Bluetooth-based electroencephalogram signal processing method and system

Abstract

The application discloses a kind of based on bluetooth's electroencephalogram signal processing method and system, the method includes: obtaining the historical channel data and annotation result of the electroencephalogram signal channel of user, and establishes pre-training model and identification model;Pre-training model is deployed to multiple participants, and historical channel data is issued to each participant, to train pre-training model based on the local data of participant in participant, obtain trained pre-training model;And identification is carried out to historical channel data based on trained pre-training model in participant, obtain identification result;Receive the identification result uploaded by each participant, and compare identification result with annotation result, determine comparison result;Identification model training participant is determined based on comparison result, and identification model is trained based on the local data of training participant, to be deployed to the identification terminal of user with trained identification model.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a Bluetooth-based method and system for processing electroencephalogram (EEG) signals. Background Technology

[0002] Brain diseases are prone to sudden onset and are difficult to detect in time. For example, epilepsy is a serious neurological disorder. Electroencephalography (EEG) signals, as an important tool for recording brain activity, are widely used in the detection of brain diseases.

[0003] Existing solutions typically use general recognition models to identify patients' EEG signals, which cannot be specifically trained and recognized based on the user's historical data, resulting in low recognition accuracy. Summary of the Invention

[0004] This invention provides a Bluetooth-based EEG signal processing method and system, which can train a model based on the user's historical EEG data and deploy the trained recognition model to the user's recognition terminal to acquire and recognize the user's EEG signals via Bluetooth communication, thereby improving recognition accuracy.

[0005] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0006] In a first aspect, this application provides a Bluetooth-based EEG signal processing method, the method comprising: acquiring historical channel data of a user's EEG signal channels and corresponding annotation results, and establishing a pre-trained model and a recognition model; deploying the pre-trained model to multiple participants and distributing historical channel data to each participant, so that each participant can train the pre-trained model based on its local data to obtain a trained pre-trained model; and having each participant recognize the historical channel data based on the trained pre-trained model to obtain a recognition result; receiving the recognition results uploaded by each participant and comparing the recognition results with the annotation results to determine the comparison result; determining the training participants for the recognition model based on the comparison result, and training the recognition model based on the local data of the training participants, so as to deploy the trained recognition model to the user's recognition terminal, wherein the recognition terminal receives the channel data of the user's EEG signal channels via Bluetooth and performs recognition based on the trained recognition model.

[0007] Preferably, the step of distributing historical channel data to each participant includes: generating a target random number and processing the historical channel data based on the target random number; adding a data identifier to the processed historical channel data; and distributing the processed historical channel data carrying the data identifier to each participant, so that the trained pre-trained model can reconstruct and identify the processed historical channel data based on the data identifier.

[0008] Preferably, the step of generating a target random number and processing the historical channel data based on the target random number, and adding a data identifier to the processed historical channel data, includes: generating a random number seed and a random batch; generating a target random number based on the random batch and the random number seed; processing the historical channel data based on the target random number and adding a data identifier to the processed historical channel data, wherein the data identifier contains the random batch; wherein the pre-trained model distributed to the participants carries a random number seed to generate the target random number based on the random number seed and the random batch in the data identifier, so as to restore the historical channel data.

[0009] Preferably, adding data identifiers to the processed historical channel data includes: generating a random difference quantity for the target participant, generating a data identifier based on the random batch and the random difference quantity, and adding a data identifier to the processed historical channel data; wherein, the pre-trained model distributed to the participant carries a random number seed and a random difference quantity, so as to restore the random batch based on the data identifier and the random difference quantity, and then generate a target random number based on the random number seed and the random batch to restore the historical channel data.

[0010] Preferably, the method further includes: receiving a pre-trained model uploaded by a participant; inputting historical channel data into the pre-trained model for identification to obtain a center identification result, and comparing the center identification result with the annotation result to determine the comparison result.

[0011] Preferably, the method further includes: determining the numerical range corresponding to the historical channel data based on the historical channel data and the annotation results corresponding to the historical channel data; and generating extended channel data as historical channel data based on the numerical range and the historical channel data.

[0012] Preferably, the step of training the recognition model based on the local data of the training participants includes: deploying the recognition model to the training participants and training the recognition model based on the local data of the training participants to obtain the local parameters of the recognition model; receiving the local parameters of each training participant and performing global analysis to obtain global parameters; determining the parameter update information of each training participant based on the global parameters and sending it to each training participant to update the recognition model of the training participants for the next round of training, until the trained recognition model is determined.

[0013] Secondly, this application provides a Bluetooth-based EEG signal processing system, comprising: a historical data acquisition module, used to acquire historical channel data of a user's EEG signal channels and corresponding annotation results, and to establish a pre-trained model and a recognition model; a training model deployment module, used to deploy the pre-trained model to multiple participants and distribute historical channel data to each participant, so that the participants can train the pre-trained model based on their local data to obtain a trained pre-trained model; and the participants can recognize the historical channel data based on the trained pre-trained model to obtain a recognition result; a comparison result acquisition module, used to receive the recognition results uploaded by each participant and compare the recognition results with the annotation results to determine the comparison result; and a recognition model training module, used to determine the training participants of the recognition model based on the comparison result and train the recognition model based on the local data of the training participants, so as to deploy the trained recognition model to the user's recognition terminal, wherein the recognition terminal receives the channel data of the user's EEG signal channels via Bluetooth and performs recognition based on the trained recognition model.

[0014] Thirdly, this application provides an electronic device, including: a memory and at least one processor; the memory is used to store computer execution instructions; the at least one processor is used to execute the computer execution instructions stored in the memory, causing the at least one processor to perform the method as described in the first aspect.

[0015] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0016] This application can be applied to scenarios involving EEG signal recognition. For example, it can acquire a user's EEG signals via Bluetooth and input them into a trained recognition model for processing to determine the user's state. This solution can select data from relevant participants based on the user's historical channel data for model training, and then deploy the trained model to the user's recognition terminal for EEG signal recognition. Specifically, this solution can acquire the user's historical EEG signal channel data and the corresponding annotation results, and establish a pre-trained model and a recognition model. The pre-trained model has a smaller size and lower complexity than the recognition model. Both the pre-trained model and the recognition model are based on the user's EEG channel data to determine the user's state. Existing solutions use data from a large number of participants for model training, but this often involves a large amount of irrelevant data, leading to excessively large model parameters. This solution utilizes pre-trained models deployed to various participants and selects the training participants for the recognition model based on the pre-trained models trained by each participant, reducing the amount of training data while maintaining recognition quality. Specifically, a pre-trained model can be deployed to multiple participants, and historical channel data can be distributed to each participant. Each participant can then train the pre-trained model using their local data, resulting in a pre-trained model. The pre-trained model can then be used to identify the historical channel data, yielding recognition results. The recognition results uploaded by each participant are received and compared with the labeled results to determine the comparison outcome. Based on the comparison outcome, the training participants for the recognition model are determined, and the recognition model is trained using their local data. The trained recognition model is then deployed to the user's recognition terminal. The recognition terminal receives the user's EEG signal channel data via Bluetooth and performs recognition based on the trained model. Compared to a solution using a general model, this solution leverages the user's historical channel data to select data more suitable for the user for model training. The trained recognition model is then deployed to the user's recognition terminal, acquiring the user's EEG signals via Bluetooth for more accurate recognition, thus improving recognition accuracy. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0018] Figure 1 This is a schematic flowchart of a Bluetooth-based EEG signal processing method according to an embodiment of this application;

[0019] Figure 2 This is a schematic diagram of the structure of a Bluetooth-based EEG signal processing system according to an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This application can be applied to scenarios involving EEG signal recognition. For example, it can acquire a user's EEG signals via Bluetooth and input them into a trained recognition model for processing to determine the user's state. This solution can select data from relevant participants based on the user's historical channel data for model training, and then deploy the trained model to the user's recognition terminal for EEG signal recognition. Specifically, this solution can acquire the user's historical EEG signal channel data and the corresponding annotation results, and establish a pre-trained model and a recognition model. The pre-trained model has a smaller size and lower complexity than the recognition model. Both the pre-trained model and the recognition model are based on the user's EEG channel data to determine the user's state. Existing solutions use data from a large number of participants for model training, but this often involves a large amount of irrelevant data, leading to excessively large model parameters. This solution utilizes pre-trained models deployed to various participants and selects the training participants for the recognition model based on the pre-trained models trained by each participant, reducing the amount of training data while maintaining recognition quality. Specifically, a pre-trained model can be deployed to multiple participants, and historical channel data can be distributed to each participant. Each participant can then train the pre-trained model using their local data, resulting in a pre-trained model. The pre-trained model can then be used to identify the historical channel data, yielding recognition results. The recognition results uploaded by each participant are received and compared with the labeled results to determine the comparison outcome. Based on the comparison outcome, the training participants for the recognition model are determined, and the recognition model is trained using their local data. The trained recognition model is then deployed to the user's recognition terminal. The recognition terminal receives the user's EEG signal channel data via Bluetooth and performs recognition based on the trained model. Compared to a solution using a general model, this solution leverages the user's historical channel data to select data more suitable for the user for model training. The trained recognition model is then deployed to the user's recognition terminal, acquiring the user's EEG signals via Bluetooth for more accurate recognition, thus improving recognition accuracy.

[0022] Specifically, this application provides a Bluetooth-based EEG signal processing method, such as... Figure 1 As shown, the method includes:

[0023] Step 102: Obtain the historical channel data of the user's EEG signal channels and the corresponding annotation results, and establish a pre-trained model and a recognition model.

[0024] Step 104: Deploy the pre-trained model to multiple participants and distribute historical channel data to each participant so that the participants can train the pre-trained model based on their local data to obtain a trained pre-trained model; and then the participants can identify the historical channel data based on the trained pre-trained model to obtain the identification results.

[0025] Step 106: Receive the recognition results uploaded by each participant, compare the recognition results with the annotation results, and determine the comparison results.

[0026] Step 108: Based on the comparison results, determine the training participants of the recognition model, and train the recognition model based on the local data of the training participants, so as to deploy the trained recognition model to the user's recognition terminal. The recognition terminal receives the channel data of the user's EEG signal channel via Bluetooth and performs recognition based on the trained recognition model.

[0027] This application can be applied to scenarios involving EEG signal recognition. For example, it can acquire a user's EEG signals via Bluetooth and input them into a pre-trained recognition model for processing to determine the user's state. This solution can train a model based on the user's historical channel data, selecting data from relevant participants, and then deploying the trained model to the user's recognition terminal for EEG signal recognition. Specifically, this solution can acquire the user's historical EEG signal channel data and corresponding annotation results, and establish a pre-trained model and a recognition model. This solution utilizes the pre-trained model deployed to various participants and selects the training participants for the recognition model based on the pre-trained models trained by each participant, which can reduce the amount of training data while ensuring recognition quality. Specifically, a pre-trained model can be deployed to multiple participants, and historical channel data can be distributed to each participant. Each participant can then train the pre-trained model using their local data to obtain a pre-trained model. The participants can then use the pre-trained model to identify the historical channel data and obtain identification results. The identification results uploaded by each participant can be received and compared with the labeled results to determine the comparison result. Based on the comparison result, the training participants for the identification model are determined, and the identification model is trained using the local data of these training participants. The trained identification model is then deployed to the user's identification terminal, which receives channel data from the user's EEG signal channels via Bluetooth and performs identification based on the trained identification model.

[0028] If historical channel data is sent directly to each participant in plaintext, each participant will receive plaintext data, potentially leading to privacy leaks. Therefore, this solution adds random numbers to the historical channel data and configures corresponding restoration processing for the pre-trained model. The pre-trained model can remove the random numbers before processing the data and perform identification based on the restored data. Specifically, as an optional embodiment, sending historical channel data to each participant includes: generating a target random number and processing the historical channel data based on the target random number; adding a data identifier to the processed historical channel data; and sending the processed historical channel data carrying the data identifier to each participant, so that the trained pre-trained model can restore the processed historical channel data based on the data identifier and then perform identification.

[0029] This solution allows for setting multiple batches of historical channel data, each with a different random number. These random numbers can be generated from the same random number seed, and data processing can be enhanced by generating different random numbers for different batches. Correspondingly, the random number seed can be added to a pre-trained model for data restoration within the model. Specifically, as an optional embodiment, generating a target random number and processing the historical channel data based on it, then adding a data identifier to the processed historical channel data, includes: generating a random number seed and random batches; generating a target random number based on the random batches and the random number seed; processing the historical channel data based on the target random number and adding a data identifier to the processed historical channel data, where the data identifier includes the random batches; wherein, the pre-trained model distributed to the participants carries a random number seed to generate the target random number based on the random number seed and the random batches in the data identifier for historical channel data restoration.

[0030] For different participants, this solution can generate random difference quantities for random batches, so that the same data has different data identifiers for different participants. When restoring random batches, the random batches can be restored based on the random number seed and random difference quantity carried by the pre-trained model, so as to generate corresponding random numbers to restore the data. Specifically, as an optional embodiment, adding data identifiers to the processed historical channel data includes: generating random difference quantities for the target participant, generating data identifiers based on the random batches and random difference quantities, and adding data identifiers to the processed historical channel data; wherein, the pre-trained model sent to the participants carries random number seeds and random difference quantities, so that after restoring random batches based on data identifiers and random difference quantities, target random numbers are generated based on random number seeds and random batches to restore the historical channel data.

[0031] This solution, besides deploying historical channel data to participants for identification, can also receive pre-trained models uploaded by participants on the server side and perform identification. For untrusted participants, this solution can utilize the server to evaluate the pre-trained models to determine whether to use their data for training. Specifically, as an optional embodiment, the method further includes: receiving the pre-trained models uploaded by participants; inputting historical channel data into the pre-trained models for identification to obtain a central identification result; and comparing the central identification result with the annotation result to determine the comparison result. This solution can also expand the data by allowing fluctuations in the historical channel data within a numerical range, enabling the identification and evaluation of a larger amount of data. Specifically, as an optional embodiment, the method further includes: determining the numerical range corresponding to the historical channel data based on the historical channel data and the corresponding annotation results; and generating expanded channel data as historical channel data based on the numerical range and the historical channel data.

[0032] After selecting a small number of training participants from multiple participants, this scheme can adopt a federated learning approach to train the recognition model, thereby protecting the data security and model security of all parties. Specifically, as an optional embodiment, training the recognition model based on the local data of the training participants includes: deploying the recognition model to the training participants and training the recognition model based on the local data of the training participants to obtain the local parameters of the recognition model; receiving the local parameters of each training participant and performing global analysis to obtain global parameters; determining the parameter update information of each training participant based on the global parameters and sending it to each training participant to update the recognition model of the training participants for the next round of training, until the trained recognition model is determined.

[0033] Based on the above embodiments, embodiments of this application also provide a Bluetooth-based EEG signal processing system, such as... Figure 2 As shown, the system includes:

[0034] The historical data acquisition module 202 is used to acquire the historical channel data of the user's EEG signal channels and the corresponding annotation results of the historical channel data, and to establish a pre-trained model and a recognition model.

[0035] The training model deployment module 204 is used to deploy the pre-trained model to multiple participants and distribute historical channel data to each participant so that the participants can train the pre-trained model based on their local data to obtain the trained pre-trained model; and the participants can identify the historical channel data based on the trained pre-trained model to obtain the identification result.

[0036] The comparison result acquisition module 206 is used to receive the recognition results uploaded by each participant, compare the recognition results with the annotation results, and determine the comparison result.

[0037] The recognition model training module 208 is used to determine the training participants of the recognition model based on the comparison results, and to train the recognition model based on the local data of the training participants, so as to deploy the trained recognition model to the user's recognition terminal. The recognition terminal receives the channel data of the user's EEG signal channel via Bluetooth and performs recognition based on the trained recognition model.

[0038] The implementation methods of this application are similar to those of the above-described method embodiments. For specific implementation methods, please refer to the specific implementation methods described above, which will not be repeated here.

[0039] This application can be applied to scenarios involving EEG signal recognition. For example, it can acquire a user's EEG signals via Bluetooth and input them into a pre-trained recognition model for processing to determine the user's state. This solution can train a model based on the user's historical channel data, selecting data from relevant participants, and then deploying the trained model to the user's recognition terminal for EEG signal recognition. Specifically, this solution can acquire the user's historical EEG signal channel data and corresponding annotation results, and establish a pre-trained model and a recognition model. This solution utilizes the pre-trained model deployed to various participants and selects the training participants for the recognition model based on the pre-trained models trained by each participant, which can reduce the amount of training data while ensuring recognition quality. Specifically, a pre-trained model can be deployed to multiple participants, and historical channel data can be distributed to each participant. Each participant can then train the pre-trained model using their local data to obtain a pre-trained model. The participants can then use the pre-trained model to identify the historical channel data and obtain identification results. The identification results uploaded by each participant can be received and compared with the labeled results to determine the comparison result. Based on the comparison result, the training participants for the identification model are determined, and the identification model is trained using the local data of these training participants. The trained identification model is then deployed to the user's identification terminal, which receives channel data from the user's EEG signal channels via Bluetooth and performs identification based on the trained identification model.

[0040] Based on the above embodiments, this application also provides an electronic device, including: a memory and at least one processor; the memory is used to store computer execution instructions; the at least one processor is used to execute the computer execution instructions stored in the memory, causing the at least one processor to perform the method as described in the above embodiments.

[0041] This invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, this computer program implements the various processes of the above-described data processing method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0043] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0044] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0045] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0046] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0047] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0048] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0049] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0050] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0051] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A Bluetooth-based EEG signal processing method, characterized in that, The method includes: Obtain historical channel data of the user's EEG signal channels and the corresponding annotation results, and establish a pre-trained model and a recognition model; The pre-trained model is deployed to multiple participants, and historical channel data is distributed to each participant so that the participants can train the pre-trained model based on their local data to obtain a trained pre-trained model; and the participants can then use the trained pre-trained model to identify the historical channel data to obtain the identification results. Receive the recognition results uploaded by each participant, compare the recognition results with the annotation results, and determine the comparison results; The training participants for the recognition model are determined based on the comparison results, and the recognition model is trained based on the local data of the training participants. The trained recognition model is then deployed to the user's recognition terminal, which receives channel data of the user's EEG signal channel via Bluetooth and performs recognition based on the trained recognition model.

2. The method according to claim 1, characterized in that, The distribution of historical channel data to each participant includes: Generate a target random number and process the historical channel data based on the target random number; add data identifiers to the processed historical channel data; The processed historical channel data, carrying data identifiers, is distributed to all participants so that the pre-trained model can reconstruct and identify the processed historical channel data based on the data identifiers.

3. The method according to claim 2, characterized in that, The process involves generating a target random number and processing historical channel data based on that target random number. Add data identifiers to the processed historical channel data, including: Generate a random number seed and a random batch, and generate the target random number based on the random batch and the random number seed; Historical channel data is processed based on a target random number, and a data identifier is added to the processed historical channel data. The data identifier contains a random batch. The pre-trained model distributed to the participants carries a random number seed, which is used to generate a target random number based on the random number seed and the random batch in the data identifier, in order to restore the historical channel data.

4. The method according to claim 3, characterized in that, The step of adding data identifiers to the processed historical channel data includes: A random difference quantity is generated for the target participants, and a data identifier is generated based on the random batch and the random difference quantity. The data identifier is also added to the processed historical channel data. The pre-trained model distributed to the participants carries a random number seed and a random difference quantity. After restoring the random batch based on the data identifier and the random difference quantity, a target random number is generated based on the random number seed and the random batch to restore the historical channel data.

5. The method according to claim 1, characterized in that, The method further includes: Receive pre-trained models uploaded by the participants; Historical channel data is input into a pre-trained model for identification to obtain the center identification result. The center identification result is then compared with the annotation result to determine the comparison result.

6. The method according to claim 5, characterized in that, The method further includes: Based on historical channel data and the corresponding annotation results, determine the numerical range corresponding to the historical channel data; Based on the numerical range and historical channel data, extended channel data is generated as historical channel data.

7. The method according to claim 1, characterized in that, The training of the recognition model based on local data from the training participants includes: The recognition model is deployed to the training participants, and trained on the local data of the training participants to obtain the local parameters of the recognition model. Receive local parameters from each training participant and perform global analysis to obtain global parameters; Based on the global parameters, the parameter update information for each training participant is determined and sent to each training participant to update their recognition model for the next round of training, until the trained recognition model is determined.

8. A Bluetooth-based EEG signal processing system, characterized in that, The system includes: The historical data acquisition module is used to acquire the user's historical channel data of EEG signals and the corresponding annotation results of the historical channel data, and to build a pre-trained model and a recognition model. The training model deployment module is used to deploy the pre-trained model to multiple participants and distribute historical channel data to each participant so that the participants can train the pre-trained model based on their local data to obtain the trained pre-trained model; and the participants can then use the trained pre-trained model to identify the historical channel data to obtain the identification results. The comparison result acquisition module is used to receive the recognition results uploaded by each participant, compare the recognition results with the annotation results, and determine the comparison result; The recognition model training module is used to determine the training participants of the recognition model based on the comparison results, and to train the recognition model based on the local data of the training participants, so as to deploy the trained recognition model to the user's recognition terminal. The recognition terminal receives the channel data of the user's EEG signal channel via Bluetooth and performs recognition based on the trained recognition model.

9. An electronic device, characterized in that, include: Memory and at least one processor; The memory is used to store computer-executed instructions; The at least one processor is configured to execute computer execution instructions stored in the memory, causing the at least one processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, A computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-7.

Images