Low signal-to-noise ratio and high nonlinearity are the characteristics of EEG signals, which bring great challenges and difficulties to the analysis and recognition of EEG signals. Convolutional neural network is a great breakthrough technology in deep learning, which can deal with highly nonlinear and noisy data, and can automatically learn the distribution characteristics of EEG data and the classification of EEG signals. Compared with traditional EEG signal classification methods, the convolutional neural network has great potential in dealing with complex EEG signals. In this paper, a deep learning algorithm model HMT based on CNN architecture is proposed to process EEG signals, and its application in EEG signal classification and recognition is discussed.