Deep learning man-machine interaction motor imagery brain-computer interface system and training method
A technology of motor imagery and brain-computer interface, applied in the fields of application, medical science, psychotherapy, etc., can solve problems such as poor stability of EEG signals, no literature and reports found, and low dimensionality of classification results, and achieve a wide range of applications and expansion Effects that apply space and improve accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0038] Brain-computer interface technology has a very promising application space in many fields such as military affairs, medical rehabilitation, education and entertainment. MI-BCI has been the first to be explored and laid out by various military powers in terms of coordinated command of military operations and control of military weapons; there is also great room for exploration in application scenarios such as medical rehabilitation training and exoskeleton control; at the same time, in education training and toy games Residential consumer markets, such as residential areas, are being tapped by many large companies and start-ups.
[0039] In the above application fields and scenarios, the accuracy and stability of MI-BCI need to be improved. Traditional research on MI-BCI decoding algorithms and paradigms cannot fundamentally solve the problem of incorrect and inaccurate imagination of users, nor can it complete the personalized adaptation of users and algorithms.
[004...
Embodiment 2
[0051] The overall composition and specific design of the human-computer adaptive motor imagery brain-computer interface system (DeCa-BCI) based on deep learning are the same as those in Embodiment 1. The construction of the STD-Net module in the present invention is as follows:
[0052] see figure 1 , the STD-Net module is divided into the ST-Net sub-module and the discrimination network layer sub-module, the ST-Net sub-module contains the ST-Net network, and the discrimination network layer sub-module contains the discrimination network layer; the ST-Net sub-module realizes the brain The electrical EEG signal is transformed from the two-dimensional matrix data of time T* lead C into a signal conversion image related to the type of motor imagery task that the user can intuitively identify, and the signal conversion image is grayscale with strong discriminative information Image and shape can be set by yourself, for example, the grayscale image is in the shape of left or right...
Embodiment 3
[0059] The overall composition and specific design of the human-computer adaptive motor imagery brain-computer interface system (DeCa-BCI) based on deep learning are the same as those in Example 1-2. The data set module is constructed as follows:
[0060] The data set of the present invention includes the original EEG data and the matching signal conversion target image and category label; the data format of the original EEG data is the same as the output of the system's EEG data acquisition module; the category label is the category of motor imagination ; see Figure 4 , the signal conversion target image is a grayscale image that contains strong discriminative information consistent with the class label. For example, define the signal conversion target image of the left and right hand motor imagery task as a grayscale image of the left or right cross star, with an aspect ratio of 2:1, a png format image with a black background and a white cross on one side.
[0061] The dat...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More