Multi-Scale Convolutional Neural Network Based Ear- Electroencephalograph Identity Recognition

Biometrics are widely used in the field of information security. Compared with other biometric technologies (such as fingerprint and iris recognition), electroencephalograph (EEG) rely on their high degree of individual specificity and are difficult to forge. EEG have gradually become a focus of identification research. However, the existing EEG recognition technology still faces problems such as complex equipment and inconvenient operation in its practical application. In particular, there are significant shortcomings in terms of portability and utility. To solve these problems, this paper innovatively proposes a multi-scale convolutional neural network based on a single-channel portable ear-EEG device. Combining existing feature extraction algorithms, computing multi-dimensional features of single-channel EEG to ultimately optimize classification performance. The experiment invited 14 subjects to participate in the identification test, and the results showed that the model achieved an average classification accuracy of 86.48%• The experimental results verify the effectiveness of ear EEG in identity recognition applications. Expanding the Application Prospects of Ear-Electroencephalography Identity Recognition Technology.