Federal cooperative optimization method, device, medium, and computer program product

Abstract

The application discloses a federal cooperation optimization method, device, medium and computer program product. The federal cooperation optimization method comprises the following steps: obtaining a cooperation data set dynamically maintained on a third-party device, and determining a sample data type supported by the cooperation data set; selecting a target feature version in the cooperation data set, and generating a federal cooperation sample set according to the sample data type and the target feature version; and performing federal cooperation with a second device corresponding to the cooperation data set based on the federal cooperation sample set, to obtain a federal cooperation result. The application solves the technical problem of low federal cooperation efficiency.

Description

Technical Field

[0001] This application relates to the field of machine learning technology in financial technology (Fintech), and more particularly to a federated cooperative optimization method, apparatus, medium, and computer program product. Background Technology

[0002] With the continuous development of fintech, especially internet fintech, more and more technologies (such as distributed systems and artificial intelligence) are being applied in the financial field. However, the financial industry is also placing higher demands on technology, such as on the distribution of tasks to be completed.

[0003] With the continuous development of computer technology, federated learning is being applied more and more widely. In federated cooperation scenarios, the participating parties usually need to match their business information in order to truly carry out federated cooperation. Currently, account information is usually used to identify different cooperative businesses with different partners. However, since the business information of the participating parties is usually dynamic, for example, data features may be added or outdated data features may no longer be used, and account information cannot detect changes in business information, such as changes in different features of the same data source. As a result, the participating parties in federated cooperation often need to negotiate offline to resolve the problem of mismatched business information, leading to low efficiency in federated cooperation. Summary of the Invention

[0004] The main purpose of this application is to provide a method, apparatus, medium, and computer program product for optimizing federated cooperation, aiming to solve the technical problem of low efficiency in federated learning in the prior art.

[0005] To achieve the above objectives, this application provides a federated cooperation optimization method, which is applied to a first device, and the federated cooperation optimization method includes:

[0006] Obtain the collaborative dataset that is dynamically maintained on the third-party device, and determine the sample data types supported by the collaborative dataset;

[0007] Select the target feature version from the collaborative dataset, and generate a federated collaborative sample set based on the sample data type and the target feature version;

[0008] Based on the federated cooperation sample set, the second device corresponding to the cooperation dataset performs federated cooperation to obtain the federated cooperation result.

[0009] To achieve the above objectives, this application also provides a federated cooperation optimization method, which is applied to a second device, and the federated cooperation optimization method includes:

[0010] Generate a collaborative dataset and upload it to a third-party device;

[0011] The third-party device initiates a federated cooperation request to the first device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset;

[0012] The third party, based on a preset local sample set, combines the federated cooperation sample set in the first device to perform federated cooperation, thereby obtaining the federated cooperation result.

[0013] To achieve the above objectives, this application also provides a federated cooperation optimization method, which is applied to a third-party device, and the federated cooperation optimization method includes:

[0014] The system receives a collaborative dataset and a federated collaboration request sent by a second device, and publishes the federated collaboration request to a first device so that the first device, after receiving the federated collaboration request, can generate a federated collaboration sample set based on the collaborative dataset.

[0015] The first device and the second device are federated in a federal coordination manner so that the first device can cooperate with the second device in a federal cooperation sample set to generate a federal cooperation result.

[0016] This application also provides a federated cooperation optimization device, which is a virtual device and is applied to a first device. The federated cooperation optimization device includes:

[0017] The determination module is used to obtain the collaborative dataset dynamically maintained on the third-party device and determine the sample data types supported by the collaborative dataset;

[0018] The generation module is used to select the target feature version in the collaborative dataset and generate a federated collaborative sample set based on the sample data type and the target feature version.

[0019] The federated cooperation module is used to perform federated cooperation with the second device corresponding to the cooperation dataset based on the federated cooperation sample set, and obtain the federated cooperation result.

[0020] To achieve the above objectives, this application also provides a federated cooperation optimization device, which is a virtual device and is applied to a second device. The federated cooperation optimization device includes:

[0021] The upload module is used to generate a collaborative dataset and upload the collaborative dataset to a third-party device;

[0022] The request initiation module is used to initiate a federated cooperation request to the first device through the third-party device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset;

[0023] The federated cooperation module is used to perform federated cooperation by the third party based on a preset local sample set and in conjunction with the federated cooperation sample set in the first device to obtain the federated cooperation result.

[0024] To achieve the above objectives, this application also provides a federated cooperation optimization device, which is a virtual device and is applied to a third-party device. The federated cooperation optimization device includes:

[0025] The receiving and sending module is used to receive the cooperative dataset and federated cooperation request sent by the second device, and to publish the federated cooperation request to the first device, so that the first device can generate a federated cooperation sample set based on the cooperative dataset after receiving the federated cooperation request;

[0026] The coordination module is used to perform federated coordination between the first device and the second device, so that the first device can perform federated cooperation with the second device based on the federated cooperation sample set and generate federated cooperation results.

[0027] This application also provides a federated cooperation optimization device, which is a physical device. The federated cooperation optimization device includes: a memory, a processor, and a program of the federated cooperation optimization method stored in the memory and executable on the processor. When the program of the federated cooperation optimization method is executed by the processor, it can implement the steps of the federated cooperation optimization method as described above.

[0028] This application also provides a medium, which is a readable storage medium, on which a program implementing the federated cooperative optimization method is stored. When the program is executed by a processor, it implements the steps of the federated cooperative optimization method as described above.

[0029] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the federated cooperative optimization method as described above.

[0030] This application provides a federated cooperation optimization method, device, medium, and computer program product. Compared to the existing technology that uses account information to identify different cooperative businesses with different partners, this application first obtains a dynamically maintained cooperative dataset on a third-party device and determines the sample data types supported by the cooperative dataset. The sample data type indicates the sample type of the selected federated cooperation sample. Then, a target feature version is selected from the cooperative dataset, and a federated cooperation sample set is generated based on the sample data type and the target feature version. The target feature version indicates the feature type of the selected federated cooperation sample. Furthermore, based on the sample data type and the target feature version, a federated cooperation sample set is generated. The aforementioned target feature version can provide another federated partner with a federated cooperation sample set that matches their business. Since the cooperation dataset is stored on a third-party device, it can be dynamically maintained. The first device can easily perceive the dynamic changes in the business of the other federated partner. Based on the federated cooperation sample set, it can efficiently conduct federated cooperation with the second device corresponding to the cooperation dataset to obtain the federated cooperation result. This overcomes the technical shortcomings of the federated cooperation, which is often inefficient because the business information of the participating parties is usually dynamic and cannot be perceived through account information. This often requires the participating parties to negotiate offline to resolve the problem of mismatched business information. This improves the efficiency of federated cooperation. Attached Figure Description

[0031] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating the first embodiment of the federal cooperation optimization method of this application;

[0034] Figure 2 This is a flowchart illustrating the second embodiment of the federal cooperation optimization method of this application;

[0035] Figure 3 This is a flowchart illustrating the third embodiment of the federal cooperation optimization method of this application;

[0036] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the federal cooperation optimization method in the embodiments of this application;

[0037] Figure 5 This is a schematic diagram of the hardware architecture involved in the embodiments of this application.

[0038] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0039] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0040] This application provides a method for optimizing federal cooperation. In the first embodiment of this method, refer to... Figure 1 The federated cooperation optimization method is applied to the first device, and the federated cooperation optimization method includes:

[0041] Step S10: Obtain the collaborative dataset dynamically maintained on the third-party device, and determine the sample data types supported by the collaborative dataset;

[0042] In this embodiment, it should be noted that the first device is a data provider in the federated cooperation, the third-party device is a federated cooperation cloud service platform and a coordinator of the federated cooperation, and the third-party device is communicatively connected to at least one second device, wherein the second device is an initiator in the federated cooperation, and the federated cooperation is a collaborative process in federated learning, including intersection matching process, federated learning modeling process, and joint model prediction process, etc.

[0043] Additionally, it should be noted that the collaborative dataset is uploaded from the second device to the third-party device. When the business information in the second device changes, the collaborative dataset on the third-party device can be maintained and updated through interaction. Thus, the first device can indirectly perceive changes in the business information in the second device through the collaborative dataset. These changes include adding new features, deleting old features, and changes to the server / algorithm system. In one implementation, the collaborative dataset is dynamically maintained at the version level to reflect changes in the business information in the second device. The collaborative dataset includes target feature versions, system versions, collaborative data type information, and collaborative data description information, etc., and each target feature version is independent to ensure that there is no confusion between target feature versions. Each target feature version includes at least one feature. The tag type includes at least one feature tag, which is an identifier of a feature. The system version includes at least one system version number, which is associated with a system resource number. For example, a one-to-one correspondence between the system version number and the system resource number can be set, so that physical resources for federated cooperation, such as servers and algorithm systems, can be accurately selected based on the system resource number. The cooperative data type information includes at least one data tag type, such as a mobile phone number tag type and an ID card number tag type. The data tag type includes at least one sample tag, which is an identifier of the sample, including a mobile phone number and an ID card number. The cooperative data description information is used to describe the data required for federated cooperation by the second device, so that the first device can initially determine whether it meets the qualifications of the federated partner required by the second device.

[0044] Obtain the dynamically maintained collaborative dataset on the third-party device and determine the sample data types supported by the collaborative dataset. Specifically, obtain the dynamically maintained collaborative dataset on the third-party device and determine the sample data types supported by the collaborative dataset based on the data type labels in the collaborative dataset. For example, assume that the data type label is A, where A indicates that the sample data type supported by the collaborative dataset is the ID card number data type.

[0045] The federated cooperation optimization method further includes, after the step of determining the sample data type corresponding to the cooperative dataset:

[0046] Step A10: Obtain the federated learning system version and verify that the federated learning system version is within the range of system versions supported by the collaborative dataset;

[0047] In this embodiment, it should be noted that the federated learning system version is a system version used for federated cooperation, wherein the federated learning system version specifies the physical information for federated cooperation, such as servers and algorithm systems.

[0048] Obtain the federated learning system version and verify whether the federated learning system version is within the range of system versions supported by the collaborative dataset. Specifically, obtain the first version number of the local federated learning system version and extract the set of system version numbers in the collaborative dataset. Determine whether there is a target system version number in the set of system version numbers that is consistent with the first version number. If there is, it is determined that the federated learning system version is within the range of system versions supported by the collaborative dataset. If not, it is determined that the federated learning system version is not within the range of system versions supported by the collaborative dataset.

[0049] Step A20, if yes, then perform the step of selecting the target feature version in the cooperative dataset:

[0050] Step A30: If not, exit federal cooperation.

[0051] In this embodiment, if so, the system will withdraw from the federated collaboration and update its local system version until it is within the range of system versions supported by the collaborative dataset, at which point it will rejoin the federated collaboration.

[0052] Step S20: Select the target feature version in the collaborative dataset, and generate a federated collaborative sample set based on the sample data type and the target feature version;

[0053] In this embodiment, a target feature version is selected from the collaborative dataset, and a federated collaborative sample set is generated based on the sample data type and the target feature version. Specifically, each target feature version number in the collaborative dataset is obtained, and the target feature version is selected from the target feature versions corresponding to each target feature version number. Then, each sample belonging to the sample data type is selected from a preset database as the target sample set, and each target feature belonging to the target feature version is selected from each data feature corresponding to the target sample set. Finally, the sample set corresponding to each target feature is used as the federated collaborative sample set. Here, the target sample is a vector composed of the feature values ​​of at least one feature, and the data feature is a vector composed of the feature values ​​of at least one sample. For example, assuming the target sample set is a matrix, the value of each bit in the matrix is ​​a feature value, then one row of the matrix is ​​a target sample, and one column of the matrix is ​​a data feature, and all values ​​in the column are the feature values ​​of the corresponding data feature.

[0054] The sample data type includes a data label type, the target feature version includes a feature label type, and the federated cooperation sample set includes at least one federated cooperation sample.

[0055] The step of generating a federated cooperation sample set based on the sample data type and the target feature version includes:

[0056] Step S21: Select each sample label corresponding to the data label type and each feature label corresponding to the feature label type;

[0057] In this embodiment, it should be noted that the data tag type is a classification type of data tag, used to identify the category of data tag. For example, if the data tag type is ID card number tag type, it means that the data tag is ID card number, etc. The feature tag type is a classification type of feature tag, used to identify the category of feature tag. For example, when building a risk control model in a federated environment, if the feature tag type is deposit feature tag type, it means that the feature tag is a tag of various deposit features.

[0058] Select each sample label corresponding to the data label type and each feature label corresponding to the feature label type. Specifically, select each sample label belonging to the data label type and each feature label belonging to the feature label type.

[0059] Step S22: Generate each of the federated cooperation samples based on each of the sample labels and each of the feature labels.

[0060] In this embodiment, each federated cooperation sample is generated based on each sample label and each feature label. Specifically, each federated cooperation sample is selected from a preset database using each sample label and each feature label as an index. The federated cooperation sample is composed of the feature values ​​of the target features corresponding to each feature label.

[0061] Step S30: Based on the federated cooperation sample set, perform federated cooperation with the second device corresponding to the cooperation dataset to obtain the federated cooperation result.

[0062] In this embodiment, it should be noted that the federated cooperation includes the intersection matching process, the federated learning modeling process, and the federated model prediction process in federated learning. The federated learning includes horizontal federated learning and vertical federated learning. The intersection matching process is used to perform sample alignment or feature alignment. The federated learning modeling process is used to build a machine learning model based on federated learning. The federated model prediction process is used to perform model prediction based on federated learning.

[0063] Based on the federated cooperation sample set, a second device corresponding to the cooperation dataset is federated to obtain federated cooperation results. Specifically, a system version supported by the cooperation dataset is selected, and federated learning interaction is performed with the second device according to the system resources corresponding to the system version, so as to obtain federated cooperation results based on the federated cooperation sample set. In the federated cooperation, the second device provides a locally selected local sample set.

[0064] Additionally, it should be noted that, in one implementation, the intersection matching process is as follows:

[0065] The first device sends each first sample ID of the federated cooperation sample set to the second device. Then, the second device calculates the intersection of each second sample ID obtained locally with each first sample ID to obtain each intersection sample ID, and sends each intersection sample ID to the first device to complete the intersection matching process.

[0066] In one implementation, the federated learning modeling process is as follows:

[0067] The first device and the second device iteratively train the local model. When the number of iterations of the local model reaches a preset number of iterations, the first device calculates the first encrypted model gradient by exchanging intermediate results with the second device, so that the second device can calculate the second encrypted model gradient. Then, the first device sends the first encrypted model gradient to a third-party device, and the second device sends the second encrypted model gradient to the third-party device. Then, the third-party device aggregates the first encrypted model gradient and the second encrypted model gradient to obtain the aggregated encrypted model gradient. Then, the first device receives the aggregated encrypted model gradient, decrypts the aggregated model gradient to obtain the aggregated model gradient, and updates the local model based on the aggregated model gradient. It is then determined whether the local model meets the preset iterative training termination condition. If it does, the local model is used as a federated model. If it does not, the iterative training of the local model is returned to the previous steps.

[0068] The step of performing federated cooperation with the second device corresponding to the federated cooperation dataset based on the federated cooperation sample set to obtain the federated cooperation result includes:

[0069] Step S31: Select the federated learning version supported by the collaborative dataset;

[0070] In this embodiment, the federated learning version supported by the collaborative dataset is selected. Specifically, the target system version number in the collaborative dataset is selected, and the system version corresponding to the target system version number is used as the federated learning version.

[0071] Step S32: Based on the system resource number matched by the federated learning version and the federated cooperation sample set, perform federated cooperation with the second device to obtain the federated cooperation result.

[0072] In this embodiment, based on the system resource number matched by the federated learning version and the federated cooperation sample set, federated cooperation is performed with the second device to obtain the federated cooperation result. Specifically, the system resource number corresponding to the federated learning version is matched, and federated interaction is performed with the second device based on the system resource number corresponding to the system resource number, so as to perform federated cooperation with the second device based on the federated cooperation sample set to obtain the federated cooperation sample set.

[0073] This application provides a federated cooperation optimization method. Compared to existing technologies that use account information to identify different cooperative businesses with different partners, this application first obtains a dynamically maintained cooperative dataset on a third-party device and determines the sample data types supported by the cooperative dataset. The sample data type indicates the sample type of the selected federated cooperation sample. Then, a target feature version is selected from the cooperative dataset, and a federated cooperation sample set is generated based on the sample data type and the target feature version. The target feature version indicates the feature type of the selected federated cooperation sample. Finally, based on the sample data type and the target feature version, a federated cooperation sample set is generated. The first version can provide another federated partner with a federated cooperation sample set that matches their business. Since the cooperation dataset is stored on a third-party device, it can be dynamically maintained. The first device can easily detect the dynamic changes in the business of the other federated partner. Based on the federated cooperation sample set, it can efficiently conduct federated cooperation with the second device corresponding to the cooperation dataset to obtain the federated cooperation result. This overcomes the technical defects of the federated cooperation, which is often inefficient because the business information of the participants is usually dynamic and cannot be detected by account information. This often requires the participants in the federated cooperation to negotiate offline to solve the problem of mismatched business information. This improves the efficiency of federated cooperation.

[0074] Furthermore, referring to Figure 2 Based on the first embodiment of this application, in another embodiment of this application, the federated learning optimization method is applied to a second device, and the federated learning optimization method includes:

[0075] Step B10: Generate a collaborative dataset and upload the collaborative dataset to a third-party device;

[0076] In this embodiment, a collaborative dataset is generated and uploaded to a third-party device. Specifically, the target feature version, system version, collaborative data type information, and collaborative data description information required for federated collaboration are determined. Based on the target feature version, system version, collaborative data type information, and collaborative data description information, a collaborative dataset is generated and uploaded to a third-party device. The specific composition of the collaborative dataset can be referred to the specific content in step S10.

[0077] The federated learning optimization method further includes, after the step of uploading the collaborative dataset to a third-party device:

[0078] Step C10: Obtain business dynamic update information and send the business dynamic update information to the third-party device, wherein the business dynamic update information is used to instruct the updating of the cooperative dataset.

[0079] In this embodiment, it should be noted that the business dynamic update information is information that dynamically updates business information and physical information, used to indicate the specific content that needs to be updated in business information and physical information. The business information includes target feature version and sample data type, etc., and the physical information includes server and algorithm system, etc.

[0080] The system acquires dynamic business update information and sends it to the third-party device. The dynamic business update information is used to instruct the updating of the collaborative dataset. Specifically, the system acquires dynamic business update information and sends it to the third-party device so that the third-party device can update the target feature version, system version, or sample data type in the collaborative dataset based on the dynamic business update information, thereby obtaining a dynamically maintained collaborative dataset.

[0081] Step B20: The third-party device initiates a federated cooperation request to the first device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset;

[0082] In this embodiment, a federated cooperation request is initiated by the third-party device to the first device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset. Specifically, the third-party device sends a federated cooperation request to the third-party device, and the third-party device publishes the federated cooperation request to the first device. After receiving the federated cooperation request, the first device obtains the cooperation dataset dynamically maintained on the third-party device, determines the sample data types supported by the cooperation dataset, selects the target feature version in the cooperation dataset, and generates a federated cooperation sample set based on the sample data types and the target feature versions. The specific process of generating the federated cooperation sample set can be referred to steps S10 to S20.

[0083] Step B30: The third party performs federated cooperation based on a preset local sample set and in conjunction with the federated cooperation sample set in the first device to obtain the federated cooperation result.

[0084] In this embodiment, it should be noted that the federated cooperation includes an intersection matching process, a federated learning modeling process, and a federated model prediction process.

[0085] The third party, based on a preset local sample set, performs federated cooperation in conjunction with the federated cooperation sample set in the first device to obtain federated cooperation results. Specifically, it interacts with the second device through federated learning to perform federated cooperation with the first device based on the preset local sample set to obtain federated cooperation results. The first device provides the federated cooperation sample set in the federated cooperation. The specific process of the federated cooperation can be referred to in step S30.

[0086] This application provides a federated learning optimization method, which generates a collaborative dataset and uploads it to a third-party device. The second device can dynamically update the collaborative dataset based on business updates, and then initiate a federated cooperation request to a first device via the third-party device. The first device then generates a federated cooperation sample set based on the collaborative dataset. The third party uses a preset local sample set, where the data type of the samples in the collaborative dataset indicates the sample type of the federated cooperation samples selected by the first device, and the target feature version in the collaborative dataset indicates the feature type of the federated cooperation samples selected by the first device. This generates a business-matched federated cooperation sample set. Since the collaborative dataset is stored on a third-party device, it can be dynamically maintained. The first device can easily perceive dynamic changes in the business of another federated partner and then perform federated cooperation with the federated cooperation sample set in the first device to obtain the federated cooperation result. This overcomes the technical shortcomings of inefficient federated cooperation, where the business information of participating parties is usually dynamically changing, and changes cannot be perceived through account information, often requiring offline negotiation among participating parties to resolve mismatches in business information. This method improves the efficiency of federated cooperation.

[0087] Furthermore, referring to Figure 3 Based on the first, second, and third embodiments of this application, in another embodiment of this application, the federated cooperation optimization method is applied to a third-party device, and the federated cooperation optimization method includes:

[0088] Step D10: Receive the collaborative dataset and federated collaboration request sent by the second device, and publish the federated collaboration request to the first device so that the first device can generate a federated collaboration sample set based on the collaborative dataset after receiving the federated collaboration request.

[0089] In this embodiment, a collaborative dataset and a federated collaboration request sent by a second device are received, and the federated collaboration request is published to a first device. After receiving the federated collaboration request, the first device generates a federated collaboration sample set based on the collaborative dataset. Specifically, the collaborative dataset and the federated collaboration request sent by the second device are received, and the federated collaboration request is published to the first device. After receiving the federated collaboration request, the first device obtains the collaborative dataset dynamically maintained on the third-party device, determines the sample data types supported by the collaborative dataset, selects the target feature version in the collaborative dataset, and generates a federated collaboration sample set based on the sample data types and the target feature versions. The specific process of generating the federated collaboration sample set can be referred to steps S10 to S20.

[0090] The federated cooperation optimization method further includes, after the step of receiving the cooperative dataset sent by the second device:

[0091] Step E10: Receive service dynamic update information sent by the second device;

[0092] In this embodiment, it should be noted that the business dynamic update information is information that dynamically updates business information and physical information, used to indicate the specific content that needs to be updated in business information and physical information. The business information includes target feature version and sample data type, etc., and the physical information includes server and algorithm system, etc.

[0093] Step E20: Based on the business dynamic update information, update at least one of the target feature version, system version, and sample data type in the collaborative dataset to maintain the collaborative dataset.

[0094] In this embodiment, based on the business dynamic update information, at least one of the target feature version, system version, and sample data type in the collaborative dataset is updated to maintain the collaborative dataset. Specifically, the business dynamic update information is parsed to obtain target feature version update information, and / or system version update information, and / or sample data type update information. Then, based on the target feature version update information, the target feature version in the collaborative dataset is updated; and / or based on the system version update information, the system version in the collaborative dataset is updated; and / or based on the sample data type update information, the sample data type in the collaborative dataset is updated to maintain the collaborative dataset and obtain a dynamically maintained collaborative dataset. The update operation includes operations such as editing, modifying, adding, and deleting.

[0095] Step D20: Perform federated coordination between the first device and the second device, so that the first device can perform federated cooperation with the second device based on the federated cooperation sample set and generate federated cooperation results.

[0096] In this embodiment, the first device and the second device are federally coordinated so that the first device can perform federal cooperation with the second device based on the federal cooperation sample set and generate a federal cooperation result. Specifically, the first device and the second device are federally coordinated so that the first device can perform federal cooperation with the second device based on the federal cooperation sample set and through the federal coordination of the third-party device and generate a federal cooperation result. The specific content of the federal cooperation can be referred to the specific content in step S30.

[0097] This application provides a federated cooperation optimization method, which involves receiving a cooperative dataset and a federated cooperation request from a second device, and then publishing the federated cooperation request to a first device. Upon receiving the federated cooperation request, the first device generates a federated cooperation sample set based on the cooperative dataset. The second device can dynamically update the cooperative dataset based on business dynamic update information to achieve dynamic maintenance of the cooperative dataset. Furthermore, the sample data type in the cooperative dataset can indicate the sample type of the federated cooperation samples selected by the first device, and the target feature version in the cooperative dataset can indicate the feature type of the federated cooperation samples selected by the first device. Therefore, due to the cooperative data... The data is stored on a third-party device and can be dynamically maintained. The first device can easily detect dynamic changes in the business of another federated partner. Then, the first device generates a federated cooperation sample set that matches the business of the second device. Then, federated coordination is performed between the first device and the second device. The first device can then cooperate with the second device based on the federated cooperation sample set to generate federated cooperation results. This overcomes the technical defects of the federated cooperation, which is often inefficient because the business information of the participating parties is usually dynamic and cannot be detected by account information. This often requires the participating parties to negotiate offline to solve the problem of mismatched business information, thus improving the efficiency of federated cooperation.

[0098] Reference Figure 4 , Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of this application.

[0099] like Figure 4 As shown, the federated cooperative optimization device may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to establish communication between the processor 1001 and the memory 1005. The memory 1005 may be high-speed RAM or stable non-volatile memory, such as disk storage. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0100] Optionally, the federally optimized device may also include a rectangular user interface, a network interface, a camera, RF (Radio Frequency) circuitry, sensors, audio circuitry, a WiFi module, etc. The rectangular user interface may include a display screen and an input submodule such as a keyboard; optionally, the rectangular user interface may also include standard wired or wireless interfaces. The network interface may optionally include standard wired or wireless interfaces (such as a Wi-Fi interface).

[0101] Those skilled in the art will understand that Figure 4 The federal cooperative optimization device structure shown in the figure does not constitute a limitation on the federal cooperative optimization device, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0102] like Figure 4 As shown, the memory 1005, as a computer storage medium, may include an operating system, a network communication module, and a federated cooperation optimization program. The operating system is a program that manages and controls the hardware and software resources of the federated cooperation optimization device, supporting the execution of the federated cooperation optimization program and other software and / or programs. The network communication module is used to enable communication between the various components within the memory 1005, as well as communication with other hardware and software in the federated cooperation optimization system.

[0103] exist Figure 4 In the federated cooperation optimization device shown, the processor 1001 is used to execute the federated cooperation optimization program stored in the memory 1005 to implement the steps of the federated cooperation optimization method described in any of the above claims.

[0104] The specific implementation of the federal cooperative optimization device in this application is basically the same as the embodiments of the above-mentioned federal cooperative optimization method, and will not be repeated here.

[0105] Reference Figure 4 , Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of this application.

[0106] like Figure 4 As shown, the federated cooperative optimization device may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to establish communication between the processor 1001 and the memory 1005. The memory 1005 may be high-speed RAM or stable non-volatile memory, such as disk storage. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0107] Optionally, the federally optimized device may also include a rectangular user interface, a network interface, a camera, RF (Radio Frequency) circuitry, sensors, audio circuitry, a WiFi module, etc. The rectangular user interface may include a display screen and an input submodule such as a keyboard; optionally, the rectangular user interface may also include standard wired or wireless interfaces. The network interface may optionally include standard wired or wireless interfaces (such as a Wi-Fi interface).

[0108] Those skilled in the art will understand that Figure 4 The federal cooperative optimization device structure shown in the figure does not constitute a limitation on the federal cooperative optimization device, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0109] like Figure 4 As shown, the memory 1005, as a computer storage medium, may include an operating system, a network communication module, and a federated cooperation optimization program. The operating system is a program that manages and controls the hardware and software resources of the federated cooperation optimization device, supporting the execution of the federated cooperation optimization program and other software and / or programs. The network communication module is used to enable communication between the various components within the memory 1005, as well as communication with other hardware and software in the federated cooperation optimization system.

[0110] exist Figure 4 In the federated cooperation optimization device shown, the processor 1001 is used to execute the federated cooperation optimization program stored in the memory 1005 to implement the steps of the federated cooperation optimization method described in any of the above claims.

[0111] The specific implementation of the federal cooperative optimization device in this application is basically the same as the embodiments of the above-mentioned federal cooperative optimization method, and will not be repeated here.

[0112] This application embodiment also provides a federated cooperation optimization device, which is applied to a first device, and the federated cooperation optimization device includes:

[0113] The determination module is used to obtain the collaborative dataset dynamically maintained on the third-party device and determine the sample data types supported by the collaborative dataset;

[0114] The generation module is used to select the target feature version in the collaborative dataset and generate a federated collaborative sample set based on the sample data type and the target feature version.

[0115] The federated cooperation module is used to perform federated cooperation with the second device corresponding to the cooperation dataset based on the federated cooperation sample set, and obtain the federated cooperation result.

[0116] Optionally, the generation module is further configured to:

[0117] Select each sample label corresponding to the data label type and each feature label corresponding to the feature label type;

[0118] Each of the aforementioned sample labels and feature labels is used to generate a federal cooperation sample.

[0119] Optionally, the federated cooperation optimization device is further used for:

[0120] Obtain the federated learning system version and verify that the federated learning system version is within the range of system versions supported by the collaborative dataset;

[0121] If so, then perform the step of selecting the target feature version from the collaborative dataset:

[0122] If not, then withdraw from federal cooperation.

[0123] Optionally, the federated cooperation module is also used for:

[0124] Select the federated learning version supported by the collaborative dataset;

[0125] Based on the system resource number matched by the federated learning version and the federated cooperation sample set, federated cooperation is performed with the second device to obtain the federated cooperation result.

[0126] The specific implementation of the federal cooperation optimization device in this application is basically the same as the embodiments of the above-mentioned federal cooperation optimization method, and will not be repeated here.

[0127] This application embodiment also provides a federated cooperation optimization device, which is applied to a second device, and the federated cooperation optimization device includes:

[0128] The upload module is used to generate a collaborative dataset and upload the collaborative dataset to a third-party device;

[0129] The request initiation module is used to initiate a federated cooperation request to the first device through the third-party device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset;

[0130] The federated cooperation module is used to perform federated cooperation by the third party based on a preset local sample set and in conjunction with the federated cooperation sample set in the first device to obtain the federated cooperation result.

[0131] Optionally, the federated cooperation optimization device is further used for:

[0132] Obtain business dynamic update information and send the business dynamic update information to the third-party device, wherein the business dynamic update information is used to instruct the updating of the cooperative dataset.

[0133] The specific implementation of the federal cooperation optimization device in this application is basically the same as the embodiments of the above-mentioned federal cooperation optimization method, and will not be repeated here.

[0134] This application embodiment also provides a federated cooperation optimization device, which is applied to a third-party device, and the federated cooperation optimization device includes:

[0135] The receiving and sending module is used to receive the cooperative dataset and federated cooperation request sent by the second device, and to publish the federated cooperation request to the first device, so that the first device can generate a federated cooperation sample set based on the cooperative dataset after receiving the federated cooperation request;

[0136] The coordination module is used to perform federated coordination between the first device and the second device, so that the first device can perform federated cooperation with the second device based on the federated cooperation sample set and generate federated cooperation results.

[0137] Optionally, the federated cooperation optimization device is further used for:

[0138] Receive service dynamic update information sent by the second device;

[0139] Based on the business dynamic update information, at least one of the target feature version, system version, and sample data type in the collaborative dataset is updated to maintain the collaborative dataset.

[0140] The specific implementation of the federal cooperation optimization device in this application is basically the same as the embodiments of the above-mentioned federal cooperation optimization method, and will not be repeated here.

[0141] This application provides a medium that is a readable storage medium, and the readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the federated cooperative optimization method described in any of the above claims.

[0142] The specific implementation of the readable storage medium in this application is basically the same as the embodiments of the above-described federal cooperation optimization method, and will not be repeated here.

[0143] This application provides a computer program product, which includes one or more computer programs. These computer programs can be executed by one or more processors to implement the steps of the federated cooperative optimization method described above.

[0144] The specific implementation of the computer program product in this application is basically the same as the embodiments of the above-mentioned federal cooperation optimization method, and will not be repeated here.

[0145] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent scope of this application.

Claims

1. A federal cooperative optimization method, characterized in that, The federated cooperation optimization method is applied to the first device, and the federated cooperation optimization method includes: The system acquires a dynamically maintained collaborative dataset on a third-party device and determines the sample data types supported by the collaborative dataset. The collaborative dataset includes a target feature version, system version, collaborative data type information, and collaborative data description information. The target feature version includes a feature tag type, which includes feature tags. The collaborative data type information includes a data tag type, which includes sample tags, which are the identifiers of the samples. The collaborative data description information describes the data required for federated collaboration by the second device, allowing the first device to determine whether it meets the qualifications of the federated collaborator required by the second device. Select the target feature version from the collaborative dataset, and generate a federated collaborative sample set based on the sample data type and the target feature version; Based on the federated cooperation sample set, a second device corresponding to the cooperation dataset performs federated cooperation to obtain federated cooperation results; wherein, the federated cooperation is a cooperation process in federated learning, including an intersection matching process, a federated learning modeling process, and a joint model prediction process.

2. The federated cooperation optimization method as described in claim 1, characterized in that, The sample data type includes data label type, and the federated cooperation sample set includes at least one federated cooperation sample. The step of generating a federated cooperation sample set based on the sample data type and the target feature version includes: Select each sample label corresponding to the data label type and each feature label corresponding to the feature label type; Each of the aforementioned sample labels and feature labels is used to generate a federal cooperation sample.

3. The federated cooperation optimization method as described in claim 1, characterized in that, Following the step of determining the sample data type corresponding to the collaborative dataset, the federated collaboration optimization method further includes: Obtain the federated learning system version and verify that the federated learning system version is within the range of system versions supported by the collaborative dataset; If so, then proceed with the step of selecting the target feature version from the collaborative dataset; If not, then withdraw from federal cooperation.

4. The federated cooperation optimization method as described in claim 1, characterized in that, The step of performing federated cooperation based on the federated cooperation sample set and the second device corresponding to the cooperation dataset to obtain the federated cooperation result includes: Select the federated learning version supported by the collaborative dataset; Based on the system resource number matched by the federated learning version and the federated cooperation sample set, federated cooperation is performed with the second device to obtain the federated cooperation result.

5. A federal cooperative optimization method, characterized in that, The federated cooperation optimization method is applied to the second device, and the federated cooperation optimization method includes: A collaborative dataset is generated and uploaded to a third-party device. The collaborative dataset includes a target feature version, system version, collaborative data type information, and collaborative data description information. The target feature version includes a feature tag type, which includes feature tags. The collaborative data type information includes a data tag type, which includes sample tags, which are the identifiers of the samples. The collaborative data description information describes the data required for federated collaboration by the second device, allowing the first device to determine whether it meets the qualifications of the federated collaborator required by the second device. The third-party device initiates a federated cooperation request to the first device, so that the first device can generate a federated cooperation sample set based on the cooperation dataset; The third party, based on a preset local sample set, performs federated cooperation in conjunction with the federated cooperation sample set in the first device to obtain federated cooperation results; wherein, the federated cooperation is a cooperation process in federated learning, including an intersection matching process, a federated learning modeling process, and a joint model prediction process.

6. The federated cooperation optimization method as described in claim 5, characterized in that, Following the step of uploading the collaborative dataset to a third-party device, the federated collaboration optimization method further includes: Obtain business dynamic update information and send the business dynamic update information to the third-party device, wherein the business dynamic update information is used to instruct the updating of the cooperative dataset.

7. A federal cooperative optimization method, characterized in that, The federated cooperation optimization method is applied to third-party devices, and the federated cooperation optimization method includes: The system receives a collaborative dataset and a federated cooperation request from a second device, and then publishes the federated cooperation request to a first device. Upon receiving the federated cooperation request, the first device generates a federated cooperation sample set based on the collaborative dataset. The collaborative dataset includes a target feature version, a system version, collaborative data type information, and collaborative data description information. The target feature version includes a feature tag type, which includes feature tags. The collaborative data type information includes a data tag type, which includes sample tags, which are the identifiers of the samples. The collaborative data description information describes the data required for federated cooperation by the second device, allowing the first device to determine whether it meets the qualifications of the federated cooperation partner required by the second device. The first device and the second device are federated and coordinated so that the first device can cooperate with the second device in federated cooperation based on the federated cooperation sample set to generate federated cooperation results; wherein, the federated cooperation is a cooperation process in federated learning, including intersection matching process, federated learning modeling process and joint model prediction process.

8. The federated cooperation optimization method as described in claim 7, characterized in that, Following the step of receiving the collaborative dataset sent by the second device, the federated cooperation optimization method further includes: Receive service dynamic update information sent by the second device; Based on the business dynamic update information, at least one of the target feature version, system version, and sample data type in the collaborative dataset is updated to maintain the collaborative dataset.

9. A federal cooperative optimization device, characterized in that, The federated cooperation optimization device includes: a memory, a processor, and a program stored in the memory for implementing the federated cooperation optimization method. The memory is used to store programs that implement the federated cooperative optimization method; The processor is configured to execute a program that implements the federated cooperative optimization method to carry out the steps of the federated cooperative optimization method as described in any one of claims 1 to 4, 5 to 6, or 7 to 8.

10. A medium, said medium being a readable storage medium, characterized in that, The readable storage medium stores a program for implementing the federated cooperative optimization method, which is executed by a processor to implement the steps of the federated cooperative optimization method as described in any one of claims 1 to 4, 5 to 6, or 7 to 8.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the federated cooperative optimization method as described in any one of claims 1 to 4, 5 to 6, or 7 to 8.

Images