WO2021159685A1

WO2021159685A1 - Task processing method, system, and device, and medium

Info

Publication number: WO2021159685A1
Application number: PCT/CN2020/110469
Authority: WO
Inventors: 周曦; 姚志强
Original assignee: 云从科技集团股份有限公司
Priority date: 2020-02-14
Filing date: 2020-08-21
Publication date: 2021-08-19
Also published as: CN111339553A

Abstract

The present invention provides a task processing method, system, and device, and a medium. The method comprises: each task processing terminal separately obtains a corresponding decomposed training task; each task processing terminal executes the decomposed training task according to local training data and random numbers or parameters obtained from other task processing terminals. After one or more participants initiate training tasks, the present invention can decompose all the training tasks by means of the task processing terminals, and according to the decomposed training tasks, each task processing terminal executes the decomposed training tasks on the basis of the local training data and the random numbers and parameters obtained from other task processing terminals. That is, the present method can aggregate or assist in aggregating the data information of multiple parties and protect the privacy data of the multiple parties in the scene that multiple data providers participate and all the participants do not trust each other.

Description

A task processing method, system, equipment and medium

Technical field

The present invention relates to the field of data processing technology, in particular to a task processing method, system, equipment and medium.

Background technique

Some companies or institutions hold some data, which may include private data and non-private data. For some companies or institutions, they hope to use these data for analysis, evaluation, etc. For example, for financial credit institutions, it may be used to evaluate corporate qualifications, corporate operating status, loan risks, etc. However, these data may belong to the sensitive private data of enterprises or institutions, and it is inconvenient for most enterprises or institutions to directly share with financial institutions, the government or the public. Therefore, how to realize data sharing while protecting data privacy is an urgent problem to be solved.

Summary of the invention

In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a task processing method, system, equipment and medium to solve the problems existing in the prior art.

In order to achieve the above objectives and other related objectives, the present invention provides a task processing method, which includes the following steps:

Each task processing terminal obtains the corresponding decomposed training task;

Each task processing terminal executes the decomposed training task based on local training data and random numbers or parameters obtained from other task processing terminals.

Optionally, before each task processing terminal separately obtains the corresponding decomposed training task, it further includes:

After receiving the training task request transmitted by the task processing terminal, the task processing platform decomposes the training task, obtains the decomposed training task, and allocates the decomposed training task to the corresponding task processing terminal.

Optionally, after performing the decomposed training task, it further includes: outputting a shared model.

Optionally, after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains random numbers or encryption parameters from other task processing terminals, and based on the random numbers or encryption parameters, decrypts encrypted data transmitted from other task processing terminals to obtain decrypted data;

Learning and training are performed through the local training data and the decrypted data.

Each task processing terminal obtains model training process random numbers or model training process parameters from other task processing terminals;

According to the model training process random number or model training process parameters and local training data, learning and training are performed.

Each task processing terminal obtains training parameters from other task processing terminals;

Perform learning training based on the training parameters and local training data.

Optionally, the training parameters include: the number of layers of the convolutional neural network and the size of the convolution kernel.

Optionally, the number of times that each task processing terminal obtains random numbers or parameters from other task processing terminals is one or more times.

Optionally, the learning algorithm used in the learning training includes at least one of the following: linear regression, logistic regression, tree model, deep neural network, graph neural network.

Optionally, the training data includes at least one of the following: social security data, provident fund data, fixed asset data, and current asset data.

Optionally, the current asset data includes at least one of the following: deposit data and loan data.

Optionally, the encrypted data is non-private data.

The present invention also provides a task processing system, including:

The acquisition module is used for each task processing end to obtain the corresponding decomposed training task, and for each task processing end to obtain local training data, and to obtain random numbers or parameters from other task processing ends;

The execution module is used to execute the decomposed training task.

Optionally, the encrypted data is non-private data.

The present invention also provides a task processing device, including:

After each task processing terminal obtains the corresponding decomposed training task, each task processing terminal executes the decomposed training task based on local training data and random numbers or parameters obtained from other task processing terminals.

The present invention also provides a device, including:

One or more processors; and

One or more machine-readable media having instructions stored thereon, when executed by the one or more processors, cause the device to perform one or more of the methods described above.

The present invention also provides one or more machine-readable media on which instructions are stored, which when executed by one or more processors, cause the device to perform one or more of the above-mentioned methods.

As mentioned above, the task processing method, system, equipment and medium provided by the present invention have the following beneficial effects: after one or more participants initiate a training task, the present invention can decompose all training tasks through the task processing terminal, and at the same time Each task processing terminal executes the decomposed training task based on the local training data and random numbers and parameters obtained from other task processing terminals according to the decomposed training task. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple data providers participate and each participant does not trust each other.

Description of the drawings

FIG. 1 is a schematic flowchart of a data sharing learning method provided by an embodiment;

Figure 2 is a schematic flow diagram of a data sharing learning method provided by another implementation;

3 is a schematic diagram of the hardware structure of a data sharing learning system provided by an embodiment;

4 is a schematic diagram of the hardware structure of a data sharing learning system provided by another embodiment;

FIG. 5 is a schematic diagram of the hardware structure of a terminal device provided by an embodiment;

FIG. 6 is a schematic diagram of the hardware structure of a terminal device provided by another embodiment.

Component label description

M10 Obtain the module

M20 Execution module

1100 Input equipment

1101 First processor

1102 Output equipment

1103 First memory

1104 Communication bus

1200 Processing components

1201 Second processor

1202 Second memory

1203 Communication components

1204 Power supply components

1205 Multimedia components

1206 Voice component

1207 Input/output interface

1208 Sensor assembly

Detailed ways

The following describes the implementation of the present invention through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the case of no conflict, the following embodiments and the features in the embodiments can be combined with each other.

It should be noted that the illustrations provided in the following embodiments only illustrate the basic idea of the present invention in a schematic manner. The figures only show the components related to the present invention instead of the actual implementation of the number, shape and number of components. For the size drawing, the type, quantity, and proportion of each component can be changed at will during actual implementation, and the component layout type may also be more complicated.

Referring to Fig. 1, the present invention provides a task processing method, which includes the following steps:

S100, each task processing terminal obtains a corresponding decomposed training task;

S200: Each task processing terminal executes the decomposed training task based on local training data and random numbers or parameters obtained from other task processing terminals.

According to the above records, after one or more participants initiate a training task, this method can decompose all training tasks through the task processing end, and each task processing end is based on the local training data and other task processing ends based on the decomposed training task. The obtained random numbers and parameters are used to perform the decomposed training task. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple data providers participate and each participant does not trust each other. Among them, the participants include: banks, enterprises, government units or organizations, etc.

In an exemplary embodiment, before each task processing terminal separately obtains the corresponding decomposed training task, it further includes: after the task processing platform receives the training task request transmitted by the task processing terminal, decomposes the training task, Obtain the decomposed training task, and assign the decomposed training task to the corresponding task processing terminal.

In an exemplary embodiment, after executing the decomposed training task, it further includes: outputting a shared model. Through the sharing model, it is possible to automatically discover rules in historical data and use the rules to apply unknown data, which can help users make better decisions using data, such as making predictions based on historical data.

In an exemplary embodiment, after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains random numbers or encryption parameters from other task processing terminals, and based on the encrypted random numbers or encryption parameters, decrypts the encrypted data transmitted from other task processing terminals to obtain the decrypted data;

Specifically, the encrypted data in the embodiment of the present application includes non-private data. In this embodiment, data that has not been disclosed to the public by individuals, enterprises, governments, and other institutions or units is called private data; data that the public can learn or obtain through general channels is called non-private data. In the embodiments of this application, the data that will leak user privacy among one or more participants cannot be released locally, ensuring that under the secure multi-party computing solution, one or more participants or participants exchange private information based on established agreements based on established agreements. Come for shared learning and training.

Specifically, the embodiment of the present application obtains random numbers in the training process or model training process parameters in the shared learning model to ensure that the steps or settings made during the training process are consistent, and that the trained model can be applied to common Training data. Among them, the model training process parameters include the number of layers of the convolutional neural network, the size of the convolution kernel of the convolutional neural network, and so on. As an example, for example, the number of layers of the convolutional neural network is 50, and the size of the convolution kernel is 5×5.

Specifically, in the embodiment of the present application, only local training data is acquired, and then combined with the training parameters of other task terminals to ensure that the initially set training conditions are consistent. Among them, the model training parameters include the number of layers of the convolutional neural network, the size of the convolution kernel of the convolutional neural network, and so on. As an example, for example, the number of layers of the convolutional neural network is 20, and the size of the convolution kernel is 3×3.

According to the above embodiment, the number of times that each task processing terminal obtains random numbers or parameters from other task processing terminals is one or more times.

In an exemplary embodiment, the data in at least two participants are exchanged one or more times, including exchanging at least one of the following: random numbers and encryption parameters.

In an exemplary embodiment, it specifically includes:

Initiate one or more training tasks;

Decompose and coordinate the one or more training tasks;

According to one or more training tasks after decomposition and coordination, the data in at least two participants will be exchanged one or more times, so that the data in one or more participants can be exchanged with one or more other participants. To share data in. It also includes obtaining a shared learning model after each participant has completed one or more learning sessions. By sharing the learning model, rules can be automatically discovered in historical data and used to apply unknown data, which can help users make better decisions using data, such as making predictions based on historical data.

In some exemplary embodiments, the learning algorithm used in the learning training includes at least one of the following: linear regression, logistic regression, tree model, deep neural network, graph neural network. Through the management of the learning algorithm, a variety of ways can be used to improve the robustness of the learning algorithm and enhance the security of the data. At the same time, the shared learning model can be trained and optimized, thereby improving the performance and generalization ability of the shared learning model; the index of training optimization includes at least one of the following: definition evaluation, algorithm strategy selection, data set division, parameter tuning, etc.

In some exemplary embodiments, the data in the one or more participants includes at least one of the following: social security data, provident fund data, fixed asset data, and current asset data. Wherein, the current asset data includes at least one of the following: deposit data and loan data.

Specifically, in a specific embodiment, if a bank needs to issue a loan to a certain company, it needs to perform loan qualification analysis, risk assessment, etc. The loan will not become a bad debt. A certain local government agency has the social security data of the certain enterprise, another bank has the data of one or more other loans of the certain enterprise, and the other enterprise has the data of its normal operation with the certain enterprise. Social security data includes the number of employees participating in the social security payment of a certain company, the base number of each employee’s social security payment, etc.; loan data includes the amount of loans, loan time, repayment amount, repayment time, etc. of the certain company; operating data includes Payment method, payment time, etc. Under the condition of distrusting each other and ensuring that private data will not be leaked, the local government agency, the other bank, and the other enterprise are each as a participant, and the data they hold are shared as data for shared learning. Learn to obtain a shared learning model after shared learning; the bank can perform loan qualification analysis and risk assessment on the company based on the shared learning model, and assist the bank in lending to the company, such as predicting the company’s future operating income and repayment Ability etc.

In summary, after one or more participants initiate training tasks, the present invention can decompose all training tasks through the task processing terminal, and each task processing terminal can process the training tasks based on local training data and other tasks according to the decomposed training tasks. The random numbers and parameters obtained by the terminal are used to perform the decomposed training task. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple participants participate and each participant does not trust each other. At the same time, this method can exchange the non-private data in one or more parties with the non-private data in the other one or more parties, so that the non-private data in one or more parties can be exchanged with the non-private data in one or more parties. The non-private data in the other one or more participants is shared; then one or more learning is performed based on the shared non-private data to realize shared learning and obtain a shared learning training model. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple data providers participate and each participant does not trust each other. At the same time, by sharing the learning model, this method can automatically discover laws in historical data, and use the laws to apply unknown data to help users make better decisions using data, such as making predictions based on historical data.

As shown in Figure 2, a task processing method is also provided, including:

S1, one or more participants are used to provide non-private data and deploy machine learning modules;

S2, one or more participants initiate one or more machine learning training tasks;

S3, the model platform decomposes and coordinates one or more machine learning training tasks,

S4: The model platform issues one or more training tasks to each participant;

S5, each participant reads the local training data and random numbers or parameters obtained from other task processing terminals to the local machine learning module;

S6, the machine learning module of each participant exchanges one or more parameters, so that the data in one or more participants (including training data and non-private data) can be exchanged with one or more other participants. Data (including training data and non-private data) for sharing;

S7: Perform one or more learning based on the shared data (including training data and non-private data); after completing one or more shared learning training, obtain a shared learning model of each participant.

Among them, the model platform is used to trigger and coordinate learning and training tasks; the local machine learning module in the participants is used to receive the decomposed and coordinated machine learning tasks issued by the model platform, based on local training data and obtained from other participants Random numbers or parameters for machine learning.

According to the above records, this method can exchange non-private data in one or more participants with non-private data in the other one or more participants, so that the non-private data in one or more participants can be exchanged. The data can be shared with the non-private data of one or more other participants; then one or more learnings are performed based on the shared non-private data to achieve shared learning and obtain a shared learning training model. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple data providers participate and each participant does not trust each other. At the same time, by sharing the learning model, this method can automatically discover rules in historical data, and use the rules to apply unknown data to help users make better decisions using data, such as making predictions based on historical data.

As shown in Figure 3, a task processing system includes:

The obtaining module M10 is used for each task processing end to obtain the corresponding decomposed training tasks, and for each task processing end to obtain local training data, and to obtain random numbers or parameters from other task processing ends;

The execution module M20 is used to execute the decomposed training task.

In an exemplary embodiment, it specifically includes:

Initiate one or more training tasks;

Decompose and coordinate the one or more training tasks;

As shown in Figure 4, a data sharing learning system is also provided, which includes a control module and a learning module; the control module is located in the model platform; and the learning module is located in the participants. Among them, the learning module of the participants exchanges random numbers or encryption parameters, and is triggered and coordinated by the control module to perform shared machine learning; each participant deploys its own machine learning module, and each participant initiates one or more machine learning Training tasks; after receiving one or more training tasks, the control module decomposes and coordinates, and sends one or more training tasks to each participant; each participant reads the non-privacy of one or more local participants Data and non-private data from other participants to the local machine learning module; enable non-private data from one or more participants to be shared with non-private data from one or more other participants; based on After sharing the non-private data, learn one or more times; after completing one or more shared learning training, obtain the shared learning model of each participant.

According to the above records, this system can exchange non-private data in one or more participants with non-private data in the other one or more participants, so that the non-private data in one or more participants The data can be shared with the non-private data of one or more other participants; then one or more learnings are performed based on the shared non-private data to achieve shared learning and obtain a shared learning training model. That is, this method can aggregate or assist in the aggregation of multi-party data information and protect multi-party private data in a scenario where multiple data providers participate and each participant does not trust each other. At the same time, by sharing the learning model, this method can automatically discover laws in historical data, and use the laws to apply unknown data to help users make better decisions using data, such as making predictions based on historical data.

The embodiment of the present application also provides a task processing device, including:

After each task processing terminal obtains the corresponding decomposed training task, each task processing terminal is based on local training data and random numbers or parameters obtained from other task processing terminals;

Perform the decomposed training task.

In this embodiment, the task processing device executes the above-mentioned system or method. For specific functions and technical effects, refer to the above-mentioned embodiment, which will not be repeated here.

The embodiment of the present application also provides a device, which may include: one or more processors; and one or more machine-readable media on which instructions are stored, when executed by the one or more processors At this time, the device is caused to execute the method described in FIG. 1. In practical applications, the device can be used as a terminal device or a server. Examples of terminal devices include: smart phones, tablets, e-book readers, MP3 (moving picture experts compress standard voice level 3, Moving Picture Experts Group Audio Layer III) Players, MP4 (Moving Picture Experts Group Audio Layer IV) Players, laptop portable computers, car computers, desktop computers, set-top boxes, smart TVs, wearable devices And so on, the embodiments of the present application do not impose restrictions on specific devices.

The embodiment of the present application also provides a non-volatile readable storage medium. The storage medium stores one or more modules (programs). When the one or more modules are applied to a device, the device can execute Instructions for the steps included in the method described in FIG. 1 of the embodiment of the present application.

FIG. 5 is a schematic diagram of the hardware structure of a terminal device provided by an embodiment of the application. As shown in the figure, the terminal device may include: an input device 1100, a first processor 1101, an output device 1102, a first memory 1103, and at least one communication bus 1104. The communication bus 1104 is used to implement communication connections between components. The first memory 1103 may include a high-speed RAM memory, or may also include a non-volatile storage NVM, such as at least one disk memory. The first memory 1103 may store various programs for completing various processing functions and implementing this embodiment. Method steps.

Optionally, the foregoing first processor 1101 may be, for example, a central processing unit (Central Processing Unit, CPU for short), an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), and A programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components are implemented, and the first processor 1101 is coupled to the aforementioned input device 1100 and via a wired or wireless connection. Output device 1102.

Optionally, the aforementioned input device 1100 may include multiple input devices, for example, it may include at least one of a user-oriented user interface, a device-oriented device interface, a software programmable interface, a camera, and a sensor. Optionally, the device-oriented device interface may be a wired interface for data transmission between the device and the device, or a hardware plug-in interface for data transmission between the device and the device (such as a USB interface, a serial port, etc.) ); Optionally, the user-oriented user interface may be, for example, user-oriented control buttons, a voice input device for receiving voice input, and a touch sensing device for receiving user touch input (such as a touch screen with touch sensing function, touch Control board, etc.); Optionally, the programmable interface of the above software may be, for example, an entry for the user to edit or modify the program, such as the input pin interface or input interface of the chip, etc.; the output device 1102 may include output devices such as a display and audio .

In this embodiment, the processor of the terminal device includes functions for executing each module of the speech recognition device in each device. For specific functions and technical effects, please refer to the above-mentioned embodiment, which will not be repeated here.

FIG. 6 is a schematic diagram of the hardware structure of a terminal device provided by an embodiment of the application. Fig. 6 is a specific embodiment of Fig. 5 in the implementation process. As shown in the figure, the terminal device of this embodiment may include a second processor 1201 and a second memory 1202.

The second processor 1201 executes the computer program code stored in the second memory 1202 to implement the method described in FIG. 1 in the foregoing embodiment.

The second memory 1202 is configured to store various types of data to support operations on the terminal device. Examples of these data include instructions for any application or method operating on the terminal device, such as messages, pictures, videos, and so on. The second memory 1202 may include a random access memory (random access memory, RAM for short), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

Optionally, the second processor 1201 is provided in the processing component 1200. The terminal device may also include: a communication component 1203, a power supply component 1204, a multimedia component 1205, a voice component 1206, an input/output interface 1207 and/or a sensor component 1208. The specific components included in the terminal device are set according to actual requirements, which is not limited in this embodiment.

The processing component 1200 generally controls the overall operation of the terminal device. The processing component 1200 may include one or more second processors 1201 to execute instructions to complete all or part of the steps in the foregoing data processing method. In addition, the processing component 1200 may include one or more modules to facilitate the interaction between the processing component 1200 and other components. For example, the processing component 1200 may include a multimedia module to facilitate the interaction between the multimedia component 1205 and the processing component 1200.

The power supply component 1204 provides power for various components of the terminal device. The power supply component 1204 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for terminal devices.

The multimedia component 1205 includes a display screen that provides an output interface between the terminal device and the user. In some embodiments, the display screen may include a liquid crystal display (LCD) and a touch panel (TP). If the display screen includes a touch panel, the display screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, sliding, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure related to the touch or slide operation.

The voice component 1206 is configured to output and/or input voice signals. For example, the voice component 1206 includes a microphone (MIC). When the terminal device is in an operating mode, such as a voice recognition mode, the microphone is configured to receive external voice signals. The received voice signal may be further stored in the second memory 1202 or transmitted via the communication component 1203. In some embodiments, the voice component 1206 further includes a speaker for outputting voice signals.

The input/output interface 1207 provides an interface between the processing component 1200 and a peripheral interface module. The peripheral interface module may be a click wheel, a button, or the like. These buttons may include, but are not limited to: volume buttons, start buttons, and lock buttons.

The sensor component 1208 includes one or more sensors, which are used to provide various aspects of state evaluation for the terminal device. For example, the sensor component 1208 can detect the open/close state of the terminal device, the relative positioning of the component, and the presence or absence of contact between the user and the terminal device. The sensor component 1208 may include a proximity sensor configured to detect the presence of nearby objects when there is no physical contact, including detecting the distance between the user and the terminal device. In some embodiments, the sensor component 1208 may also include a camera and the like.

The communication component 1203 is configured to facilitate wired or wireless communication between the terminal device and other devices. Terminal devices can access wireless networks based on communication standards, such as WiFi, 2G or 3G, or a combination of them. In one embodiment, the terminal device may include a SIM card slot for inserting a SIM card so that the terminal device can log in to the GPRS network and establish communication with the server via the Internet.

It can be seen from the above that the communication component 1203, the voice component 1206, the input/output interface 1207, and the sensor component 1208 involved in the embodiment in FIG. 6 can all be used as implementations of the input device in the embodiment in FIG. 5.

The above-mentioned embodiments only exemplarily illustrate the principles and effects of the present invention, but are not used to limit the present invention. Anyone familiar with this technology can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed in the present invention should still be covered by the claims of the present invention.

Claims

A task processing method, characterized in that it includes:

Each task processing terminal obtains the corresponding decomposed training task;

Each task processing terminal executes the decomposed training task based on local training data and random numbers or parameters obtained from other task processing terminals.
The task processing method according to claim 1, wherein before each task processing terminal separately obtains the corresponding decomposed training task, the method further comprises:

After receiving the training task request transmitted by the task processing terminal, the task processing platform decomposes the training task, obtains the decomposed training task, and allocates the decomposed training task to the corresponding task processing terminal.
The task processing method according to claim 1, wherein after executing the decomposed training task, it further comprises: outputting a shared model.
The task processing method according to claim 1, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains random numbers or encryption parameters from other task processing terminals, and based on the random numbers or encryption parameters, decrypts encrypted data transmitted from other task processing terminals to obtain decrypted data;

Learning and training are performed through the local training data and the decrypted data.
The task processing method according to claim 1, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains model training process random numbers or model training process parameters from other task processing terminals;

According to the model training process random number or model training process parameters and local training data, learning and training are performed.
The task processing method according to claim 1, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains training parameters from other task processing terminals;

Perform learning training based on the training parameters and local training data.
The task processing method according to claim 6, wherein the training parameters include: the number of layers of the convolutional neural network and the size of the convolution kernel.
The task processing method according to claim 1, wherein the number of times that each task processing terminal obtains random numbers or parameters from other task processing terminals is one or more times.
The task processing method according to any one of claims 4 to 6, wherein the learning algorithm used in the learning training includes at least one of the following: linear regression, logistic regression, tree model, deep neural network, graph neural network.
The task processing method according to any one of claims 4 to 6, wherein the training data includes at least one of the following: social security data, provident fund data, fixed asset data, and current asset data.
The task processing method according to claim 10, wherein the current asset data includes at least one of the following: deposit data and loan data.
The task processing method according to claim 4, wherein the encrypted data is non-private data.
A task processing system, characterized in that it includes:

The acquisition module is used for each task processing end to obtain the corresponding decomposed training task, and for each task processing end to obtain local training data, and to obtain random numbers or parameters from other task processing ends;

The execution module is used to execute the decomposed training task.
The task processing system according to claim 13, wherein before each task processing terminal obtains the corresponding decomposed training task, it further comprises:

After receiving the training task request transmitted by the task processing terminal, the task processing platform decomposes the training task, obtains the decomposed training task, and allocates the decomposed training task to the corresponding task processing terminal.
The task processing system according to claim 13, wherein after executing the decomposed training task, it further comprises: outputting a shared model.
The task processing system according to claim 13, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains random numbers or encryption parameters from other task processing terminals, and based on the random numbers or encryption parameters, decrypts the encrypted data transmitted from other task processing terminals to obtain decrypted data;

Learning and training are performed through the local training data and the decrypted data.
The task processing system according to claim 13, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains model training process random numbers or model training process parameters from other task processing terminals;

According to the model training process random number or model training process parameters and local training data, learning and training are performed.
The task processing system according to claim 13, wherein after each task processing terminal obtains the corresponding decomposed training task, each task processing terminal reads local training data;

Each task processing terminal obtains training parameters from other task processing terminals;

Perform learning training based on the training parameters and local training data.
The task processing system according to claim 18, wherein the training parameters include: the number of layers of the convolutional neural network and the size of the convolution kernel.
The task processing system according to claim 13, wherein the number of times that each task processing terminal obtains a random number or a parameter from another task processing terminal is one or more times.
The task processing system according to any one of claims 16 to 18, wherein the learning algorithm used in the learning training includes at least one of the following: linear regression, logistic regression, tree model, deep neural network, graph neural network.
The task processing system according to any one of claims 16 to 18, wherein the training data includes at least one of the following: social security data, provident fund data, fixed asset data, and current asset data.
The task processing system according to claim 22, wherein the current asset data includes at least one of the following: deposit data and loan data.
The task processing system according to claim 16, wherein the encrypted data is non-private data.
A task processing device, characterized in that it includes:

After each task processing terminal obtains the corresponding decomposed training task, each task processing terminal executes the decomposed training task based on local training data and random numbers or parameters obtained from other task processing terminals.
A device, characterized in that it comprises:

One or more processors; and

One or more machine-readable media having instructions stored thereon, when executed by the one or more processors, cause the device to execute the method according to one or more of claims 1-12.
One or more machine-readable media, characterized in that instructions are stored thereon, which when executed by one or more processors, cause the device to execute the method according to one or more of claims 1-12.