A method and apparatus for jointly learning task exception processing
By monitoring the progress of collaborative learning tasks and adjusting the tasks according to the reasons for interruptions, the problem of some users being unable to participate in collaborative learning was resolved, ensuring the smooth execution of tasks and user participation.
Patent Information
- Application Number
- CN202011507786.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-18
- Publication Date
- 2026-02-06
- Estimated Expiration
- 2040-12-18
AI Technical Summary
In joint learning tasks, some energy users are unable to train accurate prediction models due to insufficient data, which prevents them from participating in joint learning smoothly. A reasonable anomaly handling method is needed to ensure their smooth participation.
By monitoring the progress of target users in executing federated learning tasks, the reasons for interruption can be determined, and task updates can be controlled based on the reasons for interruption. This includes updating tasks when interruptions occur or rejoining tasks when resource status is inconsistent, ensuring that users can successfully participate in federated learning.
It enables timely adjustments when tasks are interrupted, ensuring that target users can smoothly participate in collaborative learning, thereby improving the efficiency and accuracy of collaborative learning.
Smart Images
Figure CN114648128B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of energy, in particular to a joint learning task exception processing method and device. BACKGROUND
[0002] With the rapid development of Internet technology, user data has become an increasingly important resource, and various prediction models can be trained based on user data. Accurate prediction results are the basis for efficient operation of energy systems. However, not every energy user can collect massive user data to train accurate prediction models, which makes joint learning a trend. In order to ensure that each energy user successfully participates in joint learning, it becomes increasingly important to determine a reasonable joint learning task exception processing method. SUMMARY
[0003] The present application provides a joint learning task exception processing method, device, readable medium and electronic equipment, which monitors the task process of the target joint learning task performed by the target user, determines the interruption cause when the task is interrupted, and then controls the update of the target joint learning task according to the interruption cause, to ensure that the target user successfully participates in joint learning.
[0004] In a first aspect, the present application provides a joint learning task exception processing method, comprising:
[0005] Based on the selection result of the target user on the training task, a target joint learning task is created for the target user;
[0006] The task process of the target user performing the target joint learning task is determined;
[0007] When the task process shows that the target user interrupts the execution of the target joint learning task, the interruption cause of the target user performing the target joint learning task is determined;
[0008] Based on the interruption cause, the update of the target joint learning task is controlled.
[0009] Preferably,
[0010] The update of the target joint learning task based on the interruption cause comprises:
[0011] If the interruption cause shows that the target user actively interrupts, the target joint learning task is updated;
[0012] If the interruption cause shows that the target user does not actively interrupt, the target user is added to the target joint learning task again.
[0013] Preferably,
[0014] if the interrupt cause shows that the target user initiatively interrupts, updating the target joint learning task, comprising:
[0015] if the interrupt cause shows that the target user initiatively interrupts, updating the training task corresponding to the target user;
[0016] sending the updated training task to the target user;
[0017] updating the target joint learning task based on the selection result of the target user on the updated training task.
[0018] Preferably,
[0019] if the interrupt cause shows that the target user does not initiatively interrupt, re-adding the target user to the target joint learning task, comprising:
[0020] if the interrupt cause shows that the target user does not initiatively interrupt, determining the resource node state of the target joint learning task;
[0021] if the resource node state of the target joint learning task meets the preset condition, re-adding the target user to the target joint learning task;
[0022] if the resource node state of the target joint learning task does not meet the preset condition, updating the target joint learning task.
[0023] Preferably,
[0024] Before the method further comprises:
[0025] obtaining a task creation request of a target user;
[0026] determining a training task for the target user based on the task creation request, and sending the training task to the target user.
[0027] Preferably,
[0028] The method further comprises:
[0029] when a task process shows that a target user interrupts a target joint learning task, sending an alarm information to the target user.
[0030] In a second aspect, the present application provides a joint learning task exception processing device, comprising:
[0031] The task creation module is used to create a joint learning task for the target user based on the target user's selection of training tasks.
[0032] A process determination module is used to determine the task process of the target user executing the target joint learning task.
[0033] The cause determination module is used to determine the cause of the interruption of the target user's execution of the target joint learning task when the task process displays that the target user's execution of the target joint learning task has been interrupted;
[0034] An update control module is used to control the update of the target joint learning task based on the interruption reason.
[0035] Preferably,
[0036] The update control module includes:
[0037] The task update unit is used to update the target joint learning task if the interruption reason indicates that the target user actively interrupted the process.
[0038] The task addition unit is used to add the target user back into the target joint learning task if the interruption reason indicates that the target user did not actively interrupt the task.
[0039] Thirdly, the present invention provides a readable medium including executable instructions, which, when executed by a processor of an electronic device, cause the electronic device to perform any of the methods described in the first aspect.
[0040] Fourthly, the present invention provides an electronic device including a processor and a memory storing execution instructions, wherein when the processor executes the execution instructions stored in the memory, the processor performs the method as described in any of the first aspects.
[0041] This invention provides a method, apparatus, readable medium, and electronic device for handling anomalies in a joint learning task. The method determines the target user's selection of a training task, creates a target joint learning task for the target user, monitors the target user's execution status of the target joint learning task to determine the task progress, and when the task progress indicates an interruption, determines the reason for the interruption and then controls the update of the target joint learning task based on the reason for the interruption. The technical solution provided by this invention, by monitoring the task progress of the target user in executing the target joint learning task, determining the reason for the interruption when the task is interrupted, and then controlling the update of the target joint learning task based on the reason for the interruption, ensures that the target user can successfully participate in joint learning. Attached Figure Description
[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.
[0043] Figure 1 A flowchart of a first joint learning task exception processing method provided in an embodiment of the present application;
[0044] Figure 2 A flowchart of a second joint learning task exception processing method provided in an embodiment of the present application;
[0045] Figure 3 A flowchart of a third joint learning task exception processing method provided in an embodiment of the present application;
[0046] Figure 4 A flowchart of a fourth joint learning task exception processing method provided in an embodiment of the present application;
[0047] Figure 5 A structural diagram of a joint learning task exception processing device provided in an embodiment of the present application;
[0048] Figure 6 A structural diagram of an update control module in a joint learning task exception processing device provided in an embodiment of the present application;
[0049] Figure 7 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0050] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort belong to the scope of protection of the present application.
[0051] As shown in Figure 1 The present application provides a joint learning task exception processing method, which comprises:
[0052] Step 11, creating a target joint learning task for the target user based on the selection result of the target user on the training task;
[0053] Step 12: Determine the task progress of the target user in performing the target joint learning task;
[0054] Step 13: When the task process displays that the target user has interrupted the execution of the target joint learning task, determine the reason for the interruption of the target user's execution of the target joint learning task;
[0055] Step 14: Based on the reason for the interruption, control the update of the target joint learning task.
[0056] In the above embodiments, the target user's selection result for the training task is determined, and a target joint learning task is created for the target user based on the selection result. After the target joint learning task is created, the target user executes the target joint learning task. The status of the target user's execution of the target joint learning task is monitored to determine the task progress of the target user's execution of the target joint learning task. The task progress can be in training, interrupted, or terminated. When the task progress shows that the target user's execution of the target joint learning task is interrupted, the target joint learning task has an anomaly. The reason for the interruption is determined, and then the update of the target joint learning task is controlled based on the reason for the interruption. This realizes that after the target joint learning task is created, the update of the target joint learning task can be controlled when the target user's execution of the target joint learning task is abnormal, ensuring that the target user can successfully participate in the target joint learning task.
[0057] like Figure 2 As shown, in one embodiment of the present invention, before step 11, which creates a joint learning task for the target user based on the target user's selection of the training task, the method further includes:
[0058] Step 15: Obtain the target user's task creation request;
[0059] Step 16: Based on the task creation request, determine a training task for the target user and send the training task to the target user.
[0060] In the above embodiment, when the target user needs to create a task, the target user sends a task creation request to the joint center, the joint learning center receives the task creation request of the target user, determines the requirement information carried by the task creation request, and the requirement information includes but is not limited to algorithm, computing power, and task type. The joint learning center determines a training task for the target user in a task list according to the requirement information, wherein the task list is a list corresponding to all tasks supported by the joint learning center at the current moment, the training task is a task selected from the task list according to the requirement information, and the training task is sent to the target user after being determined, so that the target user selects the training task. Of course, if the joint learning center cannot meet the requirement information carried by the task creation request, such as being unable to provide the algorithm required in the requirement information, there is no training task at this time.
[0061] As shown in FIG. 14, in one embodiment of the present application, the step 14 controls the update of the target joint learning task based on the interruption cause, including: Figure 3
[0062] Step 141, if the interruption cause shows that the target user actively interrupts, the target joint learning task is updated;
[0063] Step 142, if the interruption cause shows that the target user does not actively interrupt, the target user is added to the target joint learning task again.
[0064] In the above embodiment, there are two cases when the target joint learning task is interrupted, one is that the target user actively interrupts, and the other is that the target user does not actively interrupt, for example, interruption caused by network anomaly. When it is determined that the interruption cause shows that the target user actively interrupts, it means that the target user no longer wants to continue to execute the target joint learning task, and the target joint learning task is updated for the target user to determine the target joint learning task that meets the user's requirement. When the interruption cause shows that the target user does not actively interrupt, the target user is added to the target joint learning task again, so that the target user continues to execute the target joint learning task, and ensures that the target user executes the target joint learning task smoothly.
[0065] In a possible implementation, when the interrupt cause shows that the target user initiatively interrupts, the training task corresponding to the target user is updated; the updated training task is sent to the target user; and the target joint learning task is updated based on a selection result of the target user on the updated training task. In order to meet the demand of the target user to create a task, when the target user initiatively interrupts the execution of the target joint learning task, a training task is reselected in a task list of the joint learning center to update the training task corresponding to the target user, the updated training task is sent to the target user, the target user makes a selection again according to the updated training task, a new target joint learning task is created for the target user according to the selection result made by the target user again, and the target joint learning task is updated. Of course, the historical data of the target user can also be determined, the number of times of the target user's initiatively interrupting is determined according to the historical data of the target user, if the number of times of the target user's initiatively interrupting is greater than a preset value (for example, 5 times), the target joint learning task is no longer actively updated for the target user. When the target user initiatively interrupts the target joint learning task, there can also be other implementation manners, for example, the target user needs to send a task creation request again, the qualification of the target user to create a task is audited, a new target joint learning task is created for the target user after the qualification of the target user to create a task is qualified, and the qualification of the target user to create a task is qualified specifically, that is, the number of times of the target user's initiatively interrupting is not greater than a preset value, and the joint learning center can meet the demand information carried in the task creation request sent by the target user again.
[0066] In another possible implementation, if the interrupt cause shows that the target user does not initiatively interrupt, the target user is added to the target joint learning task again, including: if the interrupt cause shows that the target user does not initiatively interrupt, a resource node state of the target joint learning task is determined; if the resource node state of the target joint learning task meets a preset condition, the target user is added to the target joint learning task again; and if the resource node state of the target joint learning task does not meet the preset condition, the target joint learning task is updated.
[0067] When the interruption reason indicates that the target user did not actively interrupt, it means that the target user wants to continue executing the target joint learning task. At this time, it is necessary to determine the status of the resource nodes of the target joint learning task. Resource nodes refer to each registered energy user. When an energy user wants to create a task, it needs to register its resources first, such as filling in its bandwidth, computing power, and device resources. After resource registration, the energy user becomes a resource node. If the status of the resource nodes participating in the target joint learning task meets the preset conditions, the target user can be re-added to the target joint learning task, allowing the target user to continue executing the task. The preset conditions can be that the working status is normal. If the working status of the resource nodes of the target joint learning task is abnormal, even if the target user is re-added, it will not be able to continue executing the target joint learning task, thus updating the target joint learning task for the target user. Of course, when the status of the resource nodes of the target joint learning task does not meet the preset conditions, the target resource nodes that do not meet the preset conditions can be identified, and then the duration for which the target resource nodes do not meet the preset conditions can be determined. If the duration is longer than the preset duration, the target joint learning task is updated, that is, the target resource nodes are given time for repair and maintenance. If the duration is too long, the target joint learning task is updated for the target user in order to ensure that the target user can execute the target joint learning task smoothly.
[0068] like Figure 4 As shown, in one embodiment of the present invention, the method further includes:
[0069] Step 17: When the task process shows that the target user has interrupted the execution of the target joint learning task, send an alarm message to the target user.
[0070] In the above embodiments, when the task process indicates a task interruption, the target user needs to be promptly notified; therefore, an alarm message is sent to the target user. Specifically, when the reason for task interruption indicates that the target user actively interrupted the task, the alarm message content may include whether the target joint learning task needs to be updated. If the target user chooses not to update the target joint learning task, the joint learning center will stop updating the target joint learning task. When the reason for task interruption indicates that the target user did not actively interrupt the task, an alarm message is sent to the target user so that the target user can promptly check and eliminate the cause of the interruption, ensuring the smooth execution of the target joint learning task.
[0071] Based on the same inventive concept as the above method, such as Figure 5 As shown, an embodiment of the present invention provides an anomaly handling device for a joint learning task, comprising:
[0072] The task creation module 51 is used to create a joint learning task for the target user based on the target user's selection of training tasks.
[0073] Process determination module 52 is used to determine the task process of the target user performing the target joint learning task;
[0074] The cause determination module 53 is used to determine the cause of the interruption of the target user's execution of the target joint learning task when the task process displays that the target user's execution of the target joint learning task is interrupted;
[0075] The update control module 54 is used to control the update of the target joint learning task based on the interruption reason.
[0076] like Figure 6 As shown, in one embodiment of the present invention, the update control module 54 includes:
[0077] The task update unit 541 is used to update the target joint learning task if the interruption reason shows that the target user actively interrupted the task.
[0078] The task addition unit 542 is used to add the target user back into the target joint learning task if the interruption reason shows that the target user did not actively interrupt.
[0079] For ease of description, the above device embodiments are described by function as various units or modules. When implementing the present invention, the functions of each unit or module can be implemented in one or more software and / or hardware.
[0080] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. At the hardware level, the electronic device includes a processor 701 and a memory 702 storing execution instructions, and optionally also includes an internal bus 703 and a network interface 704. The memory 702 may include main memory 7021, such as high-speed random-access memory (RAM), and may also include non-volatile memory 7022, such as at least one disk storage device. The processor 701, network interface 704, and memory 702 can be interconnected via the internal bus 703, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The internal bus 703 can be divided into an address bus, a data bus, a control bus, etc., for ease of representation.Figure 7 Only one bidirectional arrow is shown for each bus, but it will be understood that the bus can include other buses that are not explicitly shown, such as address or data buses. Furthermore, if the processor 701 is a multi-core processor, then each core can have its own bus. The bus 703 carries data to the control unit 704 and control signals to the bus controller 705. The electronic device can also include other peripherals used to obtain data. When the processor 701 executes instructions to perform the steps of the methods of any embodiments of the present application, the processor 701 performs the steps of the methods of any embodiments of the present application and at least to execute the steps of the methods as shown in Figures 1 to 4 .
[0081] In one possible implementation, the processor reads the corresponding execution instructions from the non-volatile memory into the memory and then runs, and can also obtain the corresponding execution instructions from other devices to form a joint learning task exception handling apparatus at a logical level. The processor executes the execution instructions stored in the memory to implement the joint learning task exception handling method provided in any embodiment of the present application through the executed execution instructions.
[0082] The processor can be an integrated circuit chip having a processing capability of signals. In the implementation process, each step of the above method can be completed by an integrated logic circuit or an instruction in the form of software in the processor. The processor described above can be a general processor, including a central processing unit (CPU), a network processor (NP), and the like; can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. Each method, step and logic block diagram disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor.
[0083] The embodiments of the present application also provide a computer readable storage medium, including execution instructions, when the processor of the electronic device executes the execution instructions, the processor executes the method provided in any embodiment of the present application. The electronic device can be specifically the electronic device as shown in Figure 7 ; the execution instructions are a computer program corresponding to the joint learning task exception handling apparatus.
[0084] Those skilled in the art should understand that the embodiments of the present application can be provided as a method or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or a combination of software and hardware.
[0085] The various embodiments in the present application are described in a progressive manner, and the same or similar parts among the various embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments.
[0086] It should also be noted that the terms "comprising", "comprising" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or boiler including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or boiler. Without more limitations, the element defined by the statement "comprising a" does not exclude the existence of other identical elements in the process, method, article or boiler comprising the element.
[0087] The above only describes the embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of the claims of the present application.
Claims
1. A method for handling anomalies in a joint learning task, characterized in that, The method comprises the following steps: creating a target joint learning task for a target user based on the selection result of the target user on a training task; determining a task progress of the target user performing the target joint learning task; when the task progress shows that the target user interrupts the performance of the target joint learning task, determining an interruption cause of the target user performing the target joint learning task; controlling the update of the target joint learning task based on the interruption cause; the controlling the update of the target joint learning task based on the interruption cause comprises: if the interruption cause shows that the target user interrupts voluntarily, updating the target joint learning task; if the interruption cause shows that the target user interrupts involuntarily, re-adding the target user to the target joint learning task; the updating the target joint learning task if the interruption cause shows that the target user interrupts voluntarily comprises: if the interruption cause shows that the target user interrupts voluntarily, updating the training task corresponding to the target user; sending the updated training task to the target user; updating the target joint learning task based on the selection result of the target user on the updated training task; the re-adding the target user to the target joint learning task if the interruption cause shows that the target user interrupts involuntarily comprises: if the interruption cause shows that the target user interrupts involuntarily, determining a resource node state of the target joint learning task; if the resource node state of the target joint learning task meets a preset condition, re-adding the target user to the target joint learning task; if the resource node state of the target joint learning task does not meet the preset condition, updating the target joint learning task.
2. The federated learning task anomaly processing method of claim 1, wherein, Before the creating a target joint learning task for a target user based on the selection result of the target user on a training task, the method further comprises: obtaining a task creation request of a target user; determining a training task for the target user based on the task creation request, and sending the training task to the target user. 3.The federated learning task exception handling method of claim 1, wherein, The method further comprises: when the task progress shows that the target user interrupts the performance of the target joint learning task, sending an alarm information to the target user.
4. A device for federated learning task anomaly handling, the device employing the method of any one of claims 1-3, characterized in that, The device comprises: a task creation module, configured to create a target joint learning task for a target user based on the selection result of the target user on a training task; a progress determination module, configured to determine a task progress of the target user performing the target joint learning task; a cause determination module, configured to, when the task progress shows that the target user interrupts the performance of the target joint learning task, determine an interruption cause of the target user performing the target joint learning task; an update control module, configured to control the update of the target joint learning task based on the interruption cause; the update control module comprises: a task update unit, configured to, if the interruption cause shows that the target user interrupts voluntarily, update the target joint learning task; The task joining unit is configured to re-add the target user to the target joint learning task if the interrupt cause indicates that the target user is not actively interrupted. 5.A readable medium comprising execution instructions, when a processor of an electronic device executes the execution instructions, the electronic device performs the method according to any one of claims 1 to 3. 6.An electronic device comprising a processor and a memory storing execution instructions, when the processor executes the execution instructions stored in the memory, the processor performs the method according to any one of claims 1 to 3.
Citation Information
Patent Citations
Distributed task scheduling system
CN110895486A
Deep learning model training method, device and system
CN111079892A