Brainprint recognition method based on convolutional neural network

Abstract

The invention relates to a brainprint recognition method based on a convolutional neural network. The brainprint recognition method comprises the steps: S1, collecting original time sequence electroencephalogram data, and obtaining artifact-removed original frequency domain electroencephalogram data; S2, dividing the artifact-removed original frequency domain electroencephalogram data into a firstdata set and a second data set, and training and testing the convolutional neural network through the first data set to obtain a trained convolutional neural network; S3, respectively performing feature extraction on the first data set and the second data set, and storing all extracted feature sets as feature sets; and S4, extracting the first feature from the second data set, and comparing the Euclidean distance between the first feature and each feature in the stored feature set to obtain a brainprint identification result. Compared with the prior art, the brainprint recognition method hasthe advantages that the electroencephalogram coding vectors of the individuals are extracted by training the convolutional neural network, and the brainprint of the individuals can be recognized onlyby comparing the Euclidean distances between the individuals and the samples, and the problem of model retraining is avoided.

Description

technical field [0001] The invention relates to the technical field of brain science, in particular to a brain pattern recognition method based on a convolutional neural network. Background technique [0002] Brain pattern recognition using EEG data is a very popular direction in the field of EEG signal research in recent years. Brain pattern is a feature composite map based on specific brain wave signals. Brain pattern recognition technology relies on extracting the human brain when browsing and memorizing specific information. The generated brain wave signal is subjected to feature extraction, analysis and recognition. However, because EEG data has no standardized acquisition steps, is difficult to preprocess, and has the characteristics of abstract representation, it is often impossible to find a suitable model for identifying brain patterns when using machine learning methods to analyze EEG data. [0003] First of all, the EEG data collection process will be accompanied...

AI Technical Summary

Problems solved by technology

However, because EEG data has no standardized collection steps, is difficult to preprocess, and has the characteristics of abstract expression, when using machine learning methods to analyze EEG data, it is often impossible to find a suitable model for identifying brain patterns.

[0003] First of all, the EEG data collection process will be accompanied by a lot of noise, and it is difficult to filter these noises in the preprocessing process; second, when choosing a machine learning classifier, such as SVM or decision tree, different data preprocessing effects on The recognition results output by the model have different effects; in addition, when the number of collected individuals of EEG data increases, the previous model needs to be retrained to be able to recognize new individuals

Images