Massage system, massage method, electronic device, and computer-readable medium based on machine learning

Abstract

The present application provides a massage system, a massage method, an electronic device, and a computer-readable medium based on machine learning. In the massage system, a sensor group detects the status of a user to generate a user status signal and sends it to a main control chip; the main control chip encapsulates the user status signal to generate a user status data packet; the intelligent massage recommendation model parses the user status data packet to generate an intelligent massage recommendation strategy and sends it to the main control chip, so that the main control chip generates an action execution instruction according to the intelligent massage recommendation strategy; the actuator generates a massage action for a target part and a warm and synergistic action for the target part according to the action execution instruction; the human-computer interaction unit enables the user to perform human-computer interaction with a front-end device at least based on the user status signal and generates a human-computer interaction instruction, so as to adjust at least one of the massage action and the warm and synergistic action of the actuator based on the human-computer interaction instruction.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a massage system, massage method, electronic device, and computer-readable medium based on machine learning. Background Art

[0002] Existing massage devices are basically based on the motor to simply generate vibration or add a heating device to perform body massage. The user then manually adjusts the frequency, temperature, position and other settings to massage and stimulate specific parts of the body. The massage device cannot obtain the user's true feeling in a closed loop. Moreover, in a single use, the user still needs to manually operate the device to gradually explore the best usage mode, and there is a lack of intelligent settings to enable the user to achieve the best experience. Summary of the invention

[0003] The purpose of this application is to propose a massage system, massage method, electronic device, and computer-readable medium based on machine learning to solve or alleviate technical problems existing in the prior art.

[0004] A massage system based on machine learning, comprising: a cloud server and a front-end device, wherein the cloud server is communicatively connected to the front-end device; an intelligent massage recommendation model based on machine learning is deployed on the cloud server; the front-end device comprises: a sensor group, an actuator, a data interface module, a human-computer interaction unit, and a main control chip, wherein the sensor group, the actuator, the data interface module, and the human-computer interaction unit are all connected to the main control chip;

[0005] The sensor group is used to detect the status of the user to generate a user status signal and send it to the main control chip;

[0006] The main control chip is used to encapsulate the user status signal to generate a user status data packet, and upload the user status data packet to the cloud server through the data interface module;

[0007] The intelligent massage recommendation model is used to parse the user status data packet to generate an intelligent massage recommendation strategy and send it to the main control chip, so that the main control chip generates an action execution instruction according to the intelligent massage recommendation strategy;

[0008] The execution mechanism is used to generate a massage action for the target part and a warming action for the target part according to the action execution instruction;

[0009] The human-computer interaction unit is at least used to enable the user to perform human-computer interaction with the front-end device based on the user status signal and generate human-computer interaction instructions, so as to adjust at least one of the massage action and the warming enhancement action of the actuator based on the human-computer interaction instructions.

[0010] Optionally, the sensor group includes at least one of a temperature sensor, a vibration sensor, and a gloss sensor; the temperature sensor is used to monitor the temperature of the target part and generate a temperature monitoring signal; the vibration sensor is used to monitor the vibration of the target part and generate a vibration monitoring signal; the gloss sensor is used to monitor the gloss of the target part and generate a gloss monitoring signal; and the user status signal includes at least one of the temperature monitoring signal, the vibration monitoring signal, and the gloss monitoring signal.

[0011] Optionally, when the main control chip encapsulates the user status signal to generate a user status data packet, and uploads the user status data packet to the cloud server through the data interface module, the following steps are performed:

[0012] Preprocessing the user status signal to obtain a preprocessed user status signal;

[0013] Formatting the preprocessed user status signal to obtain a data packet with a standard data structure, and adding a timestamp to each data packet to obtain formatted user status data;

[0014] Encrypting the formatted user status data to obtain encrypted user status data;

[0015] compressing the encrypted user status data to obtain compressed user status data;

[0016] The compressed user status data is uploaded to the cloud server through the data interface module.

[0017] Optionally, when the intelligent massage recommendation model parses the user status data packet to generate an intelligent massage recommendation strategy, the following steps are performed:

[0018] Based on the unpacking module of the intelligent massage recommendation model, the user status data packet is decoded to obtain the original user status data;

[0019] Based on the unpacking module of the intelligent massage recommendation model, feature extraction is performed on the original user state data to obtain user feature data, and embedding processing is performed on the user feature data to obtain a user feature vector;

[0020] Based on the inference module in the intelligent massage recommendation model, inference is performed according to the user feature vector to generate an intelligent massage recommendation strategy.

[0021] Optionally, when the inference module in the intelligent massage recommendation model performs inference according to the user feature vector to generate an intelligent massage recommendation strategy, the following steps are performed:

[0022] The inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, and performs a multi-layer full-connection process on the hidden feature vector to generate an intelligent massage recommendation strategy, wherein the intelligent massage recommendation strategy defines at least one of a recommended massage type, massage intensity, massage time, and recommended temperature.

[0023] Optionally, when the inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, a multi-head attention analysis is performed on all the extracted hidden feature vectors to perform weighted processing on all the hidden feature vectors to obtain attention-weighted hidden feature vectors, so that when the hidden feature vector is subjected to multi-layer full-connection processing, the attention-weighted hidden feature vector is subjected to multi-layer full-connection processing.

[0024] Optionally, before the inference module performs multi-head attention analysis on all the extracted hidden feature vectors, it performs regularized distribution on all the hidden feature vectors so that all the hidden feature vectors are close to a preset prior distribution, so as to obtain regularized distributed hidden feature vectors, and perform multi-head attention analysis on the regularized distributed hidden feature vectors.

[0025] Optionally, based on the following objective function, the reasoning module performs regularized distribution on all hidden feature vectors:

[0026] Wherein, q(z|x) represents the distribution of the hidden feature vector before regularization, p(z) represents the preset prior distribution, which is the standard normal distribution, KL(q(z|x)||p(z)) represents the divergence term, which is used to control the distribution of the hidden feature vector to be regularized so that it is close to the prior distribution p(z), and obtain the hidden feature vector z of the regularized distribution, x represents the user state data, It measures the probability that z reconstructs the input data x. The expectation E is calculated on the distribution of the latent variable q(z|x), KL(q(z|x)||p(z)) represents the KL divergence between q(z|x) and p(z), which measures the difference between the two probability distributions, and p(x|z) represents the probability of reconstructing x given z.

[0027] A massage method based on machine learning, comprising:

[0028] Detecting the user's status to generate a user status signal;

[0029] Encapsulating the user status signal to generate a user status data packet, and uploading the user status data packet to the cloud server through the data interface module;

[0030] The intelligent massage recommendation model is used to parse the user status data packet to generate an intelligent massage recommendation strategy, and to generate an action execution instruction according to the intelligent massage recommendation strategy;

[0031] Generating a massage action for a target part and a warming action for the target part according to the action execution instruction;

[0032] Human-computer interaction is performed based on the user status signal and a human-computer interaction instruction is generated, so as to adjust at least one of the massage action and the warming synergistic action based on the human-computer interaction instruction.

[0033] An electronic device, characterized in that the device comprises:

[0034] one or more processors;

[0035] A computer readable medium configured to store one or more programs,

[0036] When the one or more programs are executed by the one or more processors, the one or more processors implement the massage method.

[0037] A computer-readable medium, characterized in that a computer program is stored thereon, and the massage method is implemented when the program is executed by a processor.

[0038] In the massage system, massage method, electronic device, and computer-readable medium based on machine learning of the present application, the massage system based on machine learning has the following technical advantages compared to traditional massage devices:

[0039] (1) Achieve closed-loop feedback and improve user experience:

[0040] Traditional massage devices lack the ability to obtain users' real feelings, while this system uses a sensor group to monitor the user's status in real time to form a user status signal and upload it to the cloud server. The intelligent massage recommendation model adjusts the massage strategy in real time based on this data to ensure that the massage effect matches the user's actual needs, achieving a true closed-loop feedback and greatly improving the user experience.

[0041] (2) Intelligent recommendations to reduce manual operations:

[0042] Users do not need to manually adjust device parameters repeatedly to explore the best mode in a single use. The intelligent massage recommendation model uses machine learning algorithms to automatically generate personalized intelligent massage recommendation strategies based on the user's historical usage data, current status, and possible preference changes. This not only reduces the user's operating burden, but also makes the massage process more efficient and comfortable.

[0043] (3) Dynamic adaptation and continuous optimization:

[0044] The system has the ability to continuously learn and optimize. As users use it, the intelligent massage recommendation model on the cloud server can continuously learn user feedback and preferences, thereby providing more accurate and personalized massage solutions in the future. This dynamic adaptability ensures that the massage effect can be continuously optimized as the user's status changes.

[0045] (4) Comprehensive massage effect, enhanced experience:

[0046] The system not only provides vibration-based massage, but also integrates thermal enhancement actions. Through the actuators, it acts on the target parts at the same time, realizing the combination of physical massage and thermal therapy, effectively improving the comprehensive effect of massage and enhancing the user's comfort.

[0047] (5) Human-computer interaction to enhance user control:

[0048] The design of the human-computer interaction unit enables users to adjust the massage and warming actions when necessary, which not only retains a certain degree of control over the user, but also reduces the user's decision-making burden through intelligent recommendations. This combination of intelligent recommendations and user-independent adjustments not only meets personalized needs, but also enhances the user's sense of participation and satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Hereinafter, some specific embodiments of the present application will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:

[0050] Figure 1 This is a structural schematic diagram of a massage system based on machine learning according to an embodiment of the present application.

[0051] Figure 2 This is a flow chart of a massage method based on machine learning according to an embodiment of the present application.

[0052] Figure 3 is a schematic diagram of the structure of the electronic device of this embodiment;

[0053] Figure 4 Detailed description of the hardware structure of the electronic device of this embodiment. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the embodiments of the present application should fall within the scope of protection of the embodiments of the present application.

[0055] Figure 1 This is a schematic diagram of the structure of a massage system based on machine learning in an embodiment of the present application. Figure 1 As shown, the massage system based on machine learning includes: a cloud server and a front-end device, wherein the cloud server is communicatively connected with the front-end device; an intelligent massage recommendation model based on machine learning is deployed on the cloud server; the front-end device includes: a sensor group, an actuator, a data interface module, a human-computer interaction unit, and a main control chip, wherein the sensor group, the actuator, the data interface module, and the human-computer interaction unit are all connected with the main control chip;

[0056] The sensor group is used to detect the status of the user to generate a user status signal and send it to the main control chip;

[0057] The main control chip is used to encapsulate the user status signal to generate a user status data packet, and upload the user status data packet to the cloud server through the data interface module;

[0058] The intelligent massage recommendation model is used to parse the user status data packet to generate an intelligent massage recommendation strategy and send it to the main control chip, so that the main control chip generates an action execution instruction according to the intelligent massage recommendation strategy;

[0059] The execution mechanism is used to generate a massage action for the target part and a warming action for the target part according to the action execution instruction;

[0060] The human-computer interaction unit is at least used to enable the user to perform human-computer interaction with the front-end device based on the user status signal and generate human-computer interaction instructions, so as to adjust at least one of the massage action and the warming enhancement action of the actuator based on the human-computer interaction instructions.

[0061] Exemplarily, the execution mechanism may include a vibration motor and a heating device, wherein the vibration motor is used to perform a massage action on the target part, and the heating device performs a warming and synergistic action on the target part.

[0062] The data interface module is, for example but not limited to, a WIFI module.

[0063] In the massage system, massage method, electronic device, and computer-readable medium based on machine learning of the present application, the massage system based on machine learning has the following technical advantages compared to traditional massage devices:

[0064] (1) Achieve closed-loop feedback and improve user experience:

[0065] Traditional massage devices lack the ability to obtain users' real feelings, while this system uses a sensor group to monitor the user's status in real time to form a user status signal and upload it to the cloud server. The intelligent massage recommendation model adjusts the massage strategy in real time based on this data to ensure that the massage effect matches the user's actual needs, achieving a true closed-loop feedback and greatly improving the user experience.

[0066] (2) Intelligent recommendations to reduce manual operations:

[0067] Users do not need to manually adjust device parameters repeatedly to explore the best mode in a single use. The intelligent massage recommendation model uses machine learning algorithms to automatically generate personalized intelligent massage recommendation strategies based on the user's historical usage data, current status, and possible preference changes. This not only reduces the user's operating burden, but also makes the massage process more efficient and comfortable.

[0068] (3) Dynamic adaptation and continuous optimization:

[0069] The system has the ability to continuously learn and optimize. As users use it, the intelligent massage recommendation model on the cloud server can continuously learn user feedback and preferences, thereby providing more accurate and personalized massage solutions in the future. This dynamic adaptability ensures that the massage effect can be continuously optimized as the user's status changes.

[0070] (4) Comprehensive massage effect, enhanced experience:

[0071] The system not only provides vibration-based massage, but also integrates thermal enhancement actions. Through the actuators, it acts on the target parts at the same time, realizing the combination of physical massage and thermal therapy, effectively improving the comprehensive effect of massage and enhancing the user's comfort.

[0072] (5) Human-computer interaction to enhance user control:

[0073] The design of the human-computer interaction unit enables users to adjust the massage and warming actions when necessary, which not only retains a certain degree of control over the user, but also reduces the user's decision-making burden through intelligent recommendations. This combination of intelligent recommendations and user-independent adjustments not only meets personalized needs, but also enhances the user's sense of participation and satisfaction.

[0074] Optionally, the sensor group includes at least one of a temperature sensor, a vibration sensor, and a gloss sensor; the temperature sensor is used to monitor the temperature of the target part and generate a temperature monitoring signal; the vibration sensor is used to monitor the vibration of the target part and generate a vibration monitoring signal; the gloss sensor is used to monitor the gloss of the target part and generate a gloss monitoring signal; and the user status signal includes at least one of the temperature monitoring signal, the vibration monitoring signal, and the gloss monitoring signal.

[0075] Optionally, when the main control chip encapsulates the user status signal to generate a user status data packet, and uploads the user status data packet to the cloud server through the data interface module, the following steps are performed:

[0076] Preprocessing the user status signal to obtain a preprocessed user status signal;

[0077] Formatting the preprocessed user status signal to obtain a data packet with a standard data structure, and adding a timestamp to each data packet to obtain formatted user status data;

[0078] Encrypting the formatted user status data to obtain encrypted user status data;

[0079] compressing the encrypted user status data to obtain compressed user status data;

[0080] The compressed user status data is uploaded to the cloud server through the data interface module.

[0081] The process of processing user status signals on the main control chip and uploading them to the cloud server is managed in detail, including preprocessing, formatting, encryption and compression, which has the following technical benefits:

[0082] (1) Preprocessing to improve data quality:

[0083] The preprocessing step can filter out noise and outliers in the user status signal and improve the accuracy and reliability of the data. This helps the intelligent massage recommendation model to more accurately analyze the user status and generate more precise massage recommendation strategies.

[0084] (2) Formatting facilitates data analysis:

[0085] Formatting the preprocessed user status signal into a data packet with a standard data structure and adding a timestamp can facilitate the intelligent massage recommendation model on the cloud server to quickly parse and process the data. This improves the efficiency and accuracy of data processing.

[0086] (3) Encryption processing ensures data security:

[0087] Encrypting the formatted user status data can ensure the security of the data during transmission, which prevents the data from being maliciously intercepted or tampered with, and protects the privacy and rights of users.

[0088] (4) Compression processing reduces transmission overhead:

[0089] Compressing the encrypted user status data can reduce the volume of data and reduce bandwidth consumption and delay during transmission, which improves the efficiency and response speed of data transmission.

[0090] The following is an example code to illustrate how to pre-process, format, encrypt, and compress the user status signal:

[0091]

[0092]

[0093]

[0094] Optionally, when the intelligent massage recommendation model parses the user status data packet to generate an intelligent massage recommendation strategy, the following steps are performed:

[0095] Based on the unpacking module of the intelligent massage recommendation model, the user status data packet is decoded to obtain the original user status data;

[0096] Based on the unpacking module of the intelligent massage recommendation model, feature extraction is performed on the original user state data to obtain user feature data, and embedding processing is performed on the user feature data to obtain a user feature vector;

[0097] Based on the inference module in the intelligent massage recommendation model, inference is performed according to the user feature vector to generate an intelligent massage recommendation strategy.

[0098] To this end, the intelligent massage recommendation model parses the user status data packet to generate the above steps when the intelligent massage recommendation strategy is implemented, which has the following technical benefits:

[0099] (1) Accuracy of data decoding and feature extraction:

[0100] By decoding the user status data packet through the unpacking module of the intelligent massage recommendation model, it is possible to accurately restore the original user status data from the compressed and encrypted data. This step is the basis for subsequent feature extraction and reasoning, ensuring the integrity and accuracy of the data.

[0101] The feature extraction process can extract key features that affect the user's massage needs from the original data, such as the user's muscle tension, skin temperature, etc. These features are crucial for generating personalized massage strategies.

[0102] (2) Effective representation of user feature vectors:

[0103] The extracted user feature data is embedded and converted into a vector representation in a high-dimensional space (i.e., user feature vector). This representation method not only retains the original feature information, but also reflects the differences and similarities between users through the distance and similarity of the vectors, providing a more effective input for the subsequent reasoning process.

[0104] (3) Intelligent reasoning and personalized recommendation:

[0105] Based on the user feature vector, the reasoning module in the intelligent massage recommendation model can use machine learning algorithms for intelligent reasoning to generate massage recommendation strategies that meet the user's current status and needs. This personalized recommendation method not only improves the accuracy and effect of massage, but also improves user satisfaction and experience.

[0106] The reasoning module can also continuously optimize the recommendation strategy based on user feedback and historical data to achieve continuous personalization and optimization.

[0107] (3) Improve system efficiency and scalability:

[0108] Through modular design (unpacking module, feature extraction module, inference module), the system can process user status data more efficiently and generate massage recommendation strategies. This design approach also facilitates system expansion and maintenance, and provides convenience for future function upgrades and algorithm updates.

[0109] The following is a simplified example code to illustrate how the intelligent massage recommendation model parses the user status data packet and generates an intelligent massage recommendation strategy:

[0110]

[0111]

[0112]

[0113] Optionally, when the inference module in the intelligent massage recommendation model performs inference according to the user feature vector to generate an intelligent massage recommendation strategy, the following steps are performed:

[0114] The inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, and performs a multi-layer full-connection process on the hidden feature vector to generate an intelligent massage recommendation strategy, wherein the intelligent massage recommendation strategy defines at least one of a recommended massage type, massage intensity, massage time, and recommended temperature.

[0115] The above reasoning module generates intelligent massage recommendation strategies through multi-layer convolution and full connection processing, which has the following technical benefits:

[0116] (1) Improve reasoning accuracy and robustness:

[0117] Through cascaded multi-layer convolution operations, the inference module can extract hidden features in the user feature vector layer by layer. These hidden features often reflect the deep information and complex relationships in the user status data. Compared with single-layer or simple feature processing methods, multi-layer convolution can capture user features more comprehensively, thereby improving the accuracy of inference.

[0118] Multi-layer fully connected processing further integrates the features extracted by convolution operations, and generates the final intelligent massage recommendation strategy through nonlinear transformation and weight adjustment. This method enhances the robustness of the model and enables the model to better adapt to the needs of different users and different scenarios.

[0119] (2) Personalized and refined recommendations:

[0120] Through multi-layer convolution and fully connected processing, the inference module can carefully analyze the various dimensions in the user feature vector, such as muscle tension, skin temperature, etc., and generate personalized massage recommendation strategies based on these features. This personalized recommendation method not only improves the comfort of massage, but also enhances user satisfaction and loyalty.

[0121] The reasoning module can also continuously optimize the parameters of the convolutional and fully connected layers based on user feedback and historical data to achieve continuous personalized and refined recommendations.

[0122] (3) Improve system scalability and flexibility:

[0123] The design of multi-layer convolution and fully connected processing makes the inference module highly scalable and flexible. When it is necessary to add new user features or adjust the recommendation strategy, it is only necessary to fine-tune or expand the existing model without rebuilding the entire model.

[0124] This design approach also facilitates the migration and deployment of the model, allowing the intelligent massage recommendation system to run on different hardware platforms and operating systems to meet a wider range of application needs.

[0125] The following is a simplified example code to illustrate how the inference module generates intelligent massage recommendation strategies through multi-layer convolution and fully connected processing:

[0126]

[0127]

[0128] Optionally, when the inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, a multi-head attention analysis is performed on all the extracted hidden feature vectors to perform weighted processing on all the hidden feature vectors to obtain attention-weighted hidden feature vectors, so that when the hidden feature vector is subjected to multi-layer full-connection processing, the attention-weighted hidden feature vector is subjected to multi-layer full-connection processing.

[0129] How the above reasoning module performs weighted processing on hidden feature vectors through multi-head attention analysis and performs multi-layer full-connection processing to generate intelligent massage recommendation strategies has the following technical benefits:

[0130] (1) Enhanced feature representation capabilities:

[0131] Through multi-head attention analysis, the inference module can perform weighted processing on all the extracted hidden feature vectors, so that important features receive more attention and unimportant features are relatively suppressed. This mechanism helps the model capture key information in user features more accurately, thereby enhancing feature representation capabilities.

[0132] (2) Improve reasoning accuracy:

[0133] The hidden feature vector processed by attention weighting focuses more on the important parts of the user's features, which helps the model make more accurate decisions when generating intelligent massage recommendation strategies. By highlighting key features, the model can better understand the user's needs and preferences, thereby providing personalized massage recommendations.

[0134] (3) Improve model robustness:

[0135] Multi-head attention analysis can capture diverse information in user features by working in parallel with multiple independent attention heads. This diversity helps the model maintain stable performance when facing different users and different scenarios, and improves the robustness of the model.

[0136] (4) Promoting feature fusion and interaction:

[0137] The attention mechanism can not only perform weighted processing on a single feature vector, but also promote the fusion and interaction between different feature vectors. This helps the model generate a more comprehensive massage recommendation strategy based on comprehensive consideration of multiple user characteristics.

[0138] The following is a simplified example code to illustrate how the inference module performs weighted processing on the hidden feature vector through multi-head attention analysis and performs multi-layer full connection processing to generate an intelligent massage recommendation strategy:

[0139]

[0140]

[0141]

[0142] In the above code, the MultiHeadAttention module implements the multi-head attention mechanism, and the ReasoningModule integrates multi-layer convolution, multi-head attention, and multi-layer full connection processing. In this way, the model can more accurately capture the key information in user characteristics and generate personalized intelligent massage recommendation strategies.

[0143] Optionally, before the inference module performs multi-head attention analysis on all the extracted hidden feature vectors, it performs regularized distribution on all the hidden feature vectors so that all the hidden feature vectors are close to a preset prior distribution, so as to obtain regularized distributed hidden feature vectors, and perform multi-head attention analysis on the regularized distributed hidden feature vectors.

[0144] The above method of regularizing the distribution of hidden feature vectors before multi-head attention analysis has the following technical benefits:

[0145] (1) Improve model stability:

[0146] Regularizing the distribution of hidden feature vectors can make them more stably close to the pre-set prior distribution. This stability helps reduce the fluctuation of the model during training, thereby improving the convergence speed and final performance of the model.

[0147] (2) Enhance generalization ability:

[0148] Regularizing the distribution prevents the model from overfitting the training data, thereby enhancing the model's generalization ability. When the model is faced with new, unseen data, it can make inferences and predictions more accurately.

[0149] (3) Improve feature representation:

[0150] Through regularization, outliers and noise in hidden feature vectors can be effectively suppressed, making feature representation clearer and more accurate. This helps the model better capture key information in user features and generate more accurate recommendation strategies.

[0151] (4) Promoting the effect of multi-head attention mechanism:

[0152] The regularized hidden feature vector is more standardized, which is conducive to the multi-head attention mechanism to play a better role. The attention mechanism can more accurately calculate the correlation between feature vectors, thereby generating more reasonable attention weights.

[0153] The following is a simplified example code to illustrate how to regularize the hidden feature vector before multi-head attention analysis:

[0154]

[0155]

[0156] In the above code, the regularize_distribution function implements the regularized distribution of the hidden feature vector. This function accepts a tensor x as input and returns the regularized tensor. In the forward method of the inference module, this function is called to regularize the hidden feature vector before the multi-head attention analysis.

[0157] Optionally, based on the following objective function, the reasoning module performs regularized distribution on all hidden feature vectors:

[0158] Wherein, q(z|x) represents the distribution of the hidden feature vector before regularization, p(z) represents the preset prior distribution, which is the standard normal distribution, KL(q(z|x)||p(z)) represents the divergence term, which is used to control the distribution of the hidden feature vector to be regularized so that it is close to the prior distribution p(z), and obtain the hidden feature vector z of the regularized distribution, x represents the user state data, It measures the probability that z reconstructs the input data x. The expectation E is calculated on the distribution of the latent variable q(z|x), KL(q(z|x)||p(z)) represents the KL divergence between q(z|x) and p(z), which measures the difference between the two probability distributions, and p(x|z) represents the probability of reconstructing x given z.

[0159] To this end, based on the above objective function, an effective representation of the input data is learned by balancing the reconstruction quality of the data and the distribution regularization of the latent variables, so that the inference module regularizes the distribution of all hidden feature vectors, which can not only improve the performance of the model, but also increase the generalization ability of the model, the ability to generate new data, and improve the flexibility and interpretability of the model.

[0160] Figure 2 This is a flow chart of a massage method based on machine learning according to an embodiment of the present application. Figure 2As shown, it includes:

[0161] Detecting the user's status to generate a user status signal;

[0162] Encapsulating the user status signal to generate a user status data packet, and uploading the user status data packet to the cloud server through the data interface module;

[0163] The intelligent massage recommendation model is used to parse the user status data packet to generate an intelligent massage recommendation strategy, and to generate an action execution instruction according to the intelligent massage recommendation strategy;

[0164] Generating a massage action for a target part and a warming action for the target part according to the action execution instruction;

[0165] Human-computer interaction is performed based on the user status signal and a human-computer interaction instruction is generated, so as to adjust at least one of the massage action and the warming synergistic action based on the human-computer interaction instruction.

[0166] In the above method, the exemplary description of each step can be found in the above Figure 1 Instructions shown.

[0167] The present application also provides an electronic device, the electronic device comprising:

[0168] one or more processors;

[0169] A computer readable medium configured to store one or more programs,

[0170] When the one or more programs are executed by the one or more processors, the one or more processors implement the massage method as described in the embodiment of the present application.

[0171] For an exemplary description of each step in the above electronic device, please refer to the above Figure 1 Instructions shown.

[0172] A computer-readable medium stores a computer program, which, when executed by a processor, implements the massage method described in the embodiment of the present application.

[0173] For an exemplary description of each step in the above-mentioned computer readable medium, please refer to the above-mentioned Figure 1 Instructions shown.

[0174] Figure 3 Schematic diagram of the structure of the electronic device of this embodiment; the electronic device may include:

[0175] One or more processors 301;

[0176] The computer readable medium 302 may be configured to store one or more programs,

[0177] When the one or more programs are executed by the one or more processors, the one or more processors implement the massage method as described in the above embodiments.

[0178] Figure 4 is the hardware structure of the electronic device of this embodiment; Figure 4 As shown, the hardware structure of the electronic device may include: a processor 401, a communication interface 402, a computer readable medium 403 and a communication bus 404;

[0179] The processor 401, the communication interface 402, and the computer readable medium 403 communicate with each other via a communication bus 404;

[0180] Optionally, the communication interface 402 may be an interface of a communication module, such as an interface of a GSM module;

[0181] The processor 401 may be specifically configured to implement some or all of the steps in the massage method described in the above embodiment.

[0182] When the steps of the front-end device are implemented, the electronic device is equivalent to the front-end device, and when the steps of the cloud server are implemented, the electronic device is equivalent to the cloud server.

[0183] The processor 401 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0184] The computer readable medium 403 can be, but is not limited to, a random access memory medium (Random Access Memory, RAM), a read-only memory medium (Read Only Memory, ROM), a programmable read-only memory medium (Programmable Read-Only Memory, PROM), an erasable read-only memory medium (Erasable Programmable Read-Only Memory, EPROM), an electrically erasable read-only memory medium (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

[0185] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code configured to execute the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed. It should be noted that the computer-readable medium described in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access storage medium (RAM), a read-only storage medium (ROM), an erasable programmable read-only storage medium (EPROM or flash memory), an optical fiber, a portable compact disk read-only storage medium (CD-ROM), an optical storage medium, a magnetic storage medium, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, an apparatus or a device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program configured to be used by or in combination with an instruction execution system, an apparatus or a device. The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0186] Computer program code configured to perform the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0187] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of the code contains one or more executable instructions configured to implement the specified logical function. There are specific sequential relationships in the above-mentioned specific embodiments, but these sequential relationships are only exemplary. When the specific implementation is implemented, these steps may be less, more, or the execution order may be adjusted. That is, in some implementations as replacements, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0188] As another aspect, the present application further provides a computer-readable medium having a computer program stored thereon, which implements the method described in the above embodiment when the program is executed by a processor.

[0189] As another aspect, the present application further provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by the device, the device implements the methods, steps and logic block diagrams disclosed in the embodiments of the present application.

[0190] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A massage system based on machine learning, characterized in that: include: A cloud server and a front-end device, wherein the cloud server is in communication connection with the front-end device; an intelligent massage recommendation model based on machine learning is deployed on the cloud server; The front-end device includes: a sensor group, an actuator, a data interface module, a human-computer interaction unit, and a main control chip, wherein the sensor group, the actuator, the data interface module, and the human-computer interaction unit are all connected to the main control chip; The sensor group is used to detect the status of the user to generate a user status signal and send it to the main control chip; The main control chip is used to encapsulate the user status signal to generate a user status data packet, and upload the user status data packet to the cloud server through the data interface module; The intelligent massage recommendation model is used to parse the user status data packet to generate an intelligent massage recommendation strategy and send it to the main control chip, so that the main control chip generates an action execution instruction according to the intelligent massage recommendation strategy; The actuator is used to generate a massage action for the target part and a warming action for the target part according to the action execution instruction; The human-computer interaction unit is at least used to enable the user to perform human-computer interaction with the front-end device based on the user status signal and generate human-computer interaction instructions, so as to adjust at least one of the massage action and the warming enhancement action of the actuator based on the human-computer interaction instructions.

2. The massage system according to claim 1, characterized in that: The sensor group includes at least one of a temperature sensor, a vibration sensor, and a gloss sensor. The temperature sensor is used to monitor the temperature of the target part and generate a temperature monitoring signal. The vibration sensor is used to monitor the vibration of the target part and generate a vibration monitoring signal. The gloss sensor is used to monitor the gloss of the target part and generate a gloss monitoring signal. The user status signal includes at least one of the temperature monitoring signal, the vibration monitoring signal, and the gloss monitoring signal.

3. The massage system according to claim 1, characterized in that: When the main control chip encapsulates the user status signal to generate a user status data packet and uploads the user status data packet to the cloud server through the data interface module, the following steps are performed: Preprocessing the user status signal to obtain a preprocessed user status signal; Formatting the preprocessed user status signal to obtain a data packet with a standard data structure, and adding a timestamp to each data packet to obtain formatted user status data; Encrypting the formatted user status data to obtain encrypted user status data; compressing the encrypted user status data to obtain compressed user status data; The compressed user status data is uploaded to the cloud server through the data interface module.

4. The massage system according to claim 1, characterized in that: When the intelligent massage recommendation model parses the user status data packet to generate an intelligent massage recommendation strategy, the following steps are performed: Based on the unpacking module of the intelligent massage recommendation model, the user status data packet is decoded to obtain the original user status data; Based on the unpacking module of the intelligent massage recommendation model, feature extraction is performed on the original user state data to obtain user feature data, and embedding processing is performed on the user feature data to obtain a user feature vector; Based on the inference module in the intelligent massage recommendation model, inference is performed according to the user feature vector to generate an intelligent massage recommendation strategy.

5. The massage system according to claim 4, characterized in that: When the inference module in the intelligent massage recommendation model performs inference according to the user feature vector to generate an intelligent massage recommendation strategy, the following steps are performed: The inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, and performs a multi-layer full-connection process on the hidden feature vector to generate an intelligent massage recommendation strategy, wherein the intelligent massage recommendation strategy defines at least one of a recommended massage type, massage intensity, massage time, and recommended temperature.

6. The massage system according to claim 5, characterized in that: The inference module performs a cascaded multi-layer convolution operation on the user feature vector to extract a hidden feature vector therefrom, and performs a multi-head attention analysis on all the extracted hidden feature vectors to perform weighted processing on all the hidden feature vectors to obtain attention-weighted hidden feature vectors, so that when performing multi-layer full-connection processing on the hidden feature vector, multi-layer full-connection processing is performed on the attention-weighted hidden feature vector.

7. The massage system according to claim 6, characterized in that: Before the inference module performs multi-head attention analysis on all the extracted hidden feature vectors, it performs regularized distribution on all the hidden feature vectors so that all the hidden feature vectors are close to a preset prior distribution, so as to obtain the regularized distributed hidden feature vectors, and perform multi-head attention analysis on the regularized distributed hidden feature vectors.

8. A massage method based on machine learning, characterized in that: include: Detecting the user's status to generate a user status signal; Encapsulating the user status signal to generate a user status data packet, and uploading the user status data packet; Parsing the user status data packet to generate an intelligent massage recommendation strategy, and generating an action execution instruction according to the intelligent massage recommendation strategy; Generating a massage action for a target part and a warming action for the target part according to the action execution instruction; Human-computer interaction is performed based on the user status signal and a human-computer interaction instruction is generated, so as to adjust at least one of the massage action and the warming synergistic action based on the human-computer interaction instruction.

9. An electronic device, characterized in that: The device comprises: one or more processors; A computer readable medium configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the massage method as claimed in claim 8.

10. A computer-readable medium, characterized in that A computer program is stored thereon, and when the program is executed by a processor, the massage method as claimed in claim 8 is implemented.

Images