Joint learning task configuration method and device, electronic equipment and storage medium

Abstract

The present disclosure provides a joint learning task configuration method and device, electronic equipment and storage medium. The method comprises: receiving a task creation request sent by a participant, in response to the task creation request, setting a new joint learning task scenario, and obtaining a parameter configuration library corresponding to the new joint learning task scenario; according to the new joint learning task scenario, calling the parameters in the parameter configuration library that match the new joint learning task scenario, wherein the parameters include basic parameters and hyperparameters; performing a basic parameter configuration operation on the new joint learning task using the basic parameters, and performing a hyperparameter configuration operation on the new joint learning task using the hyperparameters, and according to the parameter adjustment request sent by the participant, adjusting the hyperparameters based on the preset debugging range. The present disclosure can quickly configure the parameters of the joint learning task for different joint learning scenarios, improve the efficiency of joint learning and the joint learning experience of users.

Description

Technical Field

[0001] This disclosure relates to the field of federated learning technology, and in particular to a federated learning task configuration method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the development of artificial intelligence and distributed machine learning technologies, joint learning methods, which involve different stakeholders in machine learning, have become a mainstream trend in training AI models. As a novel distributed machine learning framework, joint learning meets the needs of multiple clients for model training while maintaining data security. In joint learning platforms, how to achieve rapid and personalized joint learning task configuration has gradually become one of the key issues of concern in joint learning.

[0003] In existing technologies, traditional federated learning management platforms cannot differentiate between different federated learning scenarios. Users need to determine from all federated learning tasks whether there are suitable tasks to add. If no suitable tasks are available, users need to create tasks. However, the parameter configuration methods for newly created tasks in existing federated learning task creation processes are relatively simple, cannot be combined with the federated learning scenario, and require high levels of user expertise, making the operation difficult. This not only reduces the efficiency of configuring federated learning tasks but also degrades the user's federated learning experience.

[0004] In view of the problems in the existing technology, there is a need to provide a solution that can combine joint learning scenarios with parameter configuration, quickly configure the parameters of joint learning tasks for different joint learning scenarios, and improve the efficiency of joint learning task configuration. Summary of the Invention

[0005] In view of this, the present disclosure provides a method, apparatus, electronic device and storage medium for configuring federated learning tasks, in order to solve the problems of existing technologies where parameter configuration cannot be combined with federated learning scenarios, parameter configuration is difficult, reducing the efficiency of users in configuring federated learning task parameters, and thus reducing the user's federated learning experience.

[0006] A first aspect of this disclosure provides a method for configuring a joint learning task, comprising: receiving a task creation request sent by a participant; responding to the task creation request, setting a scenario for a new joint learning task and obtaining a parameter configuration library corresponding to the scenario of the new joint learning task; retrieving parameters from the parameter configuration library that match the scenario of the new joint learning task, wherein the parameters include basic parameters and hyperparameters; performing basic parameter configuration operations on the new joint learning task using the basic parameters, and performing hyperparameter configuration operations on the new joint learning task using the hyperparameters, and, when performing hyperparameter configuration operations on the new joint learning task, adjusting the hyperparameters based on a preset adjustment range according to a parameter tuning request sent by a participant.

[0007] A second aspect of this disclosure provides a joint learning task configuration apparatus, comprising: a receiving module configured to receive a task creation request sent by a participant, and in response to the task creation request, set a scenario for a new joint learning task and obtain a parameter configuration library corresponding to the scenario of the new joint learning task; a retrieval module configured to retrieve parameters from the parameter configuration library that match the scenario of the new joint learning task, wherein the parameters include basic parameters and hyperparameters; and a configuration module configured to perform basic parameter configuration operations on the new joint learning task using the basic parameters and perform hyperparameter configuration operations on the new joint learning task using the hyperparameters, and, when performing hyperparameter configuration operations on the new joint learning task, adjust the hyperparameters based on a preset adjustment range according to a parameter adjustment request sent by a participant.

[0008] The above-described at least one technical solution adopted in the embodiments of this disclosure can achieve the following beneficial effects:

[0009] By receiving a task creation request from a participant, and responding to the request, setting a scenario for a new joint learning task, and obtaining the parameter configuration library corresponding to the scenario; based on the scenario of the new joint learning task, retrieving parameters from the parameter configuration library that match the scenario, including basic parameters and hyperparameters; performing basic parameter configuration operations on the new joint learning task using the basic parameters, and performing hyperparameter configuration operations on the new joint learning task using the hyperparameters; and while performing hyperparameter configuration operations on the new joint learning task, adjusting the hyperparameters based on a preset debugging range according to the parameter tuning requests sent by the participants. This disclosure combines joint learning scenarios with parameter configuration, enabling rapid configuration of joint learning task parameters for different joint learning scenarios, improving the efficiency of joint learning task configuration, and thus enhancing the user's joint learning experience. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of a joint learning architecture provided in an embodiment of this disclosure;

[0012] Figure 2 This is a flowchart illustrating the joint learning task configuration method provided in this embodiment of the disclosure;

[0013] Figure 3 This is a schematic diagram of the structure of the joint learning task configuration device provided in the embodiments of this disclosure;

[0014] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this disclosure. Detailed Implementation

[0015] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of this disclosure. However, those skilled in the art will understand that this disclosure may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this disclosure with unnecessary detail.

[0016] Federation learning refers to the comprehensive utilization of multiple AI (Artificial Intelligence) technologies, under the premise of ensuring data security and user privacy, to collaboratively explore the value of data and foster new intelligent business forms and models based on joint modeling. Federation learning has at least the following characteristics:

[0017] (1) Participating nodes control their own data in a weakly centralized joint training mode to ensure data privacy and security in the process of co-creating intelligence.

[0018] (2) In different application scenarios, various model aggregation optimization strategies are established by using screening and / or combination of AI algorithms and privacy-preserving computing to obtain high-level and high-quality models.

[0019] (3) Under the premise of ensuring data security and user privacy, based on multiple model aggregation optimization strategies, obtain methods to improve the performance of the federated learning engine. The performance methods can be improved by solving problems such as parallel computing architecture, information interaction under large-scale cross-domain networks, intelligent perception, and anomaly handling mechanisms.

[0020] (4) Obtain the needs of multiple users in various scenarios, determine the true contribution of each joint participant through a mutual trust mechanism, and allocate incentives accordingly.

[0021] Based on the above approach, an AI technology ecosystem based on collaborative learning can be established, fully leveraging the value of industry data and promoting the implementation of scenarios in vertical fields.

[0022] Figure 1 This is a schematic diagram of a joint learning architecture provided in an embodiment of this disclosure. Figure 1 As shown, the architecture of joint learning may include a server (central node) 101 and participants 102, 103 and 104.

[0023] In the joint learning process, a basic model can be established through server 101, which then sends this model to participants 102, 103, and 104 with whom it has established a communication connection. Alternatively, any participant can establish the basic model and upload it to server 101, which then sends it to other participants with whom it has established a communication connection. Participants 102, 103, and 104 construct models based on the downloaded basic structure and model parameters, train the models using local data, obtain updated model parameters, and encrypt and upload these updated model parameters to server 101. Server 101 aggregates the model parameters sent by participants 102, 103, and 104 to obtain global model parameters, which are then transmitted back to participants 102, 103, and 104. Participants 102, 103, and 104 iterate on their respective models based on the received global model parameters until the models converge, thus achieving model training. During the collaborative learning process, the data uploaded by participants 102, 103, and 104 are model parameters. Local data is not uploaded to server 101, and all participants can share the final model parameters. Therefore, collaborative modeling can be achieved while ensuring data privacy.

[0024] It should be noted that the number of participants is not limited to the three mentioned above, but can be set as needed, and this embodiment does not impose any restrictions on this.

[0025] A method and apparatus for configuring collaborative learning tasks according to embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings.

[0026] Figure 2 This is a flowchart illustrating the joint learning task configuration method provided in this embodiment. Figure 2 The configuration method for joint learning tasks can be executed by the joint learning server. For example... Figure 2As shown, the joint learning task configuration method may specifically include:

[0027] S201, Receive the task creation request sent by the participant, respond to the task creation request, set the scenario of the new joint learning task, and obtain the parameter configuration library corresponding to the scenario of the new joint learning task.

[0028] S202, based on the scenario of the new joint learning task, retrieve the parameters from the parameter configuration library that match the scenario of the new joint learning task, including basic parameters and hyperparameters;

[0029] S203, perform basic parameter configuration operations on the new joint learning task using basic parameters, and perform hyperparameter configuration operations on the new joint learning task using hyperparameters. When performing hyperparameter configuration operations on the new joint learning task, adjust the hyperparameters based on the parameter tuning requests sent by the participants and within a preset tuning range.

[0030] Specifically, this embodiment is based on a federated learning platform, which mainly consists of the following modules: a circle selection module, a data addition module, a task viewing module, a task creation module, a training module, a training result display module, and a model repository. The federated learning platform can be considered as a client running on the devices of each participating party. Each participating party can use the federated learning platform to select federated learning scenarios, upload training data, view the federated learning process, create federated learning tasks, and perform inference and prediction. The federated learning platform provides a unified federated learning engine based on the Industrial Internet of Things (IIoT), enabling it to redefine federated learning products and services from the perspective of ecosystem users, allowing users to experience federated learning capabilities on the platform.

[0031] Furthermore, the joint learning task list is a list displayed on the joint learning platform interface to showcase joint learning tasks. The joint learning task list includes not only the name of the joint learning task, but also, but is not limited to, the following: task initiator, task scenario, number of participants, joint learning type, joint learning algorithm, task creation time, task status, and task operation buttons. It should be noted that, in this embodiment, the scenario of a joint learning task can also be referred to as a "circle," meaning one circle corresponds to one joint learning scenario.

[0032] Furthermore, in this embodiment, the user is the participant in the federated learning, and each participant corresponds to a node in the federated learning framework. Each node corresponds to a participant device, which can be a PC, tablet, smartphone, smart wearable device, etc. Each participant device has a client terminal for the federated learning participant, but the participant device is not limited to the aforementioned devices or clients. The federated learning framework also has nodes that provide services to the clients (i.e., servers). The server can be a server used to perform aggregation operations. The server can coordinate multiple clients to perform federated learning to obtain a federated learning model. The server can be an independent physical server, a server cluster composed of multiple physical servers, or a cloud computing server.

[0033] According to the technical solution provided in this disclosure, by receiving a task creation request sent by a participant, and responding to the task creation request, a new scenario for a joint learning task is set, and a parameter configuration library corresponding to the scenario of the new joint learning task is obtained; according to the scenario of the new joint learning task, parameters matching the scenario of the new joint learning task are retrieved from the parameter configuration library, wherein the parameters include basic parameters and hyperparameters; basic parameter configuration operations are performed on the new joint learning task using the basic parameters, and hyperparameter configuration operations are performed on the new joint learning task using the hyperparameters; and when performing hyperparameter configuration operations on the new joint learning task, the hyperparameters are tuned based on a preset tuning range according to the tuning request sent by the participant. This disclosure can combine joint learning scenarios with parameter configuration, enabling rapid configuration of joint learning task parameters for different joint learning scenarios, improving the efficiency of joint learning task configuration, and thus enhancing the user's joint learning experience.

[0034] In some embodiments, before receiving a task creation request sent by a participant, the method further includes: receiving a task filtering request sent by a participant through a client; responding to the task filtering request; filtering the joint learning tasks in the joint learning task list using preset judgment conditions to obtain joint learning tasks that meet the judgment conditions; and determining whether to create a new joint learning task based on the task status of the joint learning tasks that meet the judgment conditions; wherein the judgment conditions include whether the circle matches, whether the task is allowed to be automatically added, and whether the task is a task in a training state.

[0035] Specifically, the joint learning task interface also includes an option box for "Quickly join training tasks". When the user checks this option box, a second screening will be performed based on the results of the first screening. That is, joint learning tasks that meet the task screening rules will be screened a second time, and finally, joint tasks that meet the user's own conditions will be selected.

[0036] Furthermore, the preset judgment conditions include, but are not limited to, the following: whether the circles match, whether the task allows automatic joining, and the current status of the task being "not yet started training". When using the preset judgment conditions to filter the first joint learning task, the platform will automatically obtain each first joint learning task and compare it with the judgment conditions based on the current information of each first joint learning task, thereby filtering out the second joint learning tasks that meet the judgment conditions. For example: when the quick join condition is that the initiator and the participants are in the same task circle, the initiator has set the option to allow participants to join, and the task status is "not yet started training".

[0037] Furthermore, when determining whether to create a new joint learning task based on the task status of the joint learning task that meets the judgment criteria, if the filtered list of joint learning tasks contains joint learning tasks that allow participants to join, then the determination is made whether to create a new joint learning task based on the joint learning tasks that allow participants to join; or, if the filtered list of joint learning tasks does not contain joint learning tasks that allow participants to join, then the determination is made whether to create a new joint learning task.

[0038] It should be noted that when a user does not have a suitable task to join directly, a new joint learning task can be created. The creation of a new task mainly involves the following two steps:

[0039] Step a: Task configuration management. On the configuration page, users only need to fill in the task name (maximum 10 characters), select the application scenario (i.e., the circle), and associate the training data. After the user selects the application scenario, most of the basic parameter configurations can be automatically configured by the system.

[0040] Step b: When selecting participants for collaborative learning, there are two modes: Mode 1 uses automatic matching of participants by the system, and Mode 2 uses self-selection of participants. The following detailed explanations of these two methods for selecting participants in collaborative learning, using specific examples, illustrate these methods.

[0041] In some embodiments, in response to a task creation request, setting a scenario for a new joint learning task includes: determining a joint learning task scenario corresponding to the task scenario identifier in the task creation request, and setting the joint learning task scenario as the scenario corresponding to the new joint learning task.

[0042] Specifically, before configuring parameters, users can select the circle for a new joint learning task according to their needs. In practical applications, the circles for joint learning tasks include electricity load prediction scenarios, gas load prediction scenarios, electricity and gas detection scenarios, and soft measurement scenarios. In addition to selecting the circle, users can also enter the task name of the new joint learning task and associate it with training data.

[0043] In some embodiments, according to the scenario of the new joint learning task, retrieving parameters from the parameter configuration library that match the scenario of the new joint learning task includes: obtaining the task scenario identifier of the new joint learning task, searching for parameters in the parameter configuration library according to the task scenario identifier, and obtaining parameters corresponding to the task scenario identifier, wherein the parameter configuration library contains a mapping relationship between task scenario identifiers and parameters.

[0044] Specifically, when creating a new joint learning task, task configuration (i.e., parameter configuration) is required. If the user has already selected a circle, the system will automatically configure the relevant parameters corresponding to that circle. In practical applications, a parameter configuration library is pre-set, storing the association between the identifier of each circle (i.e., joint learning scenario) and the task parameters. Based on the circle selected by the user, the system automatically matches and loads the corresponding task parameters, thereby completing the parameter configuration operation for the new joint learning task.

[0045] In some embodiments, a basic parameter configuration operation is performed on a new joint learning task using basic parameters, including: obtaining the field type corresponding to the basic parameters, and automatically selecting the field options corresponding to the field type of the basic parameters using preset default parameter settings, so as to obtain the basic parameter configuration result; wherein, the field type corresponding to the basic parameters includes joint learning type, whether the task is public, algorithm type, aggregation method, encryption method, model compression, model evaluation index, termination condition, and participant selection.

[0046] Specifically, each basic parameter corresponds to a field type and field options. Based on the scenario of the new joint learning task, the default parameter settings corresponding to that scenario are selected from a set of preset default parameter settings, and the field options for the basic parameters are selected based on these default settings. In practical applications, task parameters include basic parameters and hyperparameters. The basic parameters of a new joint learning task include, but are not limited to, the following: joint learning type, whether the task is public, algorithm type, aggregation method, encryption method, model compression, model evaluation metrics, termination conditions, and participant selection.

[0047] In some embodiments, performing hyperparameter configuration operations on a new joint learning task using hyperparameters includes: receiving a hyperparameter configuration request sent by a participant through a client; responding to the hyperparameter configuration request; and configuring the hyperparameters of the new joint learning task using hyperparameters retrieved from a parameter configuration library; wherein the hyperparameters include the number of training epochs, learning rate, sample sampling rate, feature sampling rate, maximum tree depth, L1 regularization coefficient, L2 regularization coefficient, and minimum loss function decrease value required for node splitting.

[0048] Specifically, after configuring the basic parameters, hyperparameters need to be configured for the new joint learning task. Similar to the basic parameter configuration, the system will provide default values ​​for the hyperparameters based on the joint learning scenario selected by the user, as well as limitations or ranges for manual debugging. In practical applications, hyperparameter configuration mainly includes configuring the following parameters: number of training epochs, learning rate, sample sampling rate, feature sampling rate, maximum tree depth, L1 regularization coefficient, L2 regularization coefficient, and minimum loss function decrease value required for node splitting.

[0049] In some embodiments, the hyperparameters are tuned based on a preset tuning range according to the tuning request sent by the participants, including: retrieving the preset tuning range according to the tuning request in the hyperparameter configuration request, determining the tuning method according to the tuning request, and tuning the hyperparameters within the tuning range using the tuning method; wherein the tuning method includes automatic tuning and manual tuning.

[0050] Specifically, after the hyperparameters are configured, they can be tuned, meaning the default values ​​can be adjusted within a preset range. In practical applications, the default values ​​and tuning ranges of hyperparameters differ for different joint learning scenarios. Therefore, to determine the default values ​​and tuning ranges for each hyperparameter, it is necessary to pre-collect hyperparameter configuration documents for different application scenarios. Based on the preset default values ​​and tuning ranges in the hyperparameter configuration documents, the new joint learning task in the joint learning scenario is tuned.

[0051] Furthermore, in the embodiments of this disclosure, the following two different parameter tuning methods can be adopted: manual parameter tuning and automatic parameter tuning. Manual parameter tuning is when the user manually adjusts the value of the hyperparameter based on experience, while automatic parameter tuning is when the system automatically adjusts the value of the hyperparameter within a preset parameter tuning range until the ideal state is achieved.

[0052] It should be noted that the above embodiments illustrate the principles and processes of parameter configuration (basic parameter configuration and hyperparameter configuration) during the creation of a joint learning task. Based on the content of the aforementioned embodiments, the parameters of a newly created joint learning task can be reasonably configured. In addition, for a complete joint learning task creation process, besides task parameter configuration, it should also include the selection of joint learning participants. The following, with reference to specific embodiments, details the process of selecting joint learning participants after task parameter configuration, which may include the following:

[0053] In some embodiments, an automatic matching method is used to select corresponding participants for a new joint learning task, including: determining the data contribution of each participant, sorting the participants according to their data contribution, and automatically matching corresponding participants for the new joint learning task based on the sorting results and a preset number of participants.

[0054] Specifically, when using the automatic participant matching method in Mode 1, the following matching strategy can be employed: participants are sorted according to their data contribution, with the number of participants set as a default value; based on the sorting and the default number of participants, a certain number of participants are selected as participants for the new joint learning task. It should be noted that if the user previously selected a test experience, the system will directly display the relevant test participants and begin training without requiring manual selection of participants.

[0055] In some embodiments, a self-selection method is used to select corresponding participants for a new joint learning task, including: screening all participants in the joint learning platform according to the scenario of the new joint learning task, determining the participants associated with the scenario of the new joint learning task, so that users can make self-selections of participants, and using the participants selected by the users as participants in the new joint learning task.

[0056] Specifically, when using Mode 2's autonomous participant selection method, participants meeting the criteria can be automatically identified and filtered based on the user's chosen joint learning scenario. Users can then manually select several or all of these participants; in this case, the number and types of joint learning participants are determined by the user. Additionally, users can choose whether to allow the addition of new participants. Compared to the automatic participant matching mode, the manual participant selection mode is more suitable for users with high requirements for model training performance.

[0057] According to the technical solution provided in this disclosure, a method for configuring joint learning tasks is provided. When creating a new joint learning task, corresponding parameters are automatically configured for the new task based on the joint learning scenario selected by the user in the task configuration management. This includes both basic parameter configuration and hyperparameter configuration. Furthermore, the values ​​of hyperparameters can be automatically or manually adjusted according to the hyperparameter configuration document. Therefore, this disclosure not only enables the rapid configuration of suitable joint learning task parameters for different joint learning scenarios, improving the efficiency of joint learning task configuration, but also allows users to create new joint learning tasks quickly and efficiently, enhancing the user's joint learning experience.

[0058] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein. For details not disclosed in the apparatus embodiments of this disclosure, please refer to the embodiments of the method disclosed herein.

[0059] Figure 3 This is a schematic diagram of the structure of the collaborative learning task configuration device provided in an embodiment of this disclosure. Figure 3 As shown, the joint learning task configuration device includes:

[0060] The receiving module 301 is configured to receive a task creation request sent by the participants, and in response to the task creation request, set the scenario of the new joint learning task and obtain the parameter configuration library corresponding to the scenario of the new joint learning task.

[0061] The retrieval module 302 is configured to retrieve parameters from the parameter configuration library that match the scenario of the new joint learning task, based on the scenario of the new joint learning task. The parameters include basic parameters and hyperparameters.

[0062] Configuration module 303 is configured to perform basic parameter configuration operations on a new joint learning task using basic parameters, and to perform hyperparameter configuration operations on a new joint learning task using hyperparameters. When performing hyperparameter configuration operations on a new joint learning task, hyperparameters are tuned based on preset tuning ranges according to tuning requests sent by the participants.

[0063] In some embodiments, Figure 3 Before receiving the task creation request sent by the participant, the receiving module 301 receives the task filtering request sent by the participant through the client. In response to the task filtering request, it filters the joint learning tasks in the joint learning task list using preset judgment conditions to obtain joint learning tasks that meet the judgment conditions. Based on the task status of the joint learning tasks that meet the judgment conditions, it determines whether to create a new joint learning task. The judgment conditions include whether the circle matches, whether the task is allowed to be automatically added, and whether the task is in the training state.

[0064] In some embodiments, Figure 3 The receiving module 301 determines the joint learning task scenario corresponding to the task scenario identifier in the task creation request, and sets the joint learning task scenario as the scenario corresponding to the new joint learning task.

[0065] In some embodiments, Figure 3 The retrieval module 302 obtains the task scenario identifier of the new joint learning task, searches for parameters in the parameter configuration library based on the task scenario identifier, and obtains the parameters corresponding to the task scenario identifier. The parameter configuration library contains the mapping relationship between the task scenario identifier and the parameters.

[0066] In some embodiments, Figure 3 The configuration module 303 obtains the field types corresponding to the basic parameters and automatically selects the field options corresponding to the field types of the basic parameters using the preset default parameter settings in order to obtain the basic parameter configuration results. Among them, the field types corresponding to the basic parameters include joint learning type, whether the task is public, algorithm type, aggregation method, encryption method, model compression, model evaluation index, termination condition, and participant selection.

[0067] In some embodiments, Figure 3 The configuration module 303 receives the hyperparameter configuration request sent by the participant through the client, and in response to the hyperparameter configuration request, configures the hyperparameters of the new joint learning task using the hyperparameters retrieved from the parameter configuration library; wherein, the hyperparameters include the number of training rounds, learning rate, sample sampling rate, feature sampling rate, maximum tree depth, L1 regularization coefficient, L2 regularization coefficient, and minimum loss function decrease value required for node splitting.

[0068] In some embodiments, Figure 3 The configuration module 303 retrieves the preset parameter tuning range according to the parameter tuning request in the hyperparameter configuration request, and determines the parameter tuning method according to the parameter tuning request, and uses the parameter tuning method to tune the hyperparameter within the parameter tuning range; wherein, the parameter tuning method includes automatic parameter tuning and manual parameter tuning.

[0069] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this disclosure.

[0070] Figure 4 This is a schematic diagram of the structure of the electronic device 4 provided in an embodiment of this disclosure. For example... Figure 4 As shown, the electronic device 4 of this embodiment includes a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, it implements the steps in the various method embodiments described above. Alternatively, when the processor 401 executes the computer program 403, it implements the functions of each module / unit in the various device embodiments described above.

[0071] For example, computer program 403 may be divided into one or more modules / units, which are stored in memory 402 and executed by processor 401 to perform the present disclosure. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 403 in electronic device 4.

[0072] Electronic device 4 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 4 may include, but is not limited to, processor 401 and memory 402. Those skilled in the art will understand that... Figure 4 This is merely an example of electronic device 4 and does not constitute a limitation on electronic device 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0073] Processor 401 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0074] The memory 402 can be an internal storage unit of the electronic device 4, such as a hard disk or RAM. The memory 402 can also be an external storage device of the electronic device 4, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 402 can include both internal and external storage units of the electronic device 4. The memory 402 is used to store computer programs and other programs and data required by the electronic device. The memory 402 can also be used to temporarily store data that has been output or will be output.

[0075] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0076] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0077] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0078] In the embodiments provided in this disclosure, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0079] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0080] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0081] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in a computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0082] The above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit it. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be included within the protection scope of this disclosure.

Claims

1. A method for configuring joint learning tasks, characterized in that, include: The system receives a task creation request sent by a participant, responds to the task creation request, determines a joint learning task scenario corresponding to the task scenario identifier in the task creation request, sets the joint learning task scenario as a new joint learning task scenario, and obtains a parameter configuration library corresponding to the scenario of the new joint learning task, wherein the parameter configuration library contains a mapping relationship between the task scenario identifier and the parameters. Based on the scenario of the new joint learning task, parameters that match the scenario of the new joint learning task are retrieved from the parameter configuration library, wherein the parameters include basic parameters and hyperparameters. The basic parameters are used to perform basic parameter configuration operations on the new joint learning task, and the hyperparameters are used to perform hyperparameter configuration operations on the new joint learning task. When performing hyperparameter configuration operations on the new joint learning task, the hyperparameters are tuned based on the parameter tuning requests sent by the participants and within a preset tuning range.

2. The method according to claim 1, characterized in that, Before receiving the task creation request sent by the participant, the method further includes: The system receives a task filtering request sent by the participant through the client, responds to the task filtering request, filters the joint learning tasks in the joint learning task list using preset judgment conditions, obtains joint learning tasks that meet the judgment conditions, and determines whether to create a new joint learning task based on the task status of the joint learning tasks that meet the judgment conditions. The judgment conditions include whether the circle matches, whether the task is allowed to be automatically added, and whether the task is in a training state.

3. The method according to claim 1, characterized in that, The step of retrieving parameters from the parameter configuration library that match the scenario of the new joint learning task, based on the scenario of the new joint learning task, includes: Obtain the task scenario identifier of the new joint learning task, and search for parameters in the parameter configuration library based on the task scenario identifier to obtain the parameters corresponding to the task scenario identifier.

4. The method according to claim 1, characterized in that, The step of configuring basic parameters for the new joint learning task using the basic parameters includes: Obtain the field type corresponding to the basic parameter, and automatically select the field option corresponding to the field type of the basic parameter using the preset default parameter settings, so as to obtain the basic parameter configuration result; The fields corresponding to the basic parameters include joint learning type, whether the task is public, algorithm type, aggregation method, encryption method, model compression, model evaluation index, termination condition, and participant selection.

5. The method according to claim 1, characterized in that, The step of performing hyperparameter configuration operations on the new joint learning task using the hyperparameters includes: The system receives a hyperparameter configuration request sent by the participating party through the client, and in response to the hyperparameter configuration request, configures the hyperparameters of the new joint learning task using the hyperparameters retrieved from the parameter configuration library. The hyperparameters include the number of training rounds, learning rate, sample sampling rate, feature sampling rate, maximum tree depth, L1 regularization coefficient, L2 regularization coefficient, and minimum loss function decrease value required for node splitting.

6. The method according to claim 5, characterized in that, The step of tuning the hyperparameters based on a preset tuning range according to the tuning request sent by the participating party includes: According to the parameter tuning request in the hyperparameter configuration request, a preset parameter tuning range is retrieved, and the parameter tuning method is determined according to the parameter tuning request. The hyperparameter is tuned within the parameter tuning range using the parameter tuning method. The parameter tuning method includes automatic parameter tuning and manual parameter tuning.

7. A joint learning task configuration device, characterized in that, include: The receiving module is configured to receive a task creation request sent by a participant, and in response to the task creation request, determine a joint learning task scenario corresponding to the task scenario identifier in the task creation request, set the joint learning task scenario as a new joint learning task scenario, and obtain a parameter configuration library corresponding to the new joint learning task scenario, wherein the parameter configuration library contains a mapping relationship between the task scenario identifier and the parameters. The retrieval module is configured to retrieve parameters from the parameter configuration library that match the scenario of the new joint learning task, based on the scenario of the new joint learning task, wherein the parameters include basic parameters and hyperparameters. The configuration module is configured to perform basic parameter configuration operations on the new joint learning task using the basic parameters, and to perform hyperparameter configuration operations on the new joint learning task using the hyperparameters. When performing hyperparameter configuration operations on the new joint learning task, the module adjusts the hyperparameters based on a preset debugging range according to the parameter tuning request sent by the participants.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images