Parameter configuration method, device, equipment and storage medium based on federated learning

Abstract

The present application discloses a parameter configuration method, apparatus, device and storage medium based on federated learning, the method comprising: when a first federated modeling task is detected, determining each first task flow component of the first federated modeling task; when a parameter configuration instruction of the first task flow component is detected, obtaining configuration information of each first task flow component obtained through interaction; according to the configuration information of each first task flow component, performing parameter configuration on the first task flow component; if it is detected that all first task flow components have completed parameter configuration, running the first task flow component to complete the first federated modeling task. The present application improves the efficiency of federated modeling and solves the technical problem of low efficiency of federated modeling in the prior art.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology in financial technology (Fintech), and in particular to a parameter configuration method, apparatus, device and storage medium based on federated learning. Background Art

[0002] With the continuous development of financial technology, especially Internet technology finance, more and more technologies (such as distributed, blockchain, artificial intelligence, etc.) are applied in the financial field, but the financial industry also puts forward higher requirements on technology. For example, the financial industry also has higher requirements for parameter configuration based on federated learning.

[0003] In the process of federated learning modeling, currently, the parameters of the federated learning algorithm are configured through code, such as configuring the parameters into a json file, and then initiating a modeling task to complete the federated modeling task. However, the code method of configuring the parameters into a json file, etc., has at least the following problems: first, for more complex modeling tasks, it is inconvenient to edit or check (this is because the amount of code for the algorithm component parameters is too much, such as at least several hundred lines); second, it is only possible to identify whether there are errors in the parameter configuration after the modeling task is initiated, but the running time of the federated task is usually long, which results in a long wait time to see the effect of the current operation, and then modify the parameters to enter the next round of operation; third, it leads to a high learning cost for business modelers, that is, the existing federated parameter configuration has low convenience and long correction time, resulting in the current low efficiency of federated modeling. Summary of the invention

[0004] The main purpose of this application is to provide a parameter configuration method, device, equipment and storage medium based on federated learning, aiming to solve the technical problem of low efficiency of federated modeling in the prior art.

[0005] To achieve the above objectives, the present application provides a parameter configuration method based on federated learning, and the parameter configuration method based on federated learning includes:

[0006] When a first federated modeling task is detected, determining first task flow components of the first federated modeling task;

[0007] When the parameter configuration instruction of the first task flow component is detected, the configuration information of each first task flow component obtained through interaction is acquired;

[0008] According to the configuration information of each first task flow component, parameter configuration is performed on the first task flow component;

[0009] If it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task.

[0010] Optionally, when the parameter configuration instruction of the first task flow component is detected, the step of acquiring the configuration information of each first task flow component obtained through interaction includes:

[0011] When detecting the parameter configuration instruction of the first task flow component, obtaining the operation attribute of the first task flow component;

[0012] If the running attribute is the first attribute, obtaining unified configuration information of each participant obtained through interaction, and using the unified configuration information as configuration information of the first task flow component of the first attribute;

[0013] If the running attribute is the second attribute, the role configuration information of each participant obtained through interaction is acquired, and the configuration information of the first task flow component of the second attribute is obtained according to the role configuration information.

[0014] Optionally, if it is detected that all first task flow components have completed parameter configuration, the step of running the first task flow component to complete the first federated modeling task includes:

[0015] If it is detected that all first task flow components have completed parameter configuration, run the first task flow components;

[0016] During the running of the first task flow component, if a change instruction of the parameter configuration is detected, determining a change parameter configuration of the change instruction;

[0017] A second task flow component caused by the parameter configuration change is obtained to complete the first federated modeling task based on the first task flow component and the second task flow component.

[0018] Optionally, the step of obtaining the second task flow component caused by the parameter configuration change to complete the first federated modeling task based on the first task flow component and the second task flow component includes:

[0019] Running the first task flow constructed by the first task flow component to obtain a first running result;

[0020] Acquire a second task flow component caused by the parameter configuration change, run a second task flow constructed by the second task flow component, and obtain a second running result;

[0021] Comparing and evaluating the first operation result and the second operation result to obtain a comparison result;

[0022] Based on the comparison results, a target model is determined to complete the first federated modeling task.

[0023] Optionally, if it is detected that all first task flow components have completed parameter configuration, after the step of running the first task flow component to complete the first federated modeling task, the method includes:

[0024] If a parameter template export instruction is detected, saving the parameter template;

[0025] If an import instruction of a second federated modeling task having the same structure as the first federated modeling task is detected, the saved parameter template is imported to complete the parameter configuration of the first federated modeling task.

[0026] Optionally, if it is detected that all first task flow components have completed parameter configuration, the step of running the first task flow component to complete the first federated modeling task includes:

[0027] If it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task;

[0028] Among them, the first task flow component includes at least one or more of the following data import component, federated intersection component, federated sampling component, federated normalization component, vertical federated feature binning component, vertical federated feature selection component, encoding component, vertical federated logistic regression component, vertical federated security component, and federated model evaluation component.

[0029] Optionally, when the parameter configuration instruction of the first task flow component is detected, the step of acquiring the configuration information of each first task flow component obtained through interaction includes:

[0030] When the parameter configuration instruction of the first task flow component is detected, acquiring the interactive operation task for each first task flow component;

[0031] The first task flow components are operated according to the interactive operation task to obtain configuration information of the first task flow components.

[0032] The present application also provides a parameter configuration device based on federated learning, and the parameter configuration device based on federated learning includes:

[0033] A first detection module, configured to determine each first task flow component of the first federated modeling task when a first federated modeling task is detected;

[0034] A second detection module is used to obtain configuration information of each first task flow component obtained through interaction when detecting the parameter configuration instruction of the first task flow component;

[0035] A configuration module, used to perform parameter configuration on the first task flow components according to the configuration information of each first task flow component;

[0036] The running module is used to run the first task flow components to complete the first federated modeling task if it is detected that all the first task flow components have completed parameter configuration.

[0037] Optionally, the second detection module includes:

[0038] A first detection unit, configured to obtain an operation attribute of the first task flow component when detecting a parameter configuration instruction of the first task flow component;

[0039] A first acquisition unit is used to acquire unified configuration information of each participant obtained through interaction if the running attribute is a first attribute, and use the unified configuration information as configuration information of a first task flow component of the first attribute;

[0040] The second acquisition unit is used to acquire the role configuration information of each participant obtained through interaction if the running attribute is the second attribute, and obtain the configuration information of the first task flow component of the second attribute according to the role configuration information.

[0041] Optionally, the operation module includes:

[0042] A second detection unit, configured to run the first task flow component if it is detected that all first task flow components have completed parameter configuration;

[0043] A third detection unit is used to determine the changed parameter configuration of the change instruction if a change instruction of the parameter configuration is detected during the running of the first task flow component;

[0044] The third acquisition unit is used to acquire a second task flow component caused by the change of parameter configuration, so as to complete the first federated modeling task based on the first task flow component and the second task flow component.

[0045] Optionally, the third acquiring unit includes:

[0046] An operation subunit, configured to execute a first task flow constructed by the first task flow component to obtain a first operation result;

[0047] an acquisition subunit, configured to acquire a second task flow component caused by the parameter configuration change, and to run a second task flow constructed by the second task flow component to obtain a second running result;

[0048] A comparison and evaluation subunit, used for comparing and evaluating the first operation result and the second operation result to obtain a comparison result;

[0049] A determination subunit is used to determine a target model based on the comparison result to complete the first federated modeling task.

[0050] Optionally, the parameter configuration device based on federated learning further includes:

[0051] A third detection module, configured to save the parameter template if a parameter template export instruction is detected;

[0052] The fourth detection module is used to import the saved parameter template to complete the parameter configuration of the first federal modeling task if an import instruction of the second federal modeling task with the same structure as the first federal modeling task is detected.

[0053] Optionally, the operation module further includes:

[0054] A fourth detection unit, configured to run the first task flow component to complete the first federated modeling task if it is detected that all first task flow components have completed parameter configuration;

[0055] Among them, the first task flow component includes at least one or more of the following data import component, federated intersection component, federated sampling component, federated normalization component, vertical federated feature binning component, vertical federated feature selection component, encoding component, vertical federated logistic regression component, vertical federated security component, and federated model evaluation component.

[0056] Optionally, the first detection module includes:

[0057] a fifth detection unit, configured to acquire an interactive operation task for each first task flow component when detecting a parameter configuration instruction of the first task flow component;

[0058] The operating unit is used to operate each of the first task flow components according to the interactive operation task to obtain configuration information of each of the first task flow components.

[0059] The present application also provides a parameter configuration device based on federated learning, which is a physical device. The parameter configuration device based on federated learning includes: a memory, a processor, and a program of the parameter configuration method based on federated learning stored in the memory and executable on the processor. When the program of the parameter configuration method based on federated learning is executed by the processor, the steps of the parameter configuration method based on federated learning as described above can be implemented.

[0060] The present application also provides a storage medium, on which is stored a program for implementing the above-mentioned parameter configuration method based on federated learning. When the program for the parameter configuration method based on federated learning is executed by a processor, the steps of the parameter configuration method based on federated learning are implemented.

[0061] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned parameter configuration method based on federated learning when executed by a processor.

[0062] The present application provides a parameter configuration method, device, equipment and storage medium based on federated learning. Compared with the current technical problem of configuring federated parameters through code, which leads to low efficiency of federated modeling, the present application determines each first task flow component of the first federated modeling task when a first federated modeling task is detected; when a parameter configuration instruction of the first task flow component is detected, the configuration information of each first task flow component obtained through interaction is obtained; according to the configuration information of each first task flow component, the first task flow component is parameter configured; if it is detected that all first task flow components have completed parameter configuration, the first task flow component is run to complete the first federated modeling task. In the present application, the federal parameters are no longer configured through code, but the configuration information of each first task flow component of the first federated modeling task is obtained through interaction; according to the configuration information of each first task flow component, the first task flow component is parameter configured to run the first task flow component to complete the first federated modeling task. That is, in this application, the parameter configuration of the federated modeling task is directly subdivided into the parameter configuration of the task flow component, and then the parameters of the task flow component are configured in an interactive manner to improve the convenience of parameter configuration and the timeliness of error identification of parameter configuration, thereby improving the efficiency of federated modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0064] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0065] Figure 1 This is a flow chart of the first embodiment of the parameter configuration method based on federated learning of this application;

[0066] Figure 2This is a schematic diagram of a detailed process flow of step S20 in the first embodiment of the parameter configuration method based on federated learning of this application;

[0067] Figure 3 A schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present application;

[0068] Figure 4 This is a scenario diagram of the parameter configuration method based on federated learning in this application.

[0069] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0070] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0071] The present application embodiment provides a parameter configuration method based on federated learning. In the first embodiment of the parameter configuration method based on federated learning of the present application, refer to Figure 1 , the parameter configuration method based on federated learning includes:

[0072] Step S10, when a first federated modeling task is detected, determining each first task flow component of the first federated modeling task;

[0073] Step S20, when the parameter configuration instruction of the first task flow component is detected, obtaining the configuration information of each first task flow component obtained through interaction;

[0074] Step S30, performing parameter configuration on the first task flow components according to the configuration information of each first task flow component;

[0075] Step S40: If it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task.

[0076] The specific steps are as follows:

[0077] Step S10, when a first federated modeling task is detected, determining each first task flow component of the first federated modeling task;

[0078] In this embodiment, a parameter configuration method based on federated learning is applied to a multi-party federated modeling scenario. In the multi-party federated modeling scenario, it is divided into an application party (guest party) and a data source party (host party). The application party is the party with a label, and the data source party is the party without a label. Usually, in the vertical federated modeling process, there can be one guest party and multiple host parties. In this embodiment, it should be noted that the application party and the data source party are both participants. In the parameter configuration method based on federated learning, specifically, it can also be applied to the data source party or the application party, or the parameter configuration method based on federated learning can also be applied to the server side. In this embodiment, the parameter configuration method based on federated learning is applied to the server side as an example for specific description.

[0079] When a first federated modeling task is detected, each first task flow component of the first federated modeling task is determined. Specifically, when the first federated modeling task is detected on the server side, each first task flow component of the first federated modeling task is determined, wherein a user can trigger the first federated modeling task by clicking or touching on a visual interface on the server side, wherein the first federated modeling task carries task-related information, and the task-related information includes: federated participants of the federated task, federated type, federated task flow, federated task flow components, etc. When the server side detects the first federated modeling task, it determines the first task flow of the first federated modeling task, and then determines the first task The first task flow components of the flow, where it should be noted that each first task flow component may be associated with the settings of multiple participant parameters, or may only be associated with the settings of one participant parameters, but in order to ensure that the needs of different federal modeling tasks are met, for each first task flow component, the parameter settings required for multi-party operation are fully configured, for example, for the first task flow component A, all a1, a2, a3, a4, a5, a6, etc. are also configured with b1, b2, b3, b4, b5, b6, etc., where, in order to ensure that each first task flow component meets the needs of different federal modeling tasks, custom parameters can also be set for each first task flow component, or parameter options can be added.

[0080] If it is detected that all first task flow components have completed parameter configuration, the step of running the first task flow component to complete the first federated modeling task includes:

[0081] Step S11, if it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task;

[0082] Among them, the first task flow component includes at least one or more of the following data import component, federated intersection component, federated sampling component, federated normalization component, vertical federated feature binning component, vertical federated feature selection component, encoding component, vertical federated logistic regression component, vertical federated security component, and federated model evaluation component.

[0083] That is, in this embodiment, the first task flow component may specifically include a data import component, which may include parameters such as the missing feature value ratio, minimum / maximum value, median, mean, standard deviation, skewness, kurtosis, etc. The first task flow component may specifically include a federal intersection component, and the parameters of the federal intersection component include identifiers such as user ID and user ID number. The first task flow component may specifically include a federal sampling component, and the parameters of the federal sampling component may be sampling type, sampling index and other parameters. The first task flow component may specifically include a federal normalization component, and the federal normalization component may include parameters such as normalization standard. The first task flow component may specifically include a vertical federal feature binning component, a vertical federal feature selection component, etc. The vertical federal feature binning component, the vertical federal feature selection component, etc. include parameters such as binning conditions and selection conditions. The first task flow component may specifically include a vertical federal logistic regression component, a vertical federal security component, a federal model evaluation component, etc., and the corresponding parameters may include regression conditions, evaluation criteria, etc.

[0084] Step S20, when the parameter configuration instruction of the first task flow component is detected, obtaining the configuration information of each first task flow component obtained through interaction;

[0085] When the parameter configuration instruction of the first task flow component is detected, the configuration information of each first task flow component is obtained. Specifically, when the parameter configuration instruction of the first task flow component is detected, the configuration information of each first task flow component obtained in an interactive manner is obtained, wherein the configuration information of each first task flow component obtained in an interactive manner refers to the specific content of the parameters of each first task flow component obtained in an interactive manner, and the specific range of the parameters, wherein the configuration information of the first task flow component may also include specific configuration participants.

[0086] Reference Figure 2 , when the parameter configuration instruction of the first task flow component is detected, the step of acquiring the configuration information of each first task flow component obtained through interaction includes:

[0087] Step S21, when the parameter configuration instruction of the first task flow component is detected, the operation attribute of the first task flow component is obtained;

[0088] In this embodiment, when the parameter configuration instruction of the first task flow component is detected, the operation attributes of the first task flow component are obtained, wherein the operation attributes are divided into joint operation and operation differentiated by the roles of the participants. It should be noted that the attributes of joint operation can be further divided into full joint operation and partial joint operation.

[0089] Step S22, if the running attribute is the first attribute, obtaining unified configuration information of each participant obtained through interaction, and using the unified configuration information as configuration information of the first task flow component of the first attribute;

[0090] If the running attribute is the first attribute, i.e., the attribute of common running, the unified configuration information of each participant obtained through interaction is obtained, and the unified configuration information is used as the configuration information of the first task flow component of the first attribute. Specifically, if the running attribute is the attribute of full common running in the first attribute, the unified configuration information of each participant obtained through interaction (the user inputs the selected parameters and parameter ranges on the visual interface, etc.) is obtained (select all participants in the participant option), and the unified configuration information is used as the configuration information of the first task flow component of the full common running attribute (each participant runs the first task flow component of the full common running attribute in the same way). flow component), which means that the application party and the data source party jointly run the algorithm component, and the component parameters configured by both parties are consistent; if the running attribute is a partially common running attribute in the first attribute, the unified configuration information of the associated participants obtained through interaction (the user enters the selected parameters and parameter ranges on the visual interface, etc.) is obtained (the associated participants are selected on the participant options), and the unified configuration information is used as the configuration information of the first task flow component of the partially common running attributes (each associated participant runs the first task flow component of the partially common running attributes in the same way), which means that the associated participants jointly run the algorithm component, and the component parameters configured by both parties are consistent.

[0091] Step S23: if the running attribute is the second attribute, obtain the role configuration information of each participant obtained through interaction, and obtain the configuration information of the first task flow component of the second attribute according to the role configuration information.

[0092] If the running attribute is the second attribute, that is, the attribute is distinguished by the role of the participant, the role configuration information of each participant obtained interactively is obtained (the role configuration of each participant selected by the user input on the visual interface, the corresponding parameters and parameter ranges, etc.), and the configuration information of the first task flow component of the second attribute is obtained according to the role configuration information, which means that a certain participant runs the algorithm component with different parameters from other participant components.

[0093] Specifically, when selecting a role, you need to configure different parameters for each role, such as Figure 4 As shown, first, check the partyID (participant ID) of the role to determine whether the party participates in the operation of the component. Secondly, select the parameters to be set for each role and configure the specific values ​​of the parameters. That is, each role needs to select some parameters in the component for configuration, not all parameters necessarily need to be set; thirdly, since the filling of each parameter limits the interval, the range of possible values ​​or the input format, the parameters filled in through the interface completely avoid parameter configuration errors or format errors, and the purpose of parameter verification can be achieved before the task flow starts running.

[0094] In this embodiment, it should be noted that since a certain participant may run an algorithm component with different component parameters from other participants, algorithm components of the same component type but with different component parameters can be run by different participants at the same time. That is, there can be multiple algorithm components of the same type, so that different participants can set them according to their own needs.

[0095] In this embodiment, the configuration information of each first task flow component may be manually input, or may be imported data.

[0096] Step S30, performing parameter configuration on the first task flow components according to the configuration information of each first task flow component;

[0097] After obtaining the configuration information of each first task flow component, parameter configuration is performed on the first task flow component according to the configuration information. Specifically, parameter configuration is performed on the corresponding first task flow component according to the configuration information of different first task flow components.

[0098] Step S40: If it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task.

[0099] If it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task. Specifically, the task flow formed by the first task flow components is run to complete the first federated modeling task.

[0100] If it is detected that all first task flow components have completed parameter configuration, the step of running the first task flow component to complete the first federated modeling task includes:

[0101] Step S41, if it is detected that all first task flow components have completed parameter configuration, run the first task flow components;

[0102] Step S42, during the running of the first task flow component, if a parameter configuration change instruction is detected, determining a change parameter configuration of the change instruction;

[0103] Step S43: obtaining a second task flow component caused by the parameter configuration change, so as to complete the first federated modeling task based on the first task flow component and the second task flow component.

[0104] If it is detected that all first task flow components have completed parameter configuration, run the first task flow component, and during the process of running the first task flow component, if an instruction to change the parameter configuration is detected, determine the changed parameter configuration of the change instruction, wherein after the process of running the first task flow component, if an instruction to change the parameter configuration is detected, the changed parameter configuration of the change instruction can also be determined.

[0105] That is, when the parameters of each component are set, the task flow can be initiated, and the component parameter configuration can be modified at any time. After saving, the task flow can be initiated again (obtaining the second task flow component triggered by the change of the parameter configuration). In this way, multiple task flows can be carried out at the same time, and the results of the model operation can be compared in the end. Or, if the model effect is not good after a task flow is run, the parameters can be adjusted, and the model can be re-run and the results can be compared in the end, that is, the first federal modeling task is completed based on the first task flow component and the second task flow component.

[0106] The present application provides a parameter configuration method, device, equipment and storage medium based on federated learning. Compared with the current technical problem of configuring federated parameters through code, which leads to low efficiency of federated modeling, the present application determines each first task flow component of the first federated modeling task when a first federated modeling task is detected; when a parameter configuration instruction of the first task flow component is detected, the configuration information of each first task flow component obtained through interaction is obtained; according to the configuration information of each first task flow component, the first task flow component is parameter configured; if it is detected that all first task flow components have completed parameter configuration, the first task flow component is run to complete the first federated modeling task. In the present application, the federal parameters are no longer configured through code, but the configuration information of each first task flow component of the first federated modeling task is obtained through interaction; according to the configuration information of each first task flow component, the first task flow component is parameter configured to run the first task flow component to complete the first federated modeling task. That is, in this application, the parameter configuration of the federated modeling task is directly subdivided into the parameter configuration of the task flow component, and then the parameters of the task flow component are configured in an interactive manner to improve the convenience of parameter configuration and the timeliness of error identification of parameter configuration, thereby improving the efficiency of federated modeling.

[0107] The present application embodiment provides a parameter configuration method based on federated learning. In another embodiment of the parameter configuration method based on federated learning of the present application,

[0108] The step of obtaining the second task flow component caused by the parameter configuration change to complete the first federated modeling task based on the first task flow component and the second task flow component includes:

[0109] Step A1, running the first task flow constructed by the first task flow component to obtain a first running result;

[0110] In this embodiment, the first task flow constructed by the first task flow component is run to obtain a first running result. Specifically, the first task flow constructed by the first task flow component is run to obtain a first model based on the first task flow, and a first prediction accuracy of the first model is obtained based on the first model and preset verification data.

[0111] Step A2, obtaining a second task flow component caused by the parameter configuration change, running a second task flow constructed by the second task flow component, and obtaining a second running result;

[0112] Obtain a second task flow component triggered by the change in parameter configuration, run the second task flow constructed by the second task flow component, and obtain a second running result. Specifically, run the second task flow constructed by the second task flow component, obtain a second model based on the second task flow, and obtain a second prediction accuracy of the second model based on the second model and preset verification data.

[0113] Step A3, comparing and evaluating the first operation result and the second operation result to obtain a comparison result;

[0114] The first operation result and the second operation result are compared and evaluated to obtain a comparison result. Specifically, the first prediction accuracy and the second prediction accuracy are compared and evaluated to obtain a comparison result.

[0115] Step A4: determine a target model based on the comparison result to complete the first federated modeling task.

[0116] Based on the comparison results, a target model is determined to complete the first federated modeling task. Based on the comparison results, a model with the highest prediction accuracy is determined as the target model to complete the first federated modeling task. Specifically, if the model with the highest accuracy is the first model, the first model is set as the target model to complete the first federated modeling task.

[0117] In this embodiment, a first task flow constructed by running the first task flow component is obtained to obtain a first operation result; a second task flow component caused by the parameter configuration change is obtained, and a second task flow constructed by the second task flow component is run to obtain a second operation result; the first operation result is compared and evaluated with the second operation result to obtain a comparison result; based on the comparison result, a target model is determined to complete the first federated modeling task. In this embodiment, the target model is accurately determined to complete the first federated modeling task.

[0118] The present application embodiment provides a parameter configuration method based on federated learning. In another embodiment of the parameter configuration method based on federated learning of the present application,

[0119] After the step of running the first task flow component to complete the first federated modeling task if it is detected that all first task flow components have completed parameter configuration, the method includes:

[0120] Step B1, if a parameter template export instruction is detected, save the parameter template;

[0121] Step B2: if an import instruction of a second federated modeling task having the same structure as the first federated modeling task is detected, the saved parameter template is imported to complete the parameter configuration of the first federated modeling task.

[0122] In this embodiment, parameters can also be reused (i.e., model parameters can be imported and exported): that is, if a parameter template export instruction is detected, the parameter template is saved; if an import instruction of a second federated modeling task having the same structure as the first federated modeling task is detected, the saved parameter template is imported to complete the parameter configuration of the first federated modeling task. That is, in this embodiment, the parameter template can be saved and exported, and a parameter configuration file will be generated during the export (which can be saved in multiple formats such as json and python and can be used in various computing environments), so that it can be quickly imported when setting parameters of similar components in other federated modeling tasks without reconfiguration; in addition, parameter saving can save the parameters of a certain component and the parameter configuration of the entire task flow. For example, a modeling task with the same structure (the components and the component connection order are the same) is initiated, and the parameter configuration of a certain task flow is directly imported. On this basis, the parameters are slightly adjusted to achieve the effect of quickly configuring the parameters;

[0123] In this embodiment, if a parameter template export instruction is detected, the parameter template is saved; if an import instruction of a second federated modeling task having the same structure as the first federated modeling task is detected, the saved parameter template is imported to complete the parameter configuration of the first federated modeling task. In this embodiment, parameters can be reused to improve user experience.

[0124] The embodiment of the present application provides a parameter configuration method based on federated learning. In another embodiment of the parameter configuration method based on federated learning of the present application, after the step of performing prediction processing on the data to be processed based on the preset prediction model to obtain the prediction result, the method includes:

[0125] The step of acquiring the configuration information of each first task flow component obtained through interaction when the parameter configuration instruction of the first task flow component is detected includes:

[0126] Step C1, when the parameter configuration instruction of the first task flow component is detected, obtaining the interactive operation task for each first task flow component;

[0127] Step C2: operate each of the first task flow components according to the interactive operation task to obtain configuration information of each of the first task flow components.

[0128] In this embodiment, when the parameter configuration instructions of the first task flow component are detected, the interactive operation tasks for each first task flow component are obtained, and the first task flow components are operated according to the interactive operation tasks to obtain the configuration information of each first task flow component. That is, in this implementation, the first task flow components are operated through the interactive operation tasks to obtain the configuration information of each first task flow component to improve interactivity.

[0129] Reference Figure 3 , Figure 3 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present application.

[0130] like Figure 3 As shown, the parameter configuration device based on federated learning may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the processor 1001 and the memory 1005. The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0131] Optionally, the parameter configuration device based on federated learning may also include a rectangular user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, etc. The rectangular user interface may include a display screen (Display), an input submodule such as a keyboard (Keyboard), and the optional rectangular user interface may also include a standard wired interface and a wireless interface. The network interface may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0132] Those skilled in the art will understand that Figure 3 The structure of the parameter configuration device based on federated learning shown in the figure does not constitute a limitation of the parameter configuration device based on federated learning, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0133] like Figure 3 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, and a parameter configuration program based on federated learning. The operating system is a program that manages and controls the hardware and software resources of the parameter configuration device based on federated learning, and supports the operation of the parameter configuration program based on federated learning and other software and / or programs. The network communication module is used to realize the communication between the components inside the memory 1005, and the communication with other hardware and software in the parameter configuration system based on federated learning.

[0134] exist Figure 3 In the parameter configuration device based on federated learning shown, the processor 1001 is used to execute the parameter configuration program based on federated learning stored in the memory 1005 to implement the steps of any of the above-mentioned parameter configuration methods based on federated learning.

[0135] The specific implementation of the parameter configuration device based on federated learning in the present application is basically the same as the above-mentioned embodiments of the parameter configuration method based on federated learning, and will not be repeated here.

[0136] The present application also provides a parameter configuration device based on federated learning, and the parameter configuration device based on federated learning includes:

[0137] A first detection module, configured to determine each first task flow component of the first federated modeling task when a first federated modeling task is detected;

[0138] A second detection module is used to obtain configuration information of each first task flow component obtained through interaction when detecting the parameter configuration instruction of the first task flow component;

[0139] A configuration module, used to perform parameter configuration on the first task flow components according to the configuration information of each first task flow component;

[0140] The running module is used to run the first task flow components to complete the first federated modeling task if it is detected that all the first task flow components have completed parameter configuration.

[0141] Optionally, the second detection module includes:

[0142] A first detection unit, configured to obtain an operation attribute of the first task flow component when detecting a parameter configuration instruction of the first task flow component;

[0143] A first acquisition unit is used to acquire unified configuration information of each participant obtained through interaction if the running attribute is a first attribute, and use the unified configuration information as configuration information of a first task flow component of the first attribute;

[0144] The second acquisition unit is used to acquire the role configuration information of each participant obtained through interaction if the running attribute is the second attribute, and obtain the configuration information of the first task flow component of the second attribute according to the role configuration information.

[0145] Optionally, the operation module includes:

[0146] A second detection unit, configured to run the first task flow component if it is detected that all first task flow components have completed parameter configuration;

[0147] A third detection unit is used to determine the changed parameter configuration of the change instruction if a change instruction of the parameter configuration is detected during the running of the first task flow component;

[0148] The third acquisition unit is used to acquire a second task flow component caused by the change of parameter configuration, so as to complete the first federated modeling task based on the first task flow component and the second task flow component.

[0149] Optionally, the third acquiring unit includes:

[0150] An operation subunit, configured to execute a first task flow constructed by the first task flow component to obtain a first operation result;

[0151] an acquisition subunit, configured to acquire a second task flow component caused by the parameter configuration change, and to run a second task flow constructed by the second task flow component to obtain a second running result;

[0152] A comparison and evaluation subunit, used for comparing and evaluating the first operation result and the second operation result to obtain a comparison result;

[0153] A determination subunit is used to determine a target model based on the comparison result to complete the first federated modeling task.

[0154] Optionally, the parameter configuration device based on federated learning further includes:

[0155] A third detection module, configured to save the parameter template if a parameter template export instruction is detected;

[0156] The fourth detection module is used to import the saved parameter template to complete the parameter configuration of the first federal modeling task if an import instruction of the second federal modeling task with the same structure as the first federal modeling task is detected.

[0157] Optionally, the operation module further includes:

[0158] A fourth detection unit, configured to run the first task flow component to complete the first federated modeling task if it is detected that all first task flow components have completed parameter configuration;

[0159] Among them, the first task flow component includes at least one or more of the following data import component, federated intersection component, federated sampling component, federated normalization component, vertical federated feature binning component, vertical federated feature selection component, encoding component, vertical federated logistic regression component, vertical federated security component, and federated model evaluation component.

[0160] Optionally, the first detection module includes:

[0161] a fifth detection unit, configured to acquire an interactive operation task for each first task flow component when detecting a parameter configuration instruction of the first task flow component;

[0162] The operating unit is used to operate each of the first task flow components according to the interactive operation task to obtain configuration information of each of the first task flow components.

[0163] The specific implementation of the parameter configuration device based on federated learning in the present application is basically the same as the various embodiments of the parameter configuration method based on federated learning mentioned above, and will not be repeated here.

[0164] An embodiment of the present application provides a storage medium, and the storage medium stores one or more programs, and the one or more programs can also be executed by one or more processors to implement the steps of any of the above-mentioned parameter configuration methods based on federated learning.

[0165] The specific implementation method of the storage medium of the present application is basically the same as the above-mentioned embodiments of the parameter configuration method based on federated learning, and will not be repeated here.

[0166] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned parameter configuration method based on federated learning when executed by a processor.

[0167] The specific implementation of the computer program product of the present application is basically the same as the various embodiments of the parameter configuration method based on federated learning described above, and will not be repeated here.

[0168] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A parameter configuration method based on federated learning, characterized in that: Applied to the server side, the parameter configuration method based on federated learning includes: When a first federated modeling task is detected, first task flow components of the first federated modeling task are determined, wherein the first federated modeling task is triggered by a user through a visual interface of the server, wherein the first task flow components include at least one or more of the following: a data import component, a federated intersection component, a federated sampling component, a federated normalization component, a vertical federated feature binning component, a vertical federated feature selection component, an encoding component, a vertical federated logistic regression component, a vertical federated security component, and a federated model evaluation component; When the parameter configuration instruction of the first task flow component is detected, configuration information of each first task flow component obtained through interaction is obtained, wherein the configuration information includes parameters and parameter ranges, and the interaction includes the user inputting the selected parameters and parameter ranges on the visual interface; According to the configuration information of each first task flow component, parameter configuration is performed on the first task flow component; When it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task.

2. The parameter configuration method based on federated learning as claimed in claim 1, characterized in that: The step of acquiring the configuration information of each first task flow component obtained through interaction when the parameter configuration instruction of the first task flow component is detected includes: When detecting the parameter configuration instruction of the first task flow component, obtaining the operation attribute of the first task flow component; When the running attribute is the first attribute, obtaining unified configuration information of each participant obtained through interaction, and using the unified configuration information as configuration information of the first task flow component of the first attribute; When the running attribute is the second attribute, role configuration information of each participant obtained through interaction is acquired, and configuration information of the first task flow component of the second attribute is obtained according to the role configuration information.

3. The parameter configuration method based on federated learning as claimed in claim 1, characterized in that: The step of running the first task flow component to complete the first federated modeling task when it is detected that all first task flow components have completed parameter configuration includes: When it is detected that all first task flow components have completed parameter configuration, running the first task flow components; During the running of the first task flow component, when a change instruction of the parameter configuration is detected, determining a change parameter configuration of the change instruction; A second task flow component caused by the parameter configuration change is obtained to complete the first federated modeling task based on the first task flow component and the second task flow component.

4. The parameter configuration method based on federated learning as claimed in claim 3, characterized in that: The step of obtaining the second task flow component caused by the parameter configuration change to complete the first federated modeling task based on the first task flow component and the second task flow component includes: Running the first task flow constructed by the first task flow component to obtain a first running result; Acquire a second task flow component caused by the parameter configuration change, run a second task flow constructed by the second task flow component, and obtain a second running result; Comparing and evaluating the first operation result and the second operation result to obtain a comparison result; Based on the comparison results, a target model is determined to complete the first federated modeling task.

5. The parameter configuration method based on federated learning according to any one of claims 1 to 4, characterized in that: After the step of running the first task flow component to complete the first federated modeling task when it is detected that all first task flow components have completed parameter configuration, the method includes: When a parameter template export instruction is detected, saving the parameter template; When an import instruction of a second federated modeling task having the same structure as the first federated modeling task is detected, the saved parameter template is imported to complete the parameter configuration of the first federated modeling task.

6. The parameter configuration method based on federated learning as claimed in claim 1, characterized in that: The step of running the first task flow component to complete the first federated modeling task when it is detected that all first task flow components have completed parameter configuration includes: When it is detected that all first task flow components have completed parameter configuration, the first task flow components are run to complete the first federated modeling task.

7. The parameter configuration method based on federated learning as claimed in claim 1, characterized in that: The step of acquiring the configuration information of each first task flow component obtained through interaction when the parameter configuration instruction of the first task flow component is detected includes: When the parameter configuration instruction of the first task flow component is detected, acquiring the interactive operation task for each first task flow component; The first task flow components are operated according to the interactive operation task to obtain configuration information of the first task flow components.

8. A parameter configuration device based on federated learning, characterized in that: Applied to the server side, the parameter configuration device based on federated learning includes: A first detection module is used to determine each first task flow component of the first federated modeling task when a first federated modeling task is detected, wherein the first federated modeling task is triggered by a user through a visual interface of the server, wherein the first task flow component includes at least one or more of the following: a data import component, a federated intersection component, a federated sampling component, a federated normalization component, a vertical federated feature binning component, a vertical federated feature selection component, an encoding component, a vertical federated logistic regression component, a vertical federated security component, and a federated model evaluation component; A second detection module is used to obtain configuration information of each first task flow component obtained through interaction when a parameter configuration instruction of the first task flow component is detected, wherein the configuration information includes parameters and parameter ranges, and the interaction includes a user inputting the selected parameters and parameter ranges on the visual interface; A configuration module, used to perform parameter configuration on the first task flow components according to the configuration information of each first task flow component; The running module is used to run the first task flow components to complete the first federated modeling task when it is detected that all the first task flow components have completed parameter configuration.

9. A parameter configuration device based on federated learning, characterized in that: The parameter configuration device based on federated learning includes: a memory, a processor, and a program stored in the memory for implementing the parameter configuration method based on federated learning. The memory is used to store a program for implementing a parameter configuration method based on federated learning; The processor is used to execute a program for implementing the parameter configuration method based on federated learning, so as to implement the steps of the parameter configuration method based on federated learning as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a program for implementing a parameter configuration method based on federated learning, and the program for implementing a parameter configuration method based on federated learning is executed by a processor to implement the steps of the parameter configuration method based on federated learning as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images