Method, device, terminal device and storage medium for acquiring data processing model with privacy protection

Abstract

A privacy-protected data processing model acquisition method, device, terminal device, and storage medium are provided, the method including the steps of: acquiring sensor data of preset IoT devices; training an initial data model corresponding to each sensor using the sensor data to obtain an intermediate data model corresponding to each sensor, and aggregating the intermediate data models corresponding to each sensor to form an aggregated data model; processing new data using the aggregated data model and random noise to obtain a tag type corresponding to the new data; and training the aggregated data model based on the new data and the tag type of the new data to obtain a data model. The privacy-protected data processing model acquisition method provided by the present disclosure is a method of building multiple models for multiple sensor data, integrating all the models, and adding noise disturbance, thereby solving the technical problem of low privacy protection for Internet of Things data.

Description

[Technical field]

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a method, apparatus, terminal device and storage medium for obtaining a data processing model with privacy protection. [Background technology]

[0002] At present, the Internet of Things has become an important component of the Internet, and the number of global Internet of Things devices is growing rapidly. In the future Internet of Things, any individual, group, community, organization, object, product, data, service, process and activity will be interconnected by the Internet of Things. Personal data privacy and trade secrets in the Internet of Things have become important shortcomings that limit the development of the Internet of Things. At present, research and solutions on the security and privacy technology of the Internet of Things have not played a good role in ensuring the security of the Internet of Things, and the potential security risks of the Internet of Things have caused more and more concerns. Traditional centralized deep learning requires users to submit data to a data center and then uniformly train it through a cloud server. However, these data may be misused by the model trainer to infer more relevant user privacy information. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method, apparatus, terminal device and computer-readable storage medium for obtaining a data processing model with privacy protection, and aim to solve the technical problem of low privacy protection of Internet of Things data.

[0004] A first aspect of an embodiment of the present invention provides a method for obtaining a privacy-preserving data processing model, the method comprising: acquiring sensor data from a plurality of sensors of a preset IoT device; training corresponding initial data models to obtain intermediate data models corresponding to each of the sensors, and aggregating the intermediate data models corresponding to each of the sensors to form an aggregate data model; processing new data with the aggregate data model and random noise to obtain a tag type corresponding to the new data; training the aggregate data model based on the new data and a tag type of the new data to obtain a data model.

[0005] A second aspect of the present invention provides an apparatus for obtaining a privacy-preserving data processing model, the apparatus comprising: A data collection module for acquiring sensor data from a plurality of sensors of a preset IoT device; a data integration module that trains an initial data model corresponding to each of the sensors with sensor data corresponding to the sensors to obtain an intermediate data model corresponding to each of the sensors, and aggregates the intermediate data models corresponding to each of the sensors to form an aggregate data model; a noise adding module for processing new data according to the aggregate data model and random noise to obtain a tag type corresponding to the new data; and a data processing module for training the collective data model based on the new data and a tag type of the new data to obtain a data model.

[0006] A third aspect of an embodiment of the present invention provides a terminal device, comprising a memory, a processor and a computer program stored in the memory and executable by the processor, the terminal device realizing steps of the method for obtaining a data processing model with privacy protection when the processor executes the computer program.

[0007] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored therein, the computer program being capable of implementing steps of the method for obtaining a privacy-preserving data processing model when executed by a processor.

[0008] The beneficial effects of the privacy-preserving data processing model acquisition method provided by the embodiment of the present invention are at least as follows: The embodiment of the present invention first acquires sensor data, trains an initial model using the sensor data, obtains a corresponding intermediate data model, and aggregates to form an aggregated data model, then processes new data using the aggregated data model and random noise to obtain a tag type corresponding to the new data, and finally uses the new data and the corresponding tag type to train the aggregated data model and obtain a data model, so that multiple participants can build a machine learning model without the need to share data. The present invention is a method of building multiple models for multiple sensor data, integrating all the models and adding noise disturbance, and finally obtaining one comprehensive result, thereby obtaining a higher representation effect than a single sensor model. By adding noise to the model and introducing a certain degree of uncertainty, the contribution of any individual user, any IoT individual can be sufficiently masked, and the privacy of the participants can be protected, thereby solving the technical problem of low privacy protection of Internet of Things data. [Brief description of the drawings]

[0009] In order to more clearly describe the technical solutions in the embodiments of the present invention, the following briefly introduces drawings required for describing the embodiments or prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without the need for creative work.

[0010] [Figure 1]1 is a flowchart illustrating an implementation of a privacy-preserving data processing model acquisition method provided by an embodiment of the present invention; [Diagram 2] 3 is a flowchart illustrating an implementation of a collective data model and random noise processing for processing new data and obtaining a tag type corresponding to the new data according to an embodiment of the present invention; [Diagram 3] 4 is an implementation flowchart of training a collective data model and obtaining a data model according to new data and a tag type of the new data provided by an embodiment of the present invention; [Figure 4] 1 is a flowchart illustrating an implementation of a method and apparatus for obtaining a data processing model with privacy protection provided by an embodiment of the present invention. [Diagram 5] FIG. 2 is a schematic diagram of a terminal device provided by an embodiment of the present invention; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] In the following description, specific details such as specific system structures, techniques, etc. are provided for the purpose of explanation, not limitation, to thoroughly understand the embodiments of the present invention. However, those skilled in the art should understand that the present invention can be realized in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from obscuring the description of the present invention. All other embodiments obtained by those skilled in the art from the described embodiments of the present invention are within the scope of protection of the present invention. Unless otherwise specified, the technical means used in the embodiments are common means well known by those skilled in the art.

[0012] As used in this specification and the appended claims, the term "comprising" refers to the presence of stated features, wholes, steps, operations, elements and / or assemblies, but it should be understood not to exclude the presence or addition of one or more other features, wholes, steps, operations, elements, assemblies and / or combinations thereof.

[0013] Furthermore, the terminology used herein in the specification of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used in the specification of the present application and the appended claims, the singular forms "a," "one," and "the" should also be understood to include the plural forms unless the context clearly dictates otherwise.

[0014] Furthermore, the term "and / or" as used in the specification and appended claims of this application should be understood to refer to and include any and all possible combinations of one or more of the associated listed items.

[0015] As used herein and in the appended claims, the term "upon" can be interpreted as "when" or "once," or "responsive to determining" or "responsive to detecting," depending on the context. Similarly, the phrase "upon determining" or "upon detecting [a described condition or event]" can be interpreted to mean "once determining" or "responsive to determining," or "upon detecting [a described condition or event]," depending on the context. First embodiment

[0016] FIG. 1 is a flowchart illustrating an implementation of a privacy-preserving data processing model acquisition method provided in an embodiment of the present invention.

[0017] As shown in FIG. 1, the method for obtaining the data processing model includes the following steps:

[0018] Step S10: Sensor data from a plurality of sensors of the preset IoT device is acquired.

[0019] In this embodiment, the multiple sensors of the preset IoT device are the same type of sensors, and the preset IoT device may be a gas turbine, an electric meter, a boiler, or other equipment. The sensors include a temperature sensor, a smoke sensor, a heart rate sensor, a distance sensor, a light sensor, a sound sensor, a gas sensor, a chemical sensor, a pressure sensor, a fluid sensor, or other types of sensors.

[0020] In this embodiment, the sensor data is from the same type and model of IoT devices, such as a uniform model gas turbine, electric meter, boiler or other devices. When acquiring sensor data, due to sensor failure, voltage, electrical failure or other unforeseen reasons, not every sensor can collect complete data, and if there is a choice, preferably multiple relatively complete data.

[0021] Step S20: Train an initial data model corresponding to each of the sensors using sensor data corresponding to the sensors, obtain an intermediate data model corresponding to each of the sensors, and aggregate the intermediate data models corresponding to each of the sensors to form an aggregate data model.

[0022] In this embodiment, the multiple sensors may employ the same initial data model, or the multiple sensors may employ initial data models of the same type but different hyperparameters, or the initial data models of at least two of the multiple sensors are different.

[0023] In this embodiment, by building multiple models on multiple sensor data, multiple participants can build machine learning models and do not need to share data, thereby achieving greater effectiveness than a single sensor model.

[0024] In this embodiment, the initial data model includes a neural network model, a support vector machine, a logistic regression model, a decision tree model, or other data models.

[0025] Step S30: Process new data according to the aggregate data model and random noise to obtain a tag type corresponding to the new data.

[0026] In this embodiment, new data is processed with random noise to introduce a degree of uncertainty, thereby sufficiently masking the contributions of any individual user or any IoT individual and protecting the privacy of participants.

[0027] Step S40: Train the collective data model according to the new data and the tag type of the new data to obtain a data model.

[0028] In this embodiment, the aggregate data model is trained based on the new data and the tag type of the new data, so as to improve the performance effect of the model when ensuring data security.

[0029] FIG. 2 is an implementation flowchart of processing new data according to the aggregate data model and random noise provided by an embodiment of the present invention to obtain a tag type corresponding to the new data.

[0030] In this embodiment, in order to obtain the tag type corresponding to the new data, step S30 specifically includes four aspects.

[0031] Step S301: New data is input to each intermediate data model in the aggregate data model, and the tag type output from each intermediate data model is obtained.

[0032] In this embodiment, the new data is not sensor data and is generally insensitive data, such as public data sets or other relatively insensitive data.

[0033] Step S302: Determine the initial vote number for each tag type based on the tag type.

[0034] Step S303: In addition to the initial vote count, random noise is introduced to obtain the vote count for each tag type.

[0035] In this embodiment, the random noise includes Laplacian noise, which is Lap(1 / ε), where ε is the privacy cost, and its density function is JPEG0007635219000001.jpg20160, where x is the new data and b is 1 / ε.

[0036] Preferably, the maximum number of votes for a tag type is: JPEG0007635219000002.jpg22168, where x is the new data, j is the tag type, JPEG0007635219000003.jpg19166 is the initial vote count for new data x to belong to tag type j.

[0037] For example, if there are 10 sensor datasets, each dataset trains one model, and the output tags for input x of models 1 to 10 are (2, 3, 2, 1, 2, 3, 2, 4, 1, 2) and the tag types are 1 to 5, the voting results for tag types (1, 2, 3, 4, 5) of the 10 models are JPEG0007635219000004.jpg21163.

[0038] When Laplacian noise with ε=0.1 is added to the voting result (2,5,2,1,0), the voting result changes from (2,5,2,1,0) to (2.20,4.89,1.82,1.07,0.05), and the tag type to which x was input after adding the noise disturbance is still 2 (the number of votes corresponding to tag 2 is the highest at 4.89).

[0039] Step S304: Based on the number of votes for each tag type, the tag type for which the number of votes satisfies a predetermined condition is determined as the tag type of the new data.

[0040] FIG. 3 is an implementation flowchart of training a collective data model and obtaining a data model based on new data and the tag type of the new data provided by an embodiment of the present invention.

[0041] As shown in FIG. 3, step S40 specifically includes four aspects.

[0042] Step S401: Based on a predetermined rule, new data and the tag type of the new data are adopted to train an aggregate data model.

[0043] In this embodiment, the predetermined rules include model accuracy or cycle count.

[0044] Step S402: After reaching a predetermined rule, stop the training and obtain the target data model.

[0045] Step S403: Discard the sensor data and the integrated data model.

[0046] Step S404: The target data model is distributed.

[0047] In this embodiment, the distributed target model is separated from the original sensor data, so that even if an attacker obtains the model parameters of the distributed model, he or she cannot infer the original sensor data, thereby playing a role in protecting our original sensor data.

[0048] In the embodiment of the present invention, sensor data is first obtained, and then an initial model is trained using the sensor data to obtain a corresponding intermediate data model, and then the data is aggregated to form an aggregate data model, and then new data is processed using the aggregate data model and random noise to obtain a tag type corresponding to the new data, and finally the new data and the corresponding tag type are used to train the aggregate data model and obtain a data model, so that multiple participants can build a machine learning model without the need to share data. The present invention is a method of building multiple models for multiple sensor data, integrating all the models and adding noise disturbance, and finally obtaining one comprehensive result, thereby obtaining a higher representation effect than a single sensor model. By adding noise to the model and introducing a certain degree of uncertainty, the contribution of any individual user, any IoT individual can be sufficiently masked, and the privacy of the participants can be protected, thereby solving the technical problem of low privacy protection of Internet of Things data.

[0049] It should be understood that the magnitude of the numbers of each step in the above embodiments does not indicate the order of execution, and the execution order of each process should be determined by its function and inherent logic, and does not arbitrarily limit the implementation process of the embodiments of the present invention. Second embodiment

[0050] Based on the same inventive idea as the method in the first embodiment, accordingly, this embodiment further provides an apparatus for obtaining a data processing model with privacy protection.

[0051] As shown in FIG. 4 , in this embodiment, the privacy-preserving data processing model acquisition device includes a data collection module 41, a data integration module 42, a noise addition module 43 and a data processing module 44.

[0052] Here, the data collection module 41 is for acquiring sensor data from a plurality of sensors of the preset IoT devices.

[0053] The data integration module 42 is for training an initial data model corresponding to each of the sensors with sensor data corresponding to the sensors, obtaining an intermediate data model corresponding to each of the sensors, and aggregating the intermediate data models corresponding to each of the sensors to form an aggregate data model.

[0054] The noise adding module 43 is for processing new data according to the aggregate data model and random noise, and obtaining a tag type corresponding to the new data.

[0055] The data processing module 44 is for training the collective data model based on the new data and the tag type of the new data to obtain a data model.

[0056] In one embodiment, the noise adding module 43 includes: a data acquiring module for inputting new data into each intermediate data model in the collective data model respectively and acquiring a tag type output from each intermediate data model; a first determination module for determining an initial vote number for each tag type based on the tag type; a noise adding sub-module for introducing random noise in addition to the initial vote number and acquiring a vote number for each tag type; and a second determination module for determining, based on the vote number for each tag type, the tag type whose vote number satisfies a predetermined condition as the tag type of the new data.

[0057] In one embodiment, the data processing module 44 includes a data training module for training an aggregate data model using new data and tag types of the new data based on a predetermined rule, a model acquisition module for stopping the training and acquiring a target data model after the predetermined rule is reached, a discard module for discarding the sensor data and the integrated data model, and a distribution model module for distributing the target data model.

[0058] In some embodiments, the functions possessed by or modules included in the apparatus provided in the embodiments of the present disclosure can be used to perform the methods described in the above method embodiments, and for specific implementations thereof, reference can be made to the descriptions of the above method embodiments, and for the sake of brevity, the descriptions will be omitted here. Third embodiment

[0059] The above method and apparatus can be applied to terminal equipment such as desktop computers, notebook computers, palmtop computers and cloud servers.

[0060] Fig. 5 is a schematic diagram of a terminal device provided by an embodiment of the present invention. As shown in Fig. 5, the terminal device 5 includes a memory 51, a processor 50, and a computer program 52 stored in the memory 51 and executable by the processor 50, and when the processor 50 executes the computer program 52, steps such as the data simulation method of the IoT device are realized. For example, functions of modules 41 to 44 are shown in Fig. 4.

[0061] The terminal device 5 may be a computing device such as a desktop computer, a laptop, a palmtop computer, a cloud server, etc. The terminal device includes, but is not limited to, a processor 50, and the memory 51. As can be understood by those skilled in the art, FIG. 5 is merely an example of the terminal device 5, and is not intended to limit the terminal device 5, which may include more or less components than those shown, or may combine certain components or different components, for example, the terminal device may include input / output devices, network access devices, buses, etc.

[0062] The processor 50 may be a Central Processing Unit (CPU) or other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc. The general purpose processor may be a microprocessor, any common processor, etc.

[0063] The memory 51 may be an internal storage unit of the terminal equipment 5, such as a hard disk or RAM of the terminal equipment 5. The memory 51 may be an external storage device of the terminal equipment 5, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc., arranged in the terminal equipment 5. Furthermore, the memory 51 may not only include an internal storage unit of the terminal equipment 5, but also an external storage device. The memory 51 is for storing the computer program and other programs and data required by the terminal equipment. The memory 51 may also be used to temporarily store data that has been output or is to be output.

[0064] The integrated module / unit may be realized in the form of a software functional unit and stored in a computer-readable storage medium when sold or used as an independent product. With this understanding, the present application realizes all or part of the flow of the method of the above embodiment by commanding related hardware to complete it by a computer program, and the computer program may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above embodiment of the method may be realized. Here, the computer program includes a computer program code, and the computer program code may be in a source code form, an object code form, an executable file, or some intermediate form, etc. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier wave signal, an electric communication signal, and a software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be increased or decreased as required by the legislation and patent practice of a jurisdiction, e.g., in some jurisdictions, based on the legislation and patent practice, computer-readable medium does not include electrical carrier signals and telecommunications signals.

[0065] Specifically, the present embodiment further provides a computer-readable storage medium, which may be the computer-readable storage medium included in the memory in the above embodiment, or may be a computer-readable storage medium that exists independently and is not assembled into a terminal device. One or more computer programs are stored in the computer-readable storage medium.

[0066] A computer-readable storage medium, comprising a computer program stored in the computer-readable storage medium, the computer program including a step of implementing a data simulation method for the IoT device when executed by a processor.

[0067] For convenience and simplicity of explanation, only the division of each of the above functional units and modules has been described as an example, but in actual application, the above functions can be assigned to different functional units and modules as necessary, that is, all or part of the above-described functions can be completed by dividing the internal structure of the device into different functional units or modules. Each functional unit and module in the embodiment may be integrated into one processing unit, each unit may exist physically alone, two or more units may be integrated into one unit, and the integrated unit may be realized in the form of hardware or software functional unit. Note that the specific names of each functional unit and module are provided for easy distinction from each other, and do not limit the scope of protection of the present application. For the specific operation process of the units and modules in the above system, reference may be made to the corresponding process in the embodiment of the above method, and the description will be omitted here.

[0068] In the above embodiments, the description of each embodiment has its own emphasis, and for the details or parts not described in an embodiment, reference may be made to the relevant descriptions of other embodiments.

[0069] Those skilled in the art can be aware that each example unit and algorithm step described in the embodiments disclosed herein can be realized in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software is determined by the specific application and design constraints of the technical solution. Those skilled in the art may implement the described functions using different methods for each specific application, but such implementation should not be considered to go beyond the scope of the present invention.

[0070] In the embodiments of the present invention, it should be understood that the disclosed apparatus / terminal equipment and method can be realized in other ways. For example, the above-described embodiments of the apparatus / terminal equipment are merely exemplary, and the division into modules or units is merely a logical functional division, and other division methods may be applied in actual implementation, for example, multiple units or components may be combined or integrated into other systems, or some features may be ignored or not implemented. In addition, the couplings or direct couplings or communication connections between the shown or discussed mutually may be indirect couplings or communication connections by some interfaces, devices or units, which may be electrical, mechanical or other forms.

[0071] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed among multiple network units. Some or all of the units may be selected according to actual needs to achieve the objective of the solution of the present embodiment.

[0072] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically alone, or two or more units may be integrated into one unit. The integrated unit may be realized in the form of hardware or software functional unit.

[0073] The integrated module / unit may be realized in the form of a software functional unit and stored in a computer-readable storage medium when sold or used as an independent product. With this understanding, the present invention can achieve all or part of the processes in the above-mentioned embodiment methods by instructing related hardware through a computer program, and the above-mentioned computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned embodiment methods can be realized. Here, the computer program includes a computer program code, and the computer program code may be in a source code form, an object code form, an executable file, or some intermediate form, etc. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a U-disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier wave signal, an electric communication signal, and a software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be increased or decreased as required by the legislation and patent practice of a jurisdiction, e.g., in some jurisdictions, based on the legislation and patent practice, computer-readable medium does not include electrical carrier signals and telecommunications signals.

[0074] The above embodiments are only for illustrating the technical solutions of the present invention, and are not intended to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand the following: It is still possible to amend the technical solutions described in the above embodiments, or to equivalently replace some technical features therein. These amendments or replacements ensure that the substance of the corresponding technical solutions does not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and all of them should be included in the protection scope of the present invention.

Claims

1. acquiring sensor data of a plurality of sensors of a preset IoT device; training an initial data model corresponding to each of the sensors with sensor data corresponding to the sensors to obtain an intermediate data model corresponding to each of the sensors, and aggregating the intermediate data models corresponding to each of the sensors to form an aggregate data model; processing new data with the aggregate data model and random noise to obtain a tag type corresponding to the new data; training the aggregate data model based on the new data and a tag type of the new data to obtain a data model. A method for obtaining a data processing model with privacy protection, comprising:

2. The multiple sensors of the preset IoT devices are sensors of the same type, and the sensors include one of a temperature sensor, a smoke sensor, a heart rate sensor, a distance sensor, a light sensor, a sound sensor, a gas sensor, a chemical sensor, a pressure sensor, and a fluid sensor. The method for obtaining a data processing model with privacy protection according to claim 1 .

3. the initial data models of the sensors are the same; Alternatively, the multiple sensors employ initial data models of the same type but with different hyperparameters; Alternatively, the initial data models of at least two of the plurality of sensors are different. The method for obtaining a data processing model with privacy protection according to claim 1 .

4. The initial data model includes a neural network model, a support vector machine, a logistic regression model, or a decision tree model. The method for obtaining a data processing model with privacy protection according to claim 1 .

5. the step of processing new data with the aggregate data model and random noise to obtain a tag type corresponding to the new data includes: inputting new data into each intermediate data model in the aggregate data model and acquiring a tag type output from each intermediate data model; determining an initial vote number for each of the tag types based on the tag types; In addition to the initial vote count, random noise is introduced to obtain a vote count for each of the tag types; determining, based on the number of votes for each of the tag types, a tag type whose number of votes satisfies a predetermined condition as the tag type of the new data. The method for obtaining a data processing model with privacy protection according to claim 1 .

6. The random noise includes Laplacian noise, which is Lap(1 / ε), where ε is the privacy cost, and its density function is and where x is the new data and b = 1 / ε The method for obtaining a data processing model with privacy protection according to claim 5 .

7. The maximum number of votes for the tag type is: and where x is the new data, j is the tag type, is the initial vote number that the new data x belongs to the tag type j The privacy-preserving method for obtaining a data model according to claim 6 .

8. a data collection module for acquiring sensor data of a plurality of sensors of the preset IoT device; a data integration module for training an initial data model corresponding to each of the sensors with sensor data corresponding to the sensors to obtain an intermediate data model corresponding to each of the sensors, and for aggregating the intermediate data models corresponding to each of the sensors to form an aggregate data model; a noise adding module for processing new data with the aggregate data model and random noise to obtain a tag type corresponding to the new data; and a data processing module for training the aggregate data model based on the new data and a tag type of the new data to obtain a data model. A privacy-preserving data processing model acquisition device.

9. 1. A terminal device including a memory, a processor, and a computer program stored in the memory and executable by the processor, comprising: When the processor executes the computer program, the method for obtaining a privacy-preserving data processing model according to claim 1 is performed.

1. A terminal device comprising:

10. A computer-readable storage medium on which a computer program is stored, The computer program, when executed by a processor, performs the steps of the method for obtaining a privacy-preserving data processing model according to claim 1. A computer-readable storage medium comprising:

Images