Method and device for constructing emotion classification model and air conditioner

Abstract

The invention relates to the technical field of physiological monitoring, and discloses a method for constructing an emotion classification model, which comprises the following steps: constructing a fitting network model which represents a mapping relationship between a WiFi original signal and a human physiological signal; obtaining emotion categories and WiFi signals of different users; and carrying out migration training by taking the WiFi signal as input, taking the emotion category as a label and taking the fitting network model as a reference model to obtain a classification network model for emotion classification. The method can reduce the interference of emotion classification recognition on the user emotion. The invention further discloses a device for constructing the emotion classification model and an air conditioner.

Description

Technical Field

[0001] The present application relates to the field of physiological monitoring technology, for example, to a method and device for building an emotion classification model, and an air conditioner. Background Art

[0002] At present, with the rapid development of science and technology, the functions of air conditioners are becoming more diversified. In addition to the traditional cooling, heating and dehumidification functions, air conditioners also have emotion recognition functions. Through emotion recognition, air conditioners can sense the user's emotional state and / or health status, and then adjust the air outlet temperature, wind speed or mode according to the emotional state and / or health status to provide the user with an environment that matches his or her emotional state and health status. Therefore, how to sense the user's emotions has become a technical problem that needs to be solved urgently.

[0003] In order to perceive the user's emotions, the relevant technology discloses a method for prompting people's emotions, including: obtaining historical body temperature information, historical environmental humidity and historical audio for multiple historical time periods, where the historical audio is audio containing the voice of the cared person in the historical time period; performing emotion recognition based on the multiple historical audios to obtain multiple historical emotion information, and the historical emotion information corresponds one-to-one with the historical audio; using multiple historical body temperature information, multiple historical environmental humidity and multiple historical emotion information to train a convolutional neural network to obtain an emotion analysis model, and the historical body temperature information corresponds one-to-one with the historical emotion information, and the historical environmental humidity corresponds one-to-one with the historical emotion information; the current body temperature of the cared person is converted into the historical body temperature information of the cared person. The temperature information and the current environmental humidity are input into the emotion analysis model to obtain the current emotional information of the cared person. The input emotion analysis model is obtained by machine learning training of multiple sets of training data. Each set of training data includes historical body temperature information, historical environmental humidity and historical emotional information. The historical body temperature information is the body temperature of the cared person in the historical period, the historical environmental humidity is the humidity of the environment in which the cared person is located in the historical period, and the historical emotional information is the emotional information of the cared person in the historical period. The emotional information at least includes happiness and sadness; when the current emotional information is sad, a prompt message is issued, and the prompt message is used to prompt that the environmental change will cause discomfort to the cared person. The temperature sensor is attached to multiple parts of the body of the cared person to obtain historical body temperature information, historical environmental humidity and historical audio for multiple historical periods, at least including: an acquisition step, obtaining the temperatures of multiple parts of the target historical period, and the temperatures of the multiple parts are the temperatures of multiple parts of the body of the cared person detected by the temperature sensor; a calculation step, weighted averaging the temperatures of the parts to obtain historical body temperature information; repeating the acquisition step and the calculation step at least once in sequence to obtain the historical body temperature information of all historical periods.

[0004] In the process of implementing the embodiments of the present disclosure, it is found that there are at least the following problems in the related art:

[0005] The related art requires the user to wear multiple physiological sensors (e.g., temperature sensors) to detect and obtain the user's physiological information. However, wearing physiological sensors will interfere with the user's activities and even emotions. Therefore, the human emotion recognition method adopted by the related art will have an adverse effect on the user's emotions while perceiving the user's emotions.

[0006] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present application, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0007] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical components or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0008] The embodiments of the present disclosure provide a method and device for constructing an emotion classification model, and an air conditioner to eliminate the influence of wearing physiological sensors on the user's activities or emotions, and reduce the interference of emotion classification and recognition on the user's emotions.

[0009] In some embodiments, the method includes: constructing a fitting network model, the fitting network model represents the mapping relationship between the original WiFi signal and the human physiological signal; obtaining emotion categories and WiFi signals of different users; performing migration training with the WiFi signal as input and the emotion category as label and the fitting network model as a benchmark model to obtain a classification network model for emotion classification.

[0010] In some embodiments, constructing a fitting network model includes: obtaining WiFi original signals and human physiological signals; using breathing signals and heartbeat signals as labels and using WiFi original signals as input, training the initial network model, and obtaining a trained fitting network model.

[0011] In some embodiments, the model parameters of the initial network model are configured in the following manner: the model optimizer of the initial network model is configured as an LBFGS optimizer; and / or, the loss function of the initial network model is configured as a linear function of an MSE loss function and a regularized loss function.

[0012] In some embodiments, a classification network model for emotion classification is obtained by performing migration training with WiFi signals as input and emotion categories as labels and using the fitted network model as a benchmark model, including: configuring a loss function and a model optimizer of the initial classification network model; performing migration training on the initial classification network model with WiFi signals as input and emotion categories as labels and using the fitted network model as a benchmark model to obtain a classification network model.

[0013] Optionally, obtaining the WiFi raw signal includes: obtaining the WiFi raw signal through a WiFi module. Obtaining the human physiological signal includes: during the experimental stage, the processor obtains the human physiological signal through a physiological sensor, and the physiological sensor is worn on the detection part of the experimental personnel.

[0014] Optionally, the initial network model is a KAN (Kolmogorov - Arnold Network) model.

[0015] In some embodiments, the loss function and model optimizer of the initial classification network model are configured as follows: the loss function of the initial classification network model is configured as a cross - entropy function; the model optimizer of the initial classification network model is configured as an AdamW optimizer.

[0016] In some embodiments, when using the fitting network model as the benchmark model for transfer training, the number of iterations of backpropagation and the number of iterations of backpropagation when training the initial network model satisfy the following number condition: αM1 < M2; where M1 represents the number of iterations of backpropagation when using the network model as the benchmark model for transfer training, M2 represents the number of iterations of backpropagation when training the initial network model, α represents the proportionality coefficient, and 1 < α ≤ 3.

[0017] In some embodiments, it further includes: after obtaining the classification network model for emotion classification, obtaining the current WiFi signal; inputting the current WiFi signal into the classification network model for model training to obtain a model output signal; determining the current emotional state of the user according to the model output signal; and controlling the air conditioner to execute a corresponding regulation mode according to the current emotional state.

[0018] In some embodiments, it further includes: after determining the current emotional state of the user, controlling the lighting device associated with the air conditioner to execute a corresponding lighting mode, and / or controlling the audio device associated with the air conditioner to execute a corresponding audio mode.

[0019] In some embodiments, the device includes a processor and a memory storing program instructions, and the processor is configured to execute the method for constructing an emotion classification model as described above when running the program instructions.

[0020] In some embodiments, the air conditioner includes: an air conditioner body; and the device for constructing an emotion classification model as described above, which is installed on the air conditioner body.

[0021] In some embodiments, a computer - readable storage medium stores program instructions, and when the program instructions are running, they are used to cause a computer to execute the method for constructing an emotion classification model as described above.

[0022] The method and device for constructing an emotion classification model and the air conditioner provided by the embodiments of the present disclosure can achieve the following technical effects:

[0023] After constructing the fitting network model, the embodiment of the present disclosure obtains the emotion categories and WiFi signals of different users, and uses the WiFi signal as input and the emotion category as a label, and performs migration training with the fitting network model as a benchmark model to obtain a classification network model for user emotion classification. Since the fitting network model represents the mapping relationship between the original WiFi signal and the human physiological signal, by performing migration training on the fitting network model using the WiFi signal as input and the emotion category as a label, a high-accuracy classification network model can be obtained, thereby improving the accuracy of user emotion recognition. At the same time, the embodiment of the present disclosure realizes emotion classification in a non-sensing detection manner through WiFi signals, which can eliminate the influence of wearing physiological sensors on the user's activities or emotions, and is conducive to reducing the interference of emotion classification and recognition on user emotions.

[0024] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] One or more embodiments are exemplarily described by corresponding drawings, which do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements, and the drawings do not constitute a scale limitation, and wherein:

[0026] Figure 1 is a schematic diagram of a method for constructing a sentiment classification model provided by an embodiment of the present disclosure;

[0027] Figure 2 is a schematic diagram of another method for constructing a sentiment classification model provided by an embodiment of the present disclosure;

[0028] Figure 3 is a schematic diagram of another method for constructing a sentiment classification model provided by an embodiment of the present disclosure;

[0029] Figure 4 is a schematic diagram of another method for constructing a sentiment classification model provided by an embodiment of the present disclosure;

[0030] Figure 5 is a schematic diagram of a device for constructing a sentiment classification model provided by an embodiment of the present disclosure;

[0031] Figure 6 It is a schematic diagram of an air conditioner provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0032] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0033] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0034] Unless otherwise stated, the term "plurality" means two or more.

[0035] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0036] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0037] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0038] In the embodiments of the present disclosure, smart home appliances refer to home appliances that are formed by introducing microprocessors, sensor technology, and network communication technology into home appliances. They have the characteristics of intelligent control, intelligent perception, and intelligent application. The operation process of smart home appliances often relies on the application and processing of modern technologies such as the Internet of Things, the Internet, and electronic chips. For example, smart home appliances can be connected to electronic devices to enable users to remotely control and manage the smart home appliances.

[0039] In the disclosed embodiments, the terminal device refers to an electronic device with a wireless connection function. The terminal device can communicate with the above-mentioned smart home appliances by connecting to the Internet, or can communicate with the above-mentioned smart home appliances directly through Bluetooth, WiFi (WIreless Fidelity), etc. In some embodiments, the terminal device is, for example, a mobile device, a computer, or a vehicle-mounted device built into a hover car, or any combination thereof. Mobile devices may include, for example, mobile phones, smart home devices, wearable devices, smart mobile devices, virtual reality devices, etc., or any combination thereof, wherein wearable devices include, for example: smart watches, smart bracelets, pedometers, etc.

[0040] Combination Figure 1 As shown, the embodiment of the present disclosure provides a method for constructing a sentiment classification model, comprising:

[0041] S01, the processor builds a fitting network model, and the fitting network model represents the mapping relationship between the WiFi original signal and the human physiological signal.

[0042] S02: The processor obtains emotion categories and WiFi signals of different users.

[0043] S03, the processor uses the WiFi signal as input and the emotion category as a label and the fitting network model as a benchmark model for migration training to obtain a classification network model for emotion classification.

[0044] The method for constructing an emotion classification model provided by the embodiment of the present disclosure is adopted. After the embodiment of the present disclosure constructs the fitting network model, the emotion categories and WiFi signals of different users are obtained, and the WiFi signals are used as inputs and the emotion categories are used as labels, and the fitting network model is used as a benchmark model for transfer training to obtain a classification network model for user emotion classification. Since the fitting network model represents the mapping relationship between the original WiFi signal and the human physiological signal, by using the WiFi signal as input and the emotion category as a label to perform transfer training on the fitting network model, a high-accuracy classification network model can be obtained, thereby improving the accuracy of user emotion recognition. At the same time, the embodiment of the present disclosure realizes emotion classification in a non-sensing detection manner through WiFi signals, which can eliminate the influence of wearing physiological sensors on the user's activities or emotions, and is conducive to reducing the interference of emotion classification and recognition on user emotions.

[0045] In addition, related technologies also recognize user emotions based on audio-visual technology. However, audio-visual technology requires the configuration of a camera to capture user images to obtain facial information, and requires the configuration of an audio acquisition device to collect user voices, which poses a risk of leaking user privacy in home scenarios. The disclosed embodiment obtains a classified network model through WiFi signals, without the need to capture user images and collect user voices, thereby reducing the risk of user privacy leakage in home scenarios. At the same time, the installation cost of installing cameras and audio acquisition devices is reduced.

[0046] Optionally, the processor constructs a fitting network model, including:

[0047] The processor obtains the original WiFi signal and the human physiological signal, wherein the human physiological signal includes a breathing signal and a heartbeat signal.

[0048] The processor uses the breathing signal and the heartbeat signal as labels and the original WiFi signal as input to perform model training on the initial network model to obtain a trained fitting network model.

[0049] In this way, both the breathing signal and the heartbeat signal will cause changes in the amplitude of chest wall motion. However, the change in the amplitude of chest wall motion of the breathing signal is greater than the change in the amplitude of chest wall motion of the heartbeat signal. The lower threshold of the frequency range of the heartbeat signal is higher than the upper threshold of the frequency range of the breathing signal. In addition, the frequency of the breathing signal is different under different emotions and the frequency of the heartbeat signal is different under different emotions. In addition, acceleration can characterize the influence of the heartbeat signal, and displacement can characterize the influence of the breathing signal. Normally, breathing is slow and stable, and because the heartbeat involves rapid contraction of muscles, the acceleration of breathing is significantly less than the acceleration of the heartbeat. Therefore, after obtaining the WiFi original signal and the human physiological signal, the embodiment of the present disclosure uses the breathing signal and the heartbeat signal as labels, and uses the WiFi original signal as input to perform model training on the initial network model to obtain a trained fitting network model.

[0050] Optionally, the processor obtains the WiFi original signal, including: the processor obtains the WiFi original signal through the WiFi module. The processor obtains the human physiological signal, including: during the experimental stage, the processor obtains the human physiological signal through a physiological sensor, and the physiological sensor is worn at the detection part of the experimenter. As an example, the physiological sensor is worn on the wrist or ankle of the experimenter.

[0051] Optionally, the initial network model is a KAN (Kolmogorov-Arnold Network) network model.

[0052] In this way, the core of the KAN network model is that any multivariate continuous function can be expressed as a combination of a finite number of single variable functions and addition. Compared with the MLP (Multilayer Perceptron) network model, the KAN network model is more efficient and interpretable. At the same time, the performance of the function fitting task far exceeds that of the MLP network model. Moreover, the KAN network model supports human-computer interaction, and a network model with higher performance can be trained by inputting prior data into it for model training. In addition, a single cycle of the respiratory signal and the heartbeat signal is similar to the signal of the Belset function. Based on the model characteristics of the KAN network model, it is suitable for fitting the respiratory signal and the heartbeat signal from the WiFi signal and performing emotion classification. Therefore, the embodiment of the present disclosure uses the respiratory signal and the heartbeat signal as labels and uses the WiFi original signal as input to train the KAN network model. The trained fitting network model obtained by model training can accurately and reliably represent the mapping relationship between the WiFi original signal and the human physiological signal.

[0053] Optionally, the processor configures model parameters of the initial network model in the following manner:

[0054] The processor configures the model optimizer of the initial network model to be a LBFGS (Limited-memory BFGS) optimizer. And / or, the processor configures the loss function of the initial network model to be a linear function of an MSE (mean-square error) loss function and a regularized loss function.

[0055] In this way, by configuring the LBFGS optimizer and the linear function of the MSE loss function and the regularized loss function, the MSE loss function and the regularized loss function are calculated in each round of model training, and the linear function of the MSE loss function and the regularized loss function are used as indicators to update the weights of the network model in the back propagation stage, which can improve the computational efficiency of model training and reduce memory usage.

[0056] Optionally, the loss function is a linear function of an MSE loss function and a regularized loss function, including:

[0057] The loss function is loss+10 -5 *r-loss. Among them, loss represents the MSE loss function, and r-loss represents the regularized loss function.

[0058] Optionally, combined Figure 2 As shown, the processor uses the WiFi signal as input and the emotion category as a label, and uses the fitted network model as a baseline model for migration training to obtain a classification network model for emotion classification, including:

[0059] S11, the processor configures the loss function and model optimizer of the initial classification network model.

[0060] S12, the processor uses the WiFi signal as input and the emotion category as a label and uses the fitting network model as a baseline model to perform migration training on the initial classification network model to obtain a classification network model.

[0061] In this step, the emotion categories include anger, disgust, fear, happiness, fatigue, anxiety, tension, sadness, and surprise.

[0062] In this way, after configuring the loss function and model optimizer of the initial classification network model, the embodiment of the present disclosure uses the fitting network model as the benchmark model to perform migration training on the initial classification network model, and can obtain a high-accuracy classification network model, thereby improving the accuracy of user emotion recognition. At the same time, the embodiment of the present disclosure implements emotion classification in a non-sensing detection manner through WiFi signals, which can eliminate the influence of wearing physiological sensors on the user's activities or emotions, and is conducive to reducing the interference of emotion classification and recognition on the user's emotions.

[0063] In practical applications, anger, fear, and tension are accompanied by accelerated breathing and heart rate, while happiness makes breathing and heart rate more stable, and surprise makes breathing and heart rate fluctuate significantly. The above emotion categories can be added to the last activation function of the KAN network model.

[0064] Optionally, the processor configures the loss function and model optimizer of the initial classification network model as follows:

[0065] The loss function of the processor configuration initial classification network model is the cross entropy function.

[0066] The processor configures the model optimizer of the initial classification network model to be the AdamW optimizer.

[0067] In this way, by configuring the cross entropy function as the loss function and the AdamW optimizer as the model optimizer, the cross entropy function is used as an indicator in each round of model training to update the weights of the network model in the back propagation stage, which can significantly improve the training speed of the model training and help improve the computational efficiency of model migration training.

[0068] Optionally, the number of iterations of back propagation when the processor performs migration training using the fitted network model as the benchmark model and the number of iterations of back propagation when performing model training on the initial network model meet the following number conditions: αM1 <M2。

[0069] Wherein, M1 represents the number of iterations of back propagation when performing migration training with the network model as the benchmark model, M2 represents the number of iterations of back propagation when performing model training on the initial network model, α represents the proportional coefficient, and 1<α≤3. Preferably, α is 2. As an example, M1 is 100 and M2 is 200.

[0070] This is conducive to obtaining a high-accuracy classification network model, thereby improving the accuracy of user emotion recognition.

[0071] Combination Figure 3 As shown, the embodiment of the present disclosure also provides a method for constructing a sentiment classification model, including:

[0072] S21, the processor builds a fitting network model, and the fitting network model represents the mapping relationship between the WiFi original signal and the human physiological signal.

[0073] S22, the processor obtains emotion categories and WiFi signals of different users.

[0074] S23, the processor performs migration training with the WiFi signal as input and the emotion category as label and the fitting network model as the benchmark model to obtain a classification network model for emotion classification.

[0075] S24, after the processor obtains the classification network model for emotion classification, it obtains the current WiFi signal.

[0076] S25, the processor inputs the current WiFi signal to the classification network model for model training to obtain a model output signal.

[0077] S26, the processor determines the user's current emotional state based on the model output signal.

[0078] S27, the processor controls the air conditioner to execute a corresponding control mode according to the current emotional state.

[0079] Using the method for constructing an emotion classification model provided by the embodiment of the present disclosure, the embodiment of the present disclosure can obtain a high-accuracy classification network model by using WiFi signals as input and emotion categories as labels to perform migration training on the fitting network model, thereby improving the accuracy of user emotion recognition. At the same time, the embodiment of the present disclosure realizes emotion classification in a non-sensing detection manner through WiFi signals, which can eliminate the influence of wearing physiological sensors on the user's activities or emotions, and is conducive to reducing the interference of emotion classification and recognition on the user's emotions. The embodiment of the present disclosure inputs the current WiFi signal to the classification network model for model training, and obtains the model output signal through forward propagation. Then determine the user's current emotional state based on the model output signal. Finally, according to the current emotional state, control the air conditioner to execute the corresponding control mode. In this way, after determining the current emotional state of the user through the classification network model, the embodiment of the present disclosure controls the air conditioner to perform targeted mode control to improve the comfort of the user in the home environment.

[0080] In a specific embodiment, the processor controls the air conditioner to execute a corresponding control mode according to the current emotional state, including:

[0081] When the current emotional state indicates tension or anxiety, the processor controls the air conditioner to execute a comfort mode; and / or controls the air conditioner to push a prompt message, the prompt message being used to prompt a scheduled rest.

[0082] When the current emotional state indicates fatigue, the processor controls the air conditioner to automatically switch to a soft speed mode or a sleep mode.

[0083] When the current emotional state indicates discomfort, the processor controls the air conditioner to enter the self-humidification mode.

[0084] It should be noted that the above embodiments are examples, and in actual applications, differentiated air conditioner control can be performed according to the user's emotions.

[0085] Combination Figure 4 As shown, the embodiment of the present disclosure also provides a method for constructing a sentiment classification model, including:

[0086] S31, the processor builds a fitting network model, and the fitting network model represents the mapping relationship between the WiFi original signal and the human physiological signal.

[0087] S32, the processor obtains emotion categories and WiFi signals of different users.

[0088] S33, the processor performs migration training with the WiFi signal as input and the emotion category as label and the fitting network model as the benchmark model to obtain a classification network model for emotion classification.

[0089] S34, after the processor obtains the classification network model for emotion classification, it obtains the current WiFi signal.

[0090] S35, the processor inputs the current WiFi signal to the classification network model for model training to obtain a model output signal.

[0091] S36, the processor determines the user's current emotional state based on the model output signal.

[0092] S37, the processor controls the air conditioner to execute a corresponding control mode according to the current emotional state.

[0093] S38, after determining the current emotional state of the user, the processor controls the lighting device associated with the air conditioner to execute a corresponding lighting mode, and / or controls the audio device associated with the air conditioner to execute a corresponding audio mode.

[0094] By adopting the method for constructing an emotion classification model provided by the embodiment of the present disclosure, the embodiment of the present disclosure can obtain a high-accuracy classification network model by using WiFi signals as input and emotion categories as labels to perform migration training on the fitting network model, thereby improving the accuracy of user emotion recognition. At the same time, the embodiment of the present disclosure realizes emotion classification in a non-sensing detection manner through WiFi signals, which can eliminate the influence of wearing physiological sensors on the user's activities or emotions, and is conducive to reducing the interference of emotion classification and recognition on the user's emotions. The embodiment of the present disclosure inputs the current WiFi signal to the classification network model for model training, and obtains the model output signal through forward propagation. Then determine the current emotional state of the user according to the model output signal. Finally, according to the current emotional state, control the air conditioner to execute the corresponding control mode and / or control the audio device associated with the air conditioner to execute the corresponding audio mode. In this way, after determining the current emotional state of the user through the classification network model, the embodiment of the present disclosure controls the air conditioner and other types of household appliances for targeted control, which can improve the comfort of the user's home environment in multiple aspects from the environment, lighting and sound.

[0095] Combination Figure 5 As shown, the embodiment of the present disclosure provides a device 200 for constructing an emotion classification model, including a processor 700 and a memory 701. Optionally, the device 70 may also include a communication interface 702 and a bus 703. The processor 700, the communication interface 702, and the memory 701 may communicate with each other through the bus 703. The communication interface 702 may be used for information transmission. The processor 700 may call the logic instructions in the memory 701 to execute the method for constructing an emotion classification model of the above embodiment.

[0096] In addition, the logic instructions in the memory 701 described above may be implemented in the form of software functional units and when sold or used as independent products, may be stored in a computer-readable storage medium.

[0097] The memory 701 is a computer-readable storage medium that can be used to store software programs and computer executable programs, such as program instructions / modules corresponding to the method in the embodiment of the present disclosure. The processor 700 executes the function application and data processing by running the program instructions / modules stored in the memory 701, that is, the method for building the emotion classification model in the above embodiment is implemented.

[0098] The memory 701 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory 701 may include a high-speed random access memory and may also include a non-volatile memory.

[0099] Combination Figure 6 As shown, an embodiment of the present disclosure provides an air conditioner 100, including: an air conditioner body, and the above-mentioned device 200 for constructing an emotion classification model. The device 200 for constructing an emotion classification model is installed on the air conditioner body. The installation relationship described here is not limited to placement inside the air conditioner body, but also includes installation connections with other components of the air conditioner 100, including but not limited to physical connections, electrical connections, or signal transmission connections. It can be understood by those skilled in the art that the device 200 for constructing an emotion classification model can be adapted to a feasible air conditioner body, thereby realizing other feasible embodiments.

[0100] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned method for constructing a sentiment classification model.

[0101] The technical solution of the embodiment of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiment of the present disclosure. The aforementioned storage medium may be a non-transient storage medium, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media that can store program codes.

[0102] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. The parts and features of some embodiments may be included in or replace the parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of listings containing one or more associated ones. In addition, when used in the present application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the presence of other identical elements in the process, method or device comprising the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0103] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods for each specific application to implement the described functions, but such implementations should not be considered to exceed the scope of the embodiments of the present disclosure. The technicians may clearly understand that, for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above may refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0104] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units can be only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to implement this embodiment. In addition, each functional unit in the embodiment of the present disclosure may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.

[0105] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A method for constructing an emotion classification model, characterized in that: include: Construct a fitting network model, which represents the mapping relationship between the original WiFi signal and the human physiological signal; Get the emotion categories and WiFi signals of different users; Using WiFi signals as input and emotion categories as labels, and taking the fitted network model as the baseline model, migration training is performed to obtain a classification network model for emotion classification.

2. The method according to claim 1, characterized in that Construct a fitting network model, including: Obtain WiFi original signals and human physiological signals; The breathing signal and the heartbeat signal are used as labels and the original WiFi signal is used as input to train the initial network model and obtain a trained fitting network model.

3. The method according to claim 2, characterized in that Configure the model parameters of the initial network model as follows: The model optimizer of the initial network model is configured as an LBFGS optimizer; and / or, The loss function of the initial network model is configured as a linear function of the MSE loss function and the regularization loss function.

4. The method according to claim 2, characterized in that: Using WiFi signals as input and emotion categories as labels, and the fitted network model as the baseline model for migration training, a classification network model for emotion classification is obtained, including: Configure the loss function and model optimizer of the initial classification network model; The WiFi signal is used as input and the emotion category is used as the label. The initial classification network model is transferred and trained using the fitted network model as the baseline model to obtain the classification network model.

5. The method according to claim 4, characterized in that Configure the loss function and model optimizer for the initial classification network model as follows: The loss function of the initial classification network model is configured as the cross entropy function; The model optimizer of the initial classification network model is configured as the AdamW optimizer.

6. The method according to claim 4, characterized in that The number of back propagation iterations when performing migration training using the fitted network model as the baseline model and the number of back propagation iterations when performing model training on the initial network model meet the following conditions: αM1 <M2; Among them, M1 represents the number of back propagation iterations when the network model is used as the baseline model for migration training, M2 represents the number of back propagation iterations when the initial network model is trained, α represents the proportional coefficient, and 1<α≤3.

7. The method according to any one of claims 1 to 6, characterized in that: Also includes: After obtaining the classification network model for emotion classification, obtain the current WiFi signal; Input the current WiFi signal to the classification network model for model training to obtain the model output signal; Determine the user's current emotional state based on the model output signal; According to the current emotional state, the air conditioner is controlled to execute the corresponding control mode.

8. The method according to claim 7, characterized in that Also includes: After determining the current emotional state of the user, the lighting device associated with the air conditioner is controlled to execute a corresponding lighting mode, and / or the audio device associated with the air conditioner is controlled to execute a corresponding audio mode.

9. A device for constructing an emotion classification model, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the method for constructing a sentiment classification model according to any one of claims 1 to 8 when running the program instructions.

10. An air conditioner, characterized in that: include: Air conditioner body; The device for constructing an emotion classification model as described in claim 9 is installed on the air conditioner body.

Images