Abstract:Focal EEG recognition can provide important reference value for epilepsy surgery. This paper proposes a focal EEG recognition algorithm based on deep network with transfer learning. Firstly, the continuous wavelet transform (CWT) is used to perform timefrequency analysis on the EEG signals and obtain the timefrequency map of the EEG signals. Then transfer learning is performed on the AlexNet model, and the network structure is adjusted to adapt to focal EEG recognition. The output of the seventh fully connected layer of the model is used as the characteristic presentation of the timefrequency images. Finally, the classification algorithms of SVM, BP, LSTM, SRC and LDA are used to classify the features. In this paper, based on the open source EEG dataset, the 10fold crossvalidation algorithm is adopted to verify the algorithm, the effects of the six classifiers are compared. The average specificity, sensitivity and accuracy of the SVM algorithm are 8807%, 8881% and 8844%, respectively, which proves the effectiveness of the method in focal EEG recognition. .txt