Method for identifying few samples based on brain-computer interface

Abstract

The invention discloses a few-sample identification method based on a brain-computer interface. The method comprises the following steps: acquiring EEG data of one target subject and m auxiliary subjects for executing N types of motor imagery tasks; in the pre-training stage, learning a feature encoder of the EEG by using the auxiliary tested data, and taking the feature encoder as an initial model parameter of the subsequent meta-learning stage; in the meta-learning stage, optimization is carried out by taking a task as a scale on the basis of pre-training, so that the model can better adapt to a few-sample task constructed by target tested data. According to the method, the performance is further improved by using meta-learning on the basis of the pre-training model, and the method can be applied to a few-sample classification problem in different BCI scenes by simply changing a feature coding module; in the face of a new task constructed by target tested data, the method can show better generalization performance.

Description

technical field [0001] The present invention relates to the technical field of EEG signal classification and identification, in particular to a few-sample identification method based on a brain-computer interface. Background technique [0002] Brain-Computer Interface (BCI) can convert human thinking activities or intentions into corresponding computer instructions by recognizing a series of electroencephalogram (EEG) signals, so as to realize direct communication or control of external devices by the brain. Motor Imagery (MI), as an active control paradigm, is widely used in medical rehabilitation and other fields, and has become a research hotspot in the current BCI field. [0003] Although this BCI system that can be fully autonomously controlled has been favored by many scholars and has been widely studied, its development still faces many challenges, the most important of which is the long initial calibration time of the current motor imagery-based BCI system , that is...

AI Technical Summary

Problems solved by technology

As meta-learning has gradually become a very potential learning paradigm in few-shot classification problems, relevant research has applied it to the field of brain-computer interface, but some recent work that combines the advantages of transfer learning and meta-learning is in the field of vision. Established new state-of-the-art results, however, this type of method is still less explored in the field of brain-computer interface

Images