Electroencephalogram signal classification method based on collaborative contrast regularization average teacher model

Abstract

The invention provides an EEG (electroencephalogram) signal classification method based on a collaborative contrast regularization average teacher model, which comprises the following steps of: firstly, preprocessing an EEG (electroencephalogram) signal, generating EEG data under a new view angle by utilizing a two-stage EEG data augmentation method, and generating EEG data under another view angle in a time-frequency domain; utilizing student and teacher networks in the average teacher model to respectively learn EEG data characteristics of different visual angles; the consistency loss is used for encouraging the student and teacher network to predict the consistency of EEG data at different viewing angles, the collaborative comparison loss is used for encouraging the consistency of EEG data structures at different viewing angles, and finally, weighted summation with cross entropy loss is carried out to optimize network parameters. According to the two-stage EEG data augmentation method provided by the invention, change factors of a time domain and a time-frequency domain can be captured at the same time, the proposed collaborative contrast regularization average teacher model can learn data level information under different view angles and consistency information of data structures among different view angles, and the robustness, discrimination and generalization of the model are improved.

Description

technical field [0001] The invention relates to an electroencephalogram signal identification method based on a synergistic contrast regularization average teacher model, and belongs to the field of electroencephalogram signal identification. Background technique [0002] Brain Computer Interface (BCI) refers to creating a direct connection between humans or animals and external devices to realize the exchange of information between the brain and the device. At present, BCI systems have been widely used to help patients with stroke and spinal cord injury control external devices to improve their quality of life. Electroencephalogram (EEG) is a bioelectric phenomenon produced by the transfer of information in the form of ions between brain cell groups, and is the overall reflection of neuronal electrophysiological activities on the surface of the cerebral cortex or scalp. Compared with other types of brain signals, EEG has the advantages of convenient acquisition and high te...

AI Technical Summary

Problems solved by technology

EEG contains rich time, frequency domain, space and other information, and features can be represented in multiple domains (such as time domain, time-frequency domain, etc.). However, existing EEG data augmentation methods usually generate new augmentations in a single domain. data, ignoring the variable factors of EEG data in multiple domains, the quality of augmented EEG data is not high, which brings challenges to the establishment of a robust EEG recognition model

In addition, most of the current deep learning methods only consider the information at the EEG data level from different perspectives, ignoring the information at the EEG data structure level that is invariant between different perspectives, and it is difficult to break through the performance bottleneck of the EEG recognition model.

Images