Model identification method and related device

Abstract

The embodiment of the invention provides a model identification method and a related device. The method comprises the following steps: a first device obtains at least one of a region identifier ID of a second device and a user ID of a third device, and a task ID of an artificial intelligence AI service; the first device generates a first model ID of an AI model according to at least one of the area ID and the user ID and the task ID, the AI model is used for providing the AI service, and the first model ID is used for identifying the AI model; and the first device sends the first model ID to the third device. Therefore, the third device identifies or recognizes the AI model through the first model ID.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a model identification method and related apparatus. BACKGROUND

[0002] Currently, distributed learning can complete the learning task of an AI model under the premise of fully guaranteeing user data privacy and security. Distributed learning mainly includes federated learning, split learning, and decentralized learning. However, different AI models can have different functions. For example, AI can be used for channel state information (CSI) feedback enhancement, beam management, positioning enhancement, or CSI compressed feedback between network devices and terminal devices. Therefore, for terminal devices or network devices, how to identify different AI models is a problem worth considering. SUMMARY

[0003] The present application provides a model identification method and related apparatus, for a first device to generate a first model ID of an AI model according to at least one of a region ID of a second device and a user ID of a third device and a task ID of an AI service, and send the first model ID to the third device. Thereby, the third device identifies or recognizes the AI model through the first model ID.

[0004] The first aspect of the present application provides a model identification method, which can be executed by a first device. The first device can be a first network device or an AI server, or a component (for example, a processor, a chip, or a chip system, etc.) in the first network device or the AI server, or a logic module or software capable of realizing all or part of the function of the first network device, or a logic module or software capable of realizing all or part of the AI server. The method comprises: the first device obtaining at least one of a region ID of a second device and a user ID of a third device, and a task ID of an AI service; the first device generating a first model ID of an AI model according to at least one of the region ID and the user ID and the task ID, the AI model being used to provide an AI service, and the first model ID being used to identify the AI model; and the first device sending the first model ID to the third device.

[0005] In the technical solution, the first device generates a first model ID, and the first model ID is used to identify the AI model. The first device sends the first model ID to the third device. The AI model is identified or recognized by the first model ID. Further, the first device generates the first model ID by combining at least one of the region ID and the user ID and the task ID, so that the first model ID can uniquely identify the AI model in a larger region. Thus, the AI model can be identified or recognized by the first model ID between different devices in a distributed learning scenario.

[0006] In a possible implementation manner of the first aspect, the first device generates the first model ID of the AI model according to at least one of the region ID and the user ID and the task ID, including: the first device generates the first model ID according to at least one of the region ID and the user ID, the task ID, a model version ID of the AI model, and / or a training round ID of the AI model. In this implementation manner, the first device generates the first model ID according to the model version ID of the AI model and / or the training round ID of the AI model. Thus, the AI model can be uniquely identified by the first model ID, and the AI model can be identified or recognized between different devices in a distributed learning scenario. For example, when the AI model is trained, the first device generates the first model ID by combining the training round ID of the AI model. Thus, the training round ID can represent the AI model corresponding to the corresponding training round.

[0007] In a possible implementation manner of the first aspect, before the first device generates the first model ID according to at least one of the region ID and the user ID, the task ID, and the model version ID of the AI model and / or the training round ID of the AI model, the method further includes: the first device receives the model version ID of the AI model from the fourth device. Thus, the first device can generate the first model ID based on the model version ID. Different versions of the AI model correspond to different model IDs, so that the different versions of the AI model can be identified.

[0008] In a possible implementation manner of the first aspect, the method further includes: the first device sends the first model ID to the fourth device. Thus, the fourth device can identify or recognize the AI model based on the first model ID. In this implementation manner, the first device can be a first network device, and the fourth device can be a first AI server.

[0009] In a possible implementation manner of the first aspect, the method further includes: receiving, by the first device, a registration request from a fourth device, the registration request being used to request registration of the AI service; and sending, by the first device, a task ID of the AI service to the fourth device. In this implementation manner, the first device can assign a task ID to the AI service. This facilitates the first device to generate a first model ID based on the task ID. Thus, the identification or recognition of AI models corresponding to different AI services is implemented.

[0010] In a possible implementation manner of the first aspect, the method further includes: selecting, by the first device, the third device according to at least one of local data information of the third device, computing power information of the third device, and communication information between the second device and the third device; and sending, by the first device, a user ID of the third device to the third device. This enables the first device to select a suitable third device to participate in the management operation of the AI model corresponding to the AI service. Further, the first device generates a first model ID based on the user ID, so that the model IDs of different AI models owned by different users are different. Thus, the different AI models owned by different users are identified or recognized.

[0011] In a possible implementation manner of the first aspect, the method further includes: sending, by the first device, first indication information to the third device, the first indication information being used to indicate that the AI model is trained, and the first indication information including the first model ID; training, by the first device and the third device, the AI model to obtain an updated AI model; and sending, by the first device, the updated AI model to the fourth device. Thus, the first device indicates to the third device that the AI model is trained through the first model ID in the first indication information, so that the third device identifies the AI model based on the first model ID and trains the AI model corresponding to the first model ID with the first device. Thus, the training task of the AI model is completed.

[0012] In a possible implementation manner of the first aspect, the method further includes: generating, by the first device, a second model ID of the AI model according to the first model ID, the length of the second model ID being less than the length of the first model ID. In this implementation manner, the length of the second model ID is less than the length of the first model ID. This facilitates the first device to instruct other devices to perform a corresponding management operation on the AI model based on the second model ID. Thus, the indication overhead of the first device is reduced.

[0013] In a possible implementation manner of the first aspect, the method further includes: sending, by the first device, the second model ID to the third device. This facilitates the third device to identify or recognize the AI model based on the second model ID. This facilitates the first device to instruct other devices to perform a corresponding management operation on the AI model based on the second model ID. Thus, the indication overhead of the first device is reduced.

[0014] In a possible implementation manner of the first aspect, the method further includes: receiving, by the first device, second indication information from the fourth device, the second indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the second indication information including the first model ID; and sending, by the first device, third indication information to the third device, the third indication information being used to indicate that the first management operation is performed on the AI model, and the third indication information including the second model ID. The first model ID in the third indication information is converted into the second model ID by the first device. The first device is facilitated to indicate the third device to perform the first management operation on the AI model by using the shorter second model ID. The indication overhead is reduced.

[0015] In a possible implementation manner of the first aspect, the method further includes: sending, by the first device, the second model ID to the second device. The second device is facilitated to identify or recognize the AI model. In the implementation manner, the first device can be the first AI server, and the second device can be the first network device.

[0016] In a possible implementation manner of the first aspect, the first device obtains at least one of the region ID of the second device and the user ID of the third device, and the task ID of the AI service, including: receiving, by the first device, at least one of the region ID and the user ID of the second device and the task ID from the second device. In the implementation manner, the first device can be the first AI server, and the second device can be the first network device. The first AI server obtains the above-mentioned IDs from the first network device.

[0017] In a possible implementation manner of the first aspect, the method further includes: sending, by the first device, fourth indication information to the second device and the third device respectively, the fourth indication information being used to indicate that the AI model is trained, and the fourth indication information including the first model ID; training, by the first device, the second device, and the third device, the AI model to obtain an updated AI model; and updating, by the first device, the model version ID and / or the training round ID of the AI model based on the updated AI model. It is known that the first device indicates the second device and the third device to train the AI model by using the first model ID in the fourth indication information, so that the second device and the third device identify the AI model based on the first model ID, and train the AI model corresponding to the first model ID with the first device. The training task of the AI model is completed.

[0018] In a possible implementation manner of the first aspect, the method further includes: the first device sending fifth indication information to the second device, the fifth indication information being used to indicate performing a first management operation on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the fifth indication information including the first model ID. The first device indicates performing the first management operation on the AI model corresponding to the first model ID through the first model ID in the fifth indication information. Thus, the corresponding management of the AI model is completed.

[0019] In a possible implementation manner of the first aspect, the method further includes: the first model ID corresponding to a first training path, and a device on the first training path being used to perform a training operation on the AI model. Thus, it is known that the devices on different training paths can use different model IDs to identify or train the same AI model. Thus, for the devices on different training paths, the first device can use different model IDs to indicate the devices on different training paths to perform corresponding management operations on the AI model. Thus, the first device can complete the learning of the AI model through different training paths.

[0020] The second aspect of the present application provides a model identification method, which is applied to a network side. The method can be executed by a second device. The second device can be a first network device, or a component (for example, a processor, a chip, or a chip system, etc.) in the first network device, or a logic module or software capable of realizing all or part of the functions of the first network device. The method includes: the second device receiving a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of the second device and a user ID of a third device and a task ID of an AI service, the AI model being used to provide the AI service; and the second device determining the first model ID of the AI model. Thus, the AI model is identified or labeled based on the first model ID. Further, the first model ID is generated according to at least one of the region ID of the first device and the user ID of the third device and the task ID of the AI service. Thus, the first model ID can uniquely identify the AI model in a larger region. Thus, in the distributed learning scenario, the AI model is identified based on the first model ID between different devices.

[0021] In a possible implementation manner of the second aspect, before the second device receives the first model ID of the AI model from the first device, the method further includes: receiving, by the second device, a registration request from the first device, the registration request being used to request registration of an AI service, the AI model being an AI model corresponding to the AI service; and sending, by the second device, a task ID of the AI service to the first device. In this implementation manner, the second device can assign a task ID to the AI service. Thus, the first device can generate the first model ID based on the task ID. Thus, the identification or recognition of AI models corresponding to different AI services can be implemented.

[0022] In a possible implementation manner of the second aspect, before the second device receives the first model ID of the AI model from the first device, the method further includes: sending, by the second device, a region ID of the second device to the first device. Thus, the first device can generate the first model ID. The first model ID can uniquely identify the AI model in a larger region. Thus, in a distributed learning scenario, the AI model can be identified between different devices based on the first model ID.

[0023] In a possible implementation manner of the second aspect, before the second device receives the first model ID of the AI model from the first device, the method further includes: selecting, by the second device, a third device according to at least one of local data information of the third device, computing power information of the third device, and communication information between the second device and the third device; and sending, by the second device, a user ID of the third device to the first device. In this manner, the second device can select a suitable third device to participate in the management operation of the AI model corresponding to the AI service. Further, the second device sends the user ID to the first device. In this manner, the second device can generate the first model ID based on the user ID, so that the model IDs of different AI models owned by different users are different. Thus, the different AI models owned by different users can be identified or recognized.

[0024] In a possible implementation manner of the second aspect, the method further includes: receiving, by the second device, fourth indication information from the first device, the fourth indication information being used to indicate that the AI model is trained, the fourth indication information including the first model ID; and training, by the second device, the AI model together with the first device and the third device, to obtain an updated AI model. In this manner, the second device can train the AI model together with the first device and the third device. Thus, the training task of the model can be completed.

[0025] In a possible implementation manner of the second aspect, the method further includes: generating, by the second device, a second model ID of the AI model according to the first model ID, the length of the second model ID being less than the length of the first model ID. In this implementation manner, the length of the second model ID is less than the length of the first model ID. Thus, the second device can instruct other devices to perform a corresponding management operation on the AI model based on the second model ID. Thus, the indication overhead of the second device is reduced.

[0026] In a possible implementation manner of the second aspect, the method further includes: sending, by the second device, the second model ID to a third device. The third device can identify or recognize the AI model based on the second model ID. The second device can instruct other devices to perform a corresponding management operation on the AI model based on the second model ID. Thus, the indication overhead of the second device is reduced.

[0027] In a possible implementation manner of the second aspect, the method further includes: receiving, by the second device, fifth indication information from the first device, the fifth indication information being used to instruct to perform a first management operation on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the fifth indication information including the first model ID; and sending, by the second device, sixth indication information to a third device, the sixth indication information being used to instruct to perform the first management operation on the AI model, and the sixth indication information including the second model ID. The second device can convert the first model ID in the fifth indication information into the second model ID. The second device can instruct the third device to perform the first management operation on the AI model by using the shorter second model ID.

[0028] The third aspect of the present application provides a model identification method, which is applied to a terminal side. The method can be executed by a third device, which can be a terminal device, or a component (for example, a processor, a chip, or a chip system, etc.) in the terminal device, or a logic module or software capable of realizing all or part of the functions of the terminal device. The method includes: receiving, by the third device, a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of the first device and a user ID of the third device and a task ID of an AI service, the AI model being used to provide the AI service; and determining, by the third device, the first model ID of the AI model. Thus, the AI model can be identified or recognized based on the first model ID. Further, the first model ID is generated according to at least one of the region ID of the second device and the user ID of the third device and the task ID of the AI service, and the first model ID can uniquely identify the AI model in a larger region. Thus, the AI model can be identified between different devices based on the first model ID in a distributed learning scenario.

[0029] In a possible implementation manner of the third aspect, the method further includes: the third device receiving first indication information from the first device, the first indication information being used to indicate that the AI model is trained, and the first indication information including a first model ID; and the third device training the AI model based on the first model ID and the second device to obtain an updated AI model. The third device identifies the AI model based on the first model ID, and trains the AI model corresponding to the first model ID with the first device. Thus, the training task of the model is completed.

[0030] In a possible implementation manner of the third aspect, the method further includes: the third device receiving a second model ID of the AI model from the first device, the second model ID having a length smaller than that of the first model ID; or the third device generating the second model ID of the AI model according to the first model ID, the second model ID having a length smaller than that of the first model ID. Two possible implementation manners of the third device obtaining the second model ID are shown. Thus, the first device indicates the third device to perform a corresponding management operation on the AI model based on the second model ID.

[0031] In a possible implementation manner of the third aspect, the method further includes: the third device receiving third indication information from the first device, the third indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the third indication information including the second model ID; and the third device performing the first management operation on the AI model based on the second model ID. Thus, the third device identifies the AI model based on the second model ID, and performs a corresponding management operation. The first device indicates the third device to perform the first management operation on the AI model by using the second model ID which is relatively shorter. The indication overhead is reduced.

[0032] In a possible implementation manner of the third aspect, the method further includes: the third device receiving sixth indication information from the second device, the sixth indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the sixth indication information including the second model ID; and the third device performing the first management operation on the AI model based on the second model ID. Thus, the third device identifies the AI model based on the second model ID, and performs a corresponding management operation. The second device indicates the third device to perform the first management operation on the AI model by using the second model ID which is relatively shorter. The indication overhead is reduced.

[0033] In a possible implementation manner of the third aspect, the first model ID corresponds to a first training path, and devices on the first training path are configured to perform corresponding training operations on the AI model; the method further includes: receiving, by the third device, a third model ID of the AI model from the fifth device or the first device, the third model ID corresponding to a second training path, and devices on the second training path are configured to perform training operations on the AI model; and determining, by the third device, that the first model ID and the third model ID both identify the AI model. Thus, the third device can identify the AI model based on different model IDs of the AI model in the future, and train the AI model with devices on different training paths. The training task of the AI model is completed through multiple training paths.

[0034] In a possible implementation manner of the third aspect, the method further includes: receiving, by the third device, seventh indication information from the fifth device, the seventh indication information being used to indicate that the AI model is trained, and the seventh indication information including the third model ID; training, by the third device, the AI model based on the third model ID and the fifth device; and training, by the third device, the AI model based on the third model ID and the fifth device and the first device. Thus, the training task of the AI model is completed through multiple training paths.

[0035] The fourth aspect of the present application provides a model identification method, which is applied to an AI server side. The method can be executed by a fourth device. The fourth device can be an AI server, or a component (for example, a processor, a chip, or a chip system, etc.) in the AI server, or a logic module or software capable of realizing all or part of the AI server function. The method includes: receiving, by the fourth device, a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of a second device and a user ID of a third device and a task ID of an AI service, and the AI model being used to provide the AI service; and determining, by the fourth device, the first model ID of the AI model. Thus, the AI model is identified based on the first model ID. Further, the first model ID is generated according to at least one of the region ID of the first device and the user ID of the third device and the task ID of the AI service, so that the first model ID can uniquely identify the AI model in a larger region. Thus, the AI model is identified based on the first model ID between different devices in a distributed learning scenario.

[0036] In a possible implementation manner of the fourth aspect, the first model ID is further generated according to a model version ID of the AI model. Before the fourth device receives the first model ID of the AI model from the first device, the method further includes: sending, by the fourth device, a model version ID of the AI model to the first device. Different versions of the AI model correspond to different model IDs, so that the different versions of the AI model are identified.

[0037] In a possible implementation manner based on the fourth aspect, the method further includes: the fourth device sending second indication information to the first device, the second indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the second indication information including the first model ID. The fourth device performs the first management operation on the AI model based on the first model ID in the second indication information.

[0038] The fifth aspect of the present application provides a first device, including:

[0039] a processing module, configured to acquire at least one of a region ID of a second device and a user ID of a third device, and a task ID of an AI service, and generate a first model ID of an AI model according to the at least one of the region ID and the user ID and the task ID, the AI model being used to provide the AI service, and the first model ID being used to identify the AI model;

[0040] a transceiver module, configured to send the first model ID to the third device.

[0041] In a possible implementation manner based on the fifth aspect, the processing module is further configured to generate the first model ID according to the at least one of the region ID and the user ID, the task ID, and a model version ID of the AI model and / or a training round ID of the AI model.

[0042] In a possible implementation manner based on the fifth aspect, the transceiver module is further configured to receive the model version ID of the AI model from a fourth device.

[0043] In a possible implementation manner based on the fifth aspect, the transceiver module is further configured to send the first model ID to the fourth device.

[0044] In a possible implementation manner based on the fifth aspect, the transceiver module is further configured to receive a registration request from the fourth device, the registration request being used to request registration of the AI service, and send the task ID of the AI service to the fourth device.

[0045] In a possible implementation manner based on the fifth aspect, the processing module is further configured to select the third device according to at least one of local data information of the third device, computing power information, and communication information between the second device and the third device, and the transceiver module is further configured to send the user ID of the third device to the third device.

[0046] In a possible implementation manner based on the fifth aspect, the transceiver module is further configured to send first indication information to the third device, the first indication information being used to indicate that the AI model is trained, and the first indication information including the first model ID.

[0047] The processing module is further configured to train the AI model with the third device to obtain an updated AI model.

[0048] The transceiver module is further configured to send the updated AI model to a fourth device.

[0049] In a possible implementation manner of the fifth aspect, the processing module is further configured to generate a second model ID of the AI model according to the first model ID, and a length of the second model ID is less than a length of the first model ID.

[0050] In a possible implementation manner of the fifth aspect, the transceiver module is further configured to send the second model ID to the third device.

[0051] In a possible implementation manner of the fifth aspect, the transceiver module is further configured to receive second indication information from the fourth device, the second indication information being used to instruct to perform a first management operation on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the second indication information including the first model ID; and send third indication information to the third device, the third indication information being used to instruct to perform the first management operation on the AI model, and the third indication information including the second model ID.

[0052] In a possible implementation manner of the fifth aspect, the transceiver module is further configured to send the second model ID to the second device.

[0053] In a possible implementation manner of the fifth aspect, the processing module is specifically configured to receive at least one of a region ID and a user ID and a task ID from the second device.

[0054] In a possible implementation manner of the fifth aspect, the transceiver module is further configured to send fourth indication information to the second device and the third device respectively, the fourth indication information being used to instruct to train the AI model, and the fourth indication information including the first model ID; and the processing module is further configured to train the AI model with the second device and the third device to obtain an updated AI model, and update a model version ID and / or a training round ID of the AI model based on the updated AI model.

[0055] In a possible implementation manner of the fifth aspect, the transceiver module is further configured to send fifth indication information to the second device, the fifth indication information being used to instruct to perform a first management operation on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the fifth indication information including the first model ID.

[0056] In a possible implementation manner based on the fifth aspect, the first model ID corresponds to a first training path, and an apparatus on the first training path is configured to perform a training operation on the AI model.

[0057] The sixth aspect of the present application provides a second apparatus, comprising:

[0058] a transceiver configured to receive a first model ID of an AI model from a first apparatus, the first model ID being generated according to at least one of a region ID of the second apparatus and a user ID of a third apparatus and a task ID of an AI service;

[0059] a processing module configured to determine the first model ID of the AI model.

[0060] In a possible implementation manner based on the sixth aspect, the transceiver is further configured to receive a registration request from the first apparatus, the registration request being used to request registration of the AI service, and the AI model is an AI model corresponding to the AI service; and the transceiver is further configured to send the task ID of the AI service to the first apparatus.

[0061] In a possible implementation manner based on the sixth aspect, the transceiver is further configured to send the region ID of the second apparatus to the first apparatus.

[0062] In a possible implementation manner based on the sixth aspect, the processing module is further configured to select the third apparatus according to at least one of local data information of the third apparatus, computing power information, and communication information between the second apparatus and the third apparatus; and the transceiver is further configured to send the user ID of the third apparatus to the first apparatus.

[0063] In a possible implementation manner based on the sixth aspect, the transceiver is further configured to receive fourth indication information from the first apparatus, the fourth indication information being used to indicate that the AI model is to be trained, and the fourth indication information includes the first model ID; and the processing module is further configured to train the AI model together with the first apparatus and the third apparatus to obtain an updated AI model.

[0064] In a possible implementation manner based on the sixth aspect, the processing module is further configured to generate a second model ID of the AI model according to the first model ID, the length of the second model ID being less than the length of the first model ID.

[0065] In a possible implementation manner based on the sixth aspect, the transceiver is further configured to send the second model ID to the third apparatus.

[0066] In a possible implementation manner based on the sixth aspect, the transceiver is further configured to: receive fifth indication information from the first device, the fifth indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the fifth indication information including the first model ID; and send sixth indication information to the third device, the sixth indication information being used to indicate that the first management operation is performed on the AI model, and the sixth indication information including the second model ID.

[0067] The seventh aspect of the present application provides a third device, including:

[0068] The transceiver is configured to receive a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of a second device and a user ID of the third device and a task ID of an AI service;

[0069] The processing module is configured to determine the first model ID of the AI model.

[0070] In a possible implementation manner based on the seventh aspect, the transceiver is further configured to: receive first indication information from the first device, the first indication information being used to indicate that the AI model is trained, and the first indication information including the first model ID; and the processing module is further configured to: train the AI model based on the first model ID and the second device, to obtain an updated AI model.

[0071] In a possible implementation manner based on the seventh aspect, the transceiver is further configured to: receive a second model ID of the AI model from the first device, the second model ID having a length smaller than that of the first model ID; or the processing module is further configured to: generate the second model ID of the AI model according to the first model ID, the second model ID having a length smaller than that of the first model ID.

[0072] In a possible implementation manner based on the seventh aspect, the transceiver is further configured to: receive third indication information from the first device, the third indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the third indication information including the second model ID; and the processing module is further configured to: perform the first management operation on the AI model based on the second model ID.

[0073] In a possible implementation manner based on the seventh aspect, the transceiver is further configured to receive sixth indication information from the second device, the sixth indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the sixth indication information including the second model ID; and the processor is further configured to perform the first management operation on the AI model based on the second model ID.

[0074] In a possible implementation manner based on the seventh aspect, the first model ID corresponds to a first training path, and devices on the first training path are configured to perform corresponding training operations on the AI model; the transceiver is further configured to receive a third model ID of the AI model from the fifth device or the first device, the third model ID corresponding to a second training path, and devices on the second training path are configured to perform training operations on the AI model; and the processor is further configured to determine that the first model ID and the third model ID are both used to identify the AI model.

[0075] In a possible implementation manner based on the seventh aspect, the transceiver is further configured to receive seventh indication information from the fifth device, the seventh indication information being used to indicate that the AI model is trained, and the seventh indication information including the third model ID; the processor is further configured to train the AI model based on the third model ID and the fifth device; or the processor is further configured to train the AI model based on the third model ID, the fifth device, and the first device.

[0076] An eighth aspect of the present application provides a fourth device, including:

[0077] a transceiver configured to receive a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of a second device and a user ID of a third device and a task ID of an AI service;

[0078] a processor configured to determine the first model ID of the AI model.

[0079] In a possible implementation manner based on the eighth aspect, the first model ID is further generated according to a model version ID of the AI model. The transceiver is further configured to send the model version ID of the AI model to the first device.

[0080] In a possible implementation manner based on the eighth aspect, the transceiver is further configured to send second indication information to the first device, the second indication information being used to indicate that a first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the second indication information including the first model ID.

[0081] For the fifth aspect, the first apparatus can be an AI server or a network device, or a component (e.g., a processor, a chip, or a chip system, etc.) in the AI server or the network device, or a logic module or software capable of implementing all or part of the AI server function, or a logic module or software capable of implementing all or part of the network device function. The transceiver module can be a transceiver, or an input / output interface; and the processing module can be a processor.

[0082] In an implementation manner, the first apparatus is a chip, a chip system, or a circuit configured in an AI server or a network device. When the first apparatus is a chip, a chip system, or a circuit configured in an AI server or a network device, the transceiver module can be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, or a related circuit on the chip, the chip system, or the circuit; and the processing module can be a processor, a processing circuit, or a logic circuit, etc.

[0083] For the sixth aspect, the second apparatus can be a network device, or a component (e.g., a processor, a chip, or a chip system, etc.) in the network device, or a logic module or software capable of implementing all or part of the network device function. The transceiver module can be a transceiver, or an input / output interface; and the processing module can be a processor.

[0084] In an implementation manner, the second apparatus is a chip, a chip system, or a circuit configured in a network device. When the second apparatus is a chip, a chip system, or a circuit configured in a network device, the transceiver module can be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, or a related circuit on the chip, the chip system, or the circuit; and the processing module can be a processor, a processing circuit, or a logic circuit, etc.

[0085] For the seventh aspect, the third apparatus can be a terminal device, or a component (e.g., a processor, a chip, or a chip system, etc.) in the terminal device, or a logic module or software capable of implementing all or part of the terminal device function. The transceiver module can be a transceiver, or an input / output interface; and the processing module can be a processor.

[0086] In an implementation manner, the third apparatus is a chip, a chip system, or a circuit configured in a terminal device. When the third apparatus is a chip, a chip system, or a circuit configured in a terminal device, the transceiver module can be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, or a related circuit on the chip, the chip system, or the circuit; and the processing module can be a processor, a processing circuit, or a logic circuit, etc.

[0087] For the eighth aspect, the fourth apparatus can be an AI server, or a component (e.g., a processor, a chip, or a chip system, etc.) in the AI server, or a logic module or software capable of implementing all or part of the AI server functions. The transceiver module can be a transceiver, or an input / output interface; and the processing module can be a processor.

[0088] In an implementation manner, the fourth apparatus is a chip, a chip system, or a circuit configured in the AI server. When the fourth apparatus is a chip, a chip system, or a circuit configured in the AI server, the transceiver module can be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, or a related circuit on the chip, the chip system, or the circuit; and the processing module can be a processor, a processing circuit, or a logic circuit, etc.

[0089] The ninth aspect of the present application provides a first apparatus, which comprises a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to invoke and run the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementation manners of the first aspect.

[0090] Optionally, the first apparatus further comprises a transceiver, and the processor is configured to control the transceiver to transceive signals.

[0091] The tenth aspect of the present application provides a second apparatus, which comprises a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to invoke and run the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementation manners of the second aspect.

[0092] Optionally, the second apparatus further comprises a transceiver, and the processor is configured to control the transceiver to transceive signals.

[0093] The eleventh aspect of the present application provides a third apparatus, which comprises a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to invoke and run the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementation manners of the third aspect.

[0094] Optionally, the third apparatus further comprises a transceiver, and the processor is configured to control the transceiver to transceive signals.

[0095] The twelfth aspect of the present application provides a fourth apparatus, which comprises a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to invoke and run the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementation manners of the fourth aspect.

[0096] Optionally, the fourth apparatus further comprises a transceiver, and the processor is configured to control the transceiver to transceive signals.

[0097] The thirteenth aspect of the present application provides a first apparatus comprising a processor and an interface circuit, wherein the processor is configured to communicate with other apparatuses via the interface circuit and perform the method of the first aspect. The processor comprises one or more processors.

[0098] The fourteenth aspect of the present application provides a second apparatus comprising a processor and an interface circuit, wherein the processor is configured to communicate with other apparatuses via the interface circuit and perform the method of the second aspect. The processor comprises one or more processors.

[0099] The fifteenth aspect of the present application provides a third apparatus comprising a processor and an interface circuit, wherein the processor is configured to communicate with other apparatuses via the interface circuit and perform the method of the third aspect. The processor comprises one or more processors.

[0100] The sixteenth aspect of the present application provides a fourth apparatus comprising a processor and an interface circuit, wherein the processor is configured to communicate with other apparatuses via the interface circuit and perform the method of the fourth aspect. The processor comprises one or more processors.

[0101] The seventeenth aspect of the present application provides a first apparatus comprising a processor, wherein the processor is configured to be connected with a memory, and to invoke a program stored in the memory to perform the method of the first aspect. The memory can be located in the first apparatus or outside the first apparatus. The processor comprises one or more processors.

[0102] The eighteenth aspect of the present application provides a second apparatus comprising a processor, wherein the processor is configured to be connected with a memory, and to invoke a program stored in the memory to perform the method of the second aspect. The memory can be located in the second apparatus or outside the second apparatus. The processor comprises one or more processors.

[0103] The nineteenth aspect of the present application provides a third apparatus comprising a processor, wherein the processor is configured to be connected with a memory, and to invoke a program stored in the memory to perform the method of the third aspect. The memory can be located in the third apparatus or outside the third apparatus. The processor comprises one or more processors.

[0104] The twentieth aspect of the present application provides a fourth apparatus comprising a processor, wherein the processor is configured to be connected with a memory, and to invoke a program stored in the memory to perform the method of the fourth aspect. The memory can be located in the fourth apparatus or outside the fourth apparatus. The processor comprises one or more processors.

[0105] In an implementation form of the first device according to the first aspect or the fifth aspect, the first device can be a chip or a chip system. In an implementation form of the second device according to the second aspect or the sixth aspect, the second device can be a chip or a chip system. In an implementation form of the third device according to the third aspect or the seventh aspect, the third device can be a chip or a chip system. In an implementation form of the fourth device according to the fourth aspect or the eighth aspect, the fourth device can be a chip or a chip system.

[0106] The twenty-first aspect of the present application provides a computer program product comprising computer instructions, characterized in that when the instructions are run on a computer, the computer is caused to perform any implementation form of any one of the first aspect and the fourth aspect.

[0107] The twenty-second aspect of the present application provides a computer-readable storage medium comprising computer instructions, characterized in that when the instructions are run on a computer, the computer is caused to perform any implementation form of any one of the first aspect and the fourth aspect.

[0108] The twenty-third aspect of the present application provides a chip device comprising a processor, configured to invoke a computer program or computer instructions in a memory, so that the processor performs any implementation form of any one of the first aspect and the fourth aspect.

[0109] Optionally, the processor is coupled to the memory through an interface.

[0110] The twenty-fourth aspect of the present application provides a communication system comprising the first device according to the first aspect and the third device according to the third aspect. Optionally, the communication system further comprises the second device according to the second aspect, or further comprises the fourth device according to the fourth aspect.

[0111] According to the above technical solution, the first device obtains at least one of the region ID of the second device and the user ID of the third device and the task ID of the AI service. Then, the first device generates a first model ID of an AI model according to at least one of the region ID and the user ID and the task ID. The AI model is used to provide the AI task, and the first model ID is used to identify the AI model. Then, the first device sends the first model ID to the third device. Therefore, the first device generates the first model ID of the AI model in combination with at least one of the region ID and the user ID and the task ID. The first device sends the first model ID to the third device. Thus, the AI model is identified through the first model ID. BRIEF DESCRIPTION OF DRAWINGS

[0112] Figure 1 FIG. 1 is a schematic diagram of a communication system according to an embodiment of the present application;

[0113] Figure 2Another schematic diagram of a communication system according to embodiments of the application;

[0114] Figure 3 A schematic diagram of federated learning according to embodiments of the application;

[0115] Figure 4 A schematic diagram of split learning according to embodiments of the application;

[0116] Figure 5 A schematic diagram of decentralized learning according to embodiments of the application;

[0117] Figure 6A An embodiment schematic diagram of a model identification method according to embodiments of the application;

[0118] Figure 6B A schematic diagram of a first model ID according to embodiments of the application;

[0119] Figure 7 Another embodiment schematic diagram of a model identification method according to embodiments of the application;

[0120] Figure 8 A schematic diagram of a first apparatus according to embodiments of the application;

[0121] Figure 9 A schematic diagram of a second apparatus according to embodiments of the application;

[0122] Figure 10 A schematic diagram of a third apparatus according to embodiments of the application;

[0123] Figure 11 A schematic diagram of a fourth apparatus according to embodiments of the application;

[0124] Figure 12 A schematic diagram of an apparatus according to embodiments of the application;

[0125] Figure 13 A schematic diagram of a terminal device according to embodiments of the application;

[0126] Figure 14 A schematic diagram of a network device according to embodiments of the application. DETAILED DESCRIPTION

[0127] Embodiments of the application provide a model identification method and related apparatus, for a first apparatus to generate a first model ID of an AI model according to at least one of a region ID of a second apparatus and a user ID of a third apparatus and a task ID of an AI service, and send the first model ID to the third apparatus. Thus, the AI model is identified or recognized through the first model ID.

[0128] With reference to the drawings and the embodiments described herein, it will be understood that the drawings and embodiments are illustrative of only a portion of the application and are not therefore to be considered to be limiting the scope of the application in any way. Various changes and modifications can be made with respect to the embodiments described herein and the scope of the application should not be considered limited to the embodiments described herein.

[0129] Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment, but can refer to different embodiments. The terms "including", "comprising", "having" and variations thereof are meant to encompass the terms "consisting of" and "consisting essentially of" unless otherwise noted.

[0130] In the description of the application, unless otherwise stated, " / " means "or", for example, A / B can mean A or B. "And / or" in this text is only a description of the relationship between the associated objects, which means that there can be three relationships, for example, A and / or B, which can mean: A alone, A and B together, B alone, these three cases. In addition, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a and b and c. Where a, b, c can be single or multiple.

[0131] It can be understood that in this application, "indication" can include direct indication, indirect indication, display indication, and implicit indication. When describing a certain indication information for indicating A, it can be understood that the indication information carries A, directly indicates A, or indirectly indicates A.

[0132] In the present application, the information indicated by the indication information is referred to as to-be-indicated information. In the specific implementation process, there are many ways to indicate the to-be-indicated information, for example, but not limited to, the to-be-indicated information can be directly indicated, such as the to-be-indicated information itself or the index of the to-be-indicated information, or the to-be-indicated information can be indirectly indicated by indicating other information, wherein the other information and the to-be-indicated information have an association relationship. It can also only indicate part of the to-be-indicated information, and the other part of the to-be-indicated information is known or agreed in advance. For example, the indication of a specific information can also be achieved by means of the arrangement order of each information agreed in advance (for example, the protocol stipulates), thereby reducing the indication overhead to a certain extent.

[0133] The to-be-indicated information can be sent together as a whole, or can be sent separately into multiple sub-information, and the sending period and / or sending occasion of the sub-information can be the same or different. The specific sending method is not limited in the present application. The sending period and / or sending occasion of the sub-information can be predefined, for example, predefined according to the protocol, or configured by the transmitting end device by sending configuration information to the receiving end device.

[0134] It can be understood that the "sending" and "receiving" in the present application represent the direction of signal transmission. For example, "sending information to XX" can be understood as that the destination of the information is XX, which can include direct sending through the air interface, or indirect sending through the air interface by other units or modules. "Receiving information from YY" can be understood as that the source of the information is YY, which can include direct receiving from YY through the air interface, or indirect receiving from YY through the air interface from other units or modules. "Sending" can also be understood as "output" of the chip interface, and "receiving" can also be understood as "input" of the chip interface.

[0135] In other words, the sending and receiving can be between devices, for example, between network devices and terminal devices, or within a device, for example, between components, modules, chips, software modules or hardware modules in the device through a bus, wire or interface.

[0136] It can be understood that the information between the source and the destination of the information transmission can be processed as necessary, such as encoding, modulation, etc., but the destination can understand the effective information from the source. Similar expressions in the present application can be understood similarly, and will not be repeated here.

[0137] The technical solutions of the present application can be applied to a cellular communication system related to the 3rd generation partnership project (3GPP). For example, a fourth generation (4G) communication system, a fifth generation (5G) communication system, a communication system after the fifth generation communication system. For example, a sixth generation communication system. For example, the fourth generation communication system can include a long term evolution (LTE) communication system. The fifth generation communication system can include a new radio (NR) communication system. The technical solutions of the present application can also be applied to a wireless fidelity (WiFi) system, a communication system supporting multiple wireless technology fusion, a device-to-device (D2D) system, a vehicle to everything (V2X) communication system, and the like.

[0138] The communication system to which the technical solutions of the present application are applicable includes a first device and a third device. Two possible implementation manners of the first device are introduced below.

[0139] In one possible implementation manner, the first device is a first network device, or a chip, a chip system, or a processor in the first network device, or a logic module or software for realizing part or all functions of the first network device.

[0140] Optionally, the third device is a terminal device, or a chip, a chip system, or a processor in the terminal device, or a logic module or software for realizing part or all functions of the terminal device.

[0141] Optionally, the communication system further includes a fourth device. Optionally, the fourth device is an AI server, or a chip, a chip system, or a processor in the AI server, or a logic module or software for realizing part or all functions of the AI server. The AI server is a device with AI function. The AI server can be an AI server of a third-party AI service manufacturer. The AI server can be referred to as an AI device or an AI application server, and the present application does not limit the name of the AI server.

[0142] Optionally, the communication system further includes a fifth device. Optionally, the fifth device is a second network device, or a chip, a chip system, or a processor in the second network device, or a logic module or software for realizing part or all functions of the second network device.

[0143] In another possible implementation, the first device is an AI server, or a chip, chip system, or processor within an AI server; or a logic module or software that implements part or all of the AI ​​server. Please refer to the foregoing related introduction regarding AI servers.

[0144] Optionally, the third device is a terminal device, or a chip, chip system, or processor in the terminal device; or a logic module or software that implements some or all of the functions of the terminal device.

[0145] Optionally, the communication system further includes a second device. Optionally, the second device is the first network device, or a chip, chip system, or processor in the first network device; or a logic module or software that implements some or all of the functions of the first network device, etc.

[0146] Optionally, the communication system may also include a fifth device. Optionally, the fifth device may be a second network device, or a chip, chip system, or processor in the second network device; or a logic module or software that implements some or all of the functions of the second network device, etc.

[0147] The following describes some communication systems to which this application applies. This application also applies to other communication systems, but no specific limitations are imposed.

[0148] Figure 1 This is a schematic diagram of a communication system according to an embodiment of this application. Please refer to... Figure 1 The communication system includes a terminal device 101, a network device 102, and an AI server 103. The technical solutions of this application can be executed between the terminal device 101, the network device 102, and the AI ​​server 103.

[0149] In one possible implementation, the first device is a network device 102, the third device is a terminal device 101, and the fourth device is an AI server 103. In another possible implementation, the first device may be an AI server 103, the second device may be a network device 102, and the third device may be a terminal device 101.

[0150] It should be noted that the above Figure 1 The communication system shown is merely an example. In practical applications, the communication system can also include more terminal devices, more network devices, and more AI servers; this application does not limit the specifics.

[0151] Figure 2 This is another schematic diagram of the communication system according to an embodiment of this application. Please refer to... Figure 2 The communication system includes terminal equipment 201, terminal equipment 202, network equipment 203, network equipment 204, network equipment 205, AI server 206, and AI server 207.

[0152] The AI model 1 can be trained by devices on the training path 1 or the training path 2. The training path 1 is the terminal device 201-network device 203-AI server 206. The AI model 1 can be trained among the terminal device 201, the network device 203 and the AI server 206. The training path 2 is the terminal device 201-network device 204-AI server 206. The AI model 1 can be trained among the terminal device 201, the network device 204 and the AI server 206. The terminal device 201, the network device 205 and the AI server 207 can train the AI model 2. The terminal device 202, the network device 205 and the AI server 207 can train the AI model 3.

[0153] In a possible implementation, the first device is the network device 203, the third device is the terminal device 201, and the fourth device is the AI server 206. Optionally, the fifth device is the network device 204.

[0154] In another possible implementation, the first device is the network device 205, the third device is the terminal device 201 or the terminal device 202, and the fourth device is the AI server 207.

[0155] In yet another possible implementation, the first device is the AI server 206, the second device is the network device 203 or the network device 204, and the third device is the terminal device 201. Optionally, if the second device is the network device 203, the fifth device can be the network device 204; or if the second device is the network device 204, the fifth device can be the network device 203.

[0156] In yet another possible implementation, the first device is the AI server 207, the second device is the network device 205, and the third device is the terminal device 201 or the terminal device 202.

[0157] It should be noted that, Figure 2 The communication system shown is only an example. In actual applications, the communication system includes at least one network device, at least one terminal device and at least one AI server, and the specific application is not limited in the present application.

[0158] The terminal device, the network device and the AI server involved in the present application are introduced as follows.

[0159] The terminal device can be a wireless terminal device capable of receiving scheduling information and indication information of the network device. The wireless terminal device can be a device that provides voice and / or data connectivity to a user, or a handheld device with wireless connection function, or other processing devices connected to a wireless modem.

[0160] The terminal device can communicate with one or more core networks or the Internet via an access network. The terminal device can be a mobile terminal device, such as a mobile phone (or called "cellular" phone, mobile phone), a computer and a data card, for example, can be a portable, pocket-sized, handheld, computer-built-in or vehicle-mounted mobile device, which exchanges voice and / or data with a wireless access network. For example, personal communication service (PCS) phones, cordless phones, session initiation protocol phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), tablets (Pads), computers with wireless transceiver functions, and the like. The wireless terminal device can also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile station (MS), a remote station, an access point (AP), a remote terminal, an access terminal, a user terminal, a user agent, a subscriber station (SS), customer premises equipment (CPE), a terminal, user equipment (UE), a mobile terminal (MT), and the like.

[0161] By way of example and not limitation, in embodiments of the present application, the terminal device can also be a wearable device. The wearable device can also be referred to as a wearable smart device or a smart wearable device, etc. It is a general term for devices that are designed and developed by applying wearable technology to daily wear. For example, glasses, gloves, watches, clothing, and shoes, etc. The wearable device is a portable device that is directly worn on the body or integrated into the user's clothes or accessories. The wearable device is not only a hardware device, but also a powerful function realized through software support and data interaction, cloud interaction. The broad sense of wearable smart devices includes devices with full functions, large sizes, and the ability to realize complete or partial functions without relying on smart phones, such as smart watches or smart glasses, etc., and devices that focus on a certain application function and need to be used with other devices such as smart phones, such as various smart wristbands, smart helmets, smart jewelry, etc. for monitoring vital signs.

[0162] The terminal device can also be a drone, a robot, a terminal device in device-to-device (D2D) communication, a terminal device in vehicle to everything (V2X), a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal device in industrial control, a wireless terminal device in self driving, a wireless terminal device in remote medical treatment, a wireless terminal in smart grid, a wireless terminal device in transportation safety, a wireless terminal device in smart city, a wireless terminal device in smart home, and the like.

[0163] In addition, the terminal device can also be a terminal device in a communication system evolved after the 5th generation (5G) communication system (for example, a 6th generation (6G) communication system, etc.), or a terminal device in a future evolved public land mobile network (PLMN), and the like. For example, the 6G network can further expand the form and function of the 5G communication terminal, and the 6G terminal includes but is not limited to a vehicle, a cellular network terminal (integrating satellite terminal function), a drone, or an internet of things (IoT) device.

[0164] In the embodiments of the present application, the terminal device has artificial intelligence (AI) capability. For example, the terminal device can obtain AI services provided by a network device or a server. The terminal device also has AI processing capability.

[0165] It should be noted that the terminal device can be a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, module or control unit in the above-mentioned devices or apparatus, and the specific embodiments of the present application are not limited.

[0166] The network device can be a device in a wireless network. For example, the network device can be an access network node (or an access network device) that accesses a terminal device to a wireless network, which can also be referred to as a base station. Currently, some examples of the access network device are: a base station (gNodeB, gNB) in a 5G communication system, a transmission reception point (TRP), an evolved Node B (eNB), a radio network controller (RNC), a Node B (NB), a home base station (for example, a home evolved Node B, or a home Node B, HNB), a baseband unit (BBU), or a wireless fidelity (Wi-Fi) access point AP, and the like. In addition, in a network structure, the network device can include a centralized unit (CU) node, a distributed unit (DU) node, a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU), or a RAN device including the CU node and the DU node. The CU and the DU can be separately arranged, or can also be included in the same network element, such as a baseband unit (BBU). The RU can be included in a radio frequency device or a radio frequency unit, such as a remote radio unit (RRU), an active antenna processing unit (AAU), or a remote radio head (RRH). In different systems, the CU (or CU-CP and CU-UP), DU or RU can also have different names, but those skilled in the art can understand their meanings. For example, in an open RAN (ORAN) system, the CU can also be referred to as an open CU (O-CU), the DU can also be referred to as an open DU (O-DU), the CU-CP can also be referred to as an open CU-CP (O-CU-CP), the CU-UP can also be referred to as an open CU-UP (O-CU-UP), and the RU can also be referred to as an open RU (O-RU). Among them, any one of the CU (or CU-CP, CU-UP), DU and RU can be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0167] The network device can be other apparatuses providing wireless communication functions for terminal devices. Embodiments of the present application do not limit the specific technology and specific device form of the network device. For the convenience of description, embodiments of the present application do not limit.

[0168] The network device can also include a core network device, for example, including a mobility management entity (MME) in a fourth generation (4th generation, 4G) network, a home subscriber server (HSS), a serving gateway (S-GW), a policy and charging rules function (PCRF), a public data network gateway (PDN gateway, P-GW), a network element such as an access and mobility management function (AMF), a user plane function (UPF), or a session management function (SMF) in a 5G network. In addition, the core network device can also include other core network devices in a 5G network and a next-generation network (such as a 6G network) of the 5G network.

[0169] In embodiments of the present application, the network device described above can also be an AI-capable network node, which can provide AI services for terminal devices or other network devices. For example, the network device can be an AI node, a computing power node, an AI-capable access network node, or an AI-capable core network element of the network side (access network or core network).

[0170] It should be noted that the network device can be the device or apparatus shown above, or a component (such as a chip), module, or unit of the device or apparatus shown above, and the specific application does not limit this.

[0171] The AI server is a device with AI function, which can instruct the network device and the terminal device to perform corresponding management operations on the AI model.

[0172] Distributed learning can complete the learning task of the AI model under the premise of fully guaranteeing the privacy and security of user data. Distributed learning mainly includes federated learning, split learning, and decentralized learning. Federated learning, split learning, and decentralized learning are introduced below.

[0173] I. Federated Learning

[0174] Federated learning is a typical distributed learning method, which promotes the cooperation of each distributed node and the center node to efficiently complete the learning task of the model under the premise of fully guaranteeing the privacy and security of user data. Figure 3 As shown in FIG. 1, the communication system includes a center node and one or more distributed nodes. For example, as shown in FIG. 2, the communication system includes distributed node n, distributed node k and distributed node m. Each distributed node collects a local data set and performs local training of the model to obtain local parameters. Then, the distributed node sends the local parameters to the center node. The center node itself has no data set, and the center node collects the local parameters reported by the plurality of distributed nodes and performs fusion processing on the local parameters reported by the plurality of distributed nodes to obtain global parameters, and distributes the global parameters to each distributed node. Thus, the training and learning of the model are realized. Figure 3 As shown in FIG. 1, the communication system includes a center node and one or more distributed nodes. For example, as shown in FIG. 2, the communication system includes distributed node n, distributed node k and distributed node m. Each distributed node collects a local data set and performs local training of the model to obtain local parameters. Then, the distributed node sends the local parameters to the center node. The center node itself has no data set, and the center node collects the local parameters reported by the plurality of distributed nodes and performs fusion processing on the local parameters reported by the plurality of distributed nodes to obtain global parameters, and distributes the global parameters to each distributed node. Thus, the training and learning of the model are realized.

[0175] II. Split learning.

[0176] In split learning, the complete neural network model is split into multiple parts. For example, taking two parts as an example, that is, the neural network model is split into two sub-networks. One part is deployed on the distributed node, and the other part is deployed on the center node. The place where the complete neural network model is split can be referred to as a split layer.

[0177] As shown in FIG. 3, in the forward inference of the model, distributed node 1 inputs the local data into the local sub-network and infers to the split layer to obtain the result F1 of the split layer output. Distributed node 1 sends F1 to the center node through the communication link. The center node inputs the received F1 into another sub-network deployed by the center node and continues to perform forward inference to obtain the final inference result. In the gradient back propagation of the model training, the center node performs back propagation through another sub-network deployed by the center node to the split layer to obtain the back propagation result G1. Then, the center node sends G1 to distributed node 1. Distributed node 1 continues to perform gradient back propagation through the sub-network deployed by the distributed node based on G1. The forward inference and gradient back propagation between the other distributed nodes and the center node are similar. Figure 4 As can be seen, the forward inference process and the gradient back propagation process of split learning only involve one distributed node and one center node. The trained sub-network on the distributed node can be saved locally on the distributed node or on a specific model storage server. When a new distributed node joins the split learning, the new distributed node can download the trained sub-network and further train the sub-network using the local data of the new distributed node.

[0178] III. Decentralized learning.

[0179]

[0180] ​Unlike federated learning, decentralized learning is a learning method without a central node. As shown in Figure 5 , the design goal f(x) of decentralized learning is generally the average of the local goals f i (x) obtained by each node, i.e. where n is the number of distributed nodes, and x is the parameter to be optimized. In machine learning, x is the parameter of a machine learning (such as a neural network) model. Each node calculates the local gradient i using the local data and the local goal f and sends the local gradient to its neighbor nodes. After receiving the local gradient of its neighbor nodes, any node can update the parameter x of the local model according to the following formula 1. Thus, the learning task of the model is completed through the information interaction between nodes.

[0181]

[0182] where N i is the neighbor node set of node i, |N i | represents the number of elements in the neighbor node set of node i, i.e. the number of neighbor nodes of node i, and a k is the fusion weight of the kth round of training. is the model parameter of node i obtained in the k+1th round of training. is the model parameter of node i obtained in the kth round of training. is the model parameter of neighbor node j in the neighbor node set obtained in the kth round of training. k is an integer greater than or equal to 1.

[0183] However, different AI models can have different functions. For example, an AI model can be used for CSI feedback enhancement between network devices and terminal devices, beam management, positioning enhancement, or CSI compressed feedback, etc. Therefore, for terminal devices or network devices, how to identify different AI models is a problem worth considering.

[0184] The present application provides corresponding technical solutions for a first device to generate an AI model first model ID according to at least one of a region ID and a user ID and a task ID, and send the first model ID to a third device. Thus, the third device can identify the AI model through the first model ID. For details, please refer to the related introduction in the embodiments below.

[0185] The technical solutions of the present application will be described below in conjunction with specific embodiments.

[0186] Figure 6A An embodiment of the model identification method of the present application is shown in the figure. Please refer to Figure 6A , the method includes the following steps. Optionally,Figure 6A In the illustrated embodiment, the first device can be a first network device, or a chip, a chip system, or a processor in the first network device, or a logic module or software for implementing part or all of the functions of the first network device. The third device is a terminal device, or a chip, a chip system, or a processor in the terminal device, or a logic module or software for implementing part or all of the functions of the terminal device. The fourth device can be an AI server, or a chip, a chip system, or a processor in the AI server, or a logic module or software for implementing part or all of the functions of the AI server. The fifth device is a second network device, or a chip, a chip system, or a processor in the second network device, or a logic module or software for implementing part or all of the functions of the second network device.

[0187] 601. The first device obtains at least one of a region ID of the second device, a user ID of the third device, and a task ID of the AI service.

[0188] The first device and the second device are the same device. For example, the first device and the second device are both first network devices. The region ID is used to indicate the region in which the second device is located. The region ID of the second device can be a cell ID of the first network device.

[0189] Optionally, Figure 6A The illustrated embodiment further includes step 601a. Step 601a can be performed before step 601.

[0190] 601a. The fourth device sends a registration request to the first device. Correspondingly, the first device receives the registration request from the fourth device.

[0191] The registration request is used to request registration of the AI service. For example, the AI service can be an AI model training service or an AI model inference service.

[0192] Optionally, in step 601, the first device obtaining the task ID of the AI service specifically includes: the first device assigning the task ID to the AI service.

[0193] Optionally, Figure 6A The illustrated embodiment further includes step 601b. Step 601b can be performed before or after step 601.

[0194] 601b. The first device sends the task ID to the fourth device. Correspondingly, the fourth device receives the task ID from the first device.

[0195] Optionally, Figure 6A The illustrated embodiment further includes step 601c. Step 601c can be performed before or after step 601.

[0196] 601c, the first device sends a user ID of the third device to the third device. Correspondingly, the third device receives the user ID from the first device.

[0197] Optionally, the first device selects the third device according to local data information, computing power information of the third device, and / or communication information between the first device and the third device, and allocates a user ID to the third device. For example, the local data information of the third device can be used to train an AI model corresponding to the AI service, etc. The first device can preferentially select the third device. For another example, the third device has stronger computing power, and the channel quality of the communication link between the first device and the third device is better, so the first device can preferentially select the third device to participate in training, inference, etc. of the AI model. Optionally, the first device sends the user ID to the third device.

[0198] It should be noted that, if Figure 6A The embodiments shown in FIG. 6 include step 601b, and there is no fixed execution order between step 601b and step 601c. Step 601b can be executed first, and then step 601c can be executed. Alternatively, step 601c can be executed first, and then step 601b can be executed. Alternatively, steps 601b and 601c can be executed simultaneously according to the situation, and the specific application is not limited.

[0199] It should be noted that, in the above step 601, the first device can obtain at least one of the region ID of the second device and the user ID of the third device and the task ID at the same time, or can obtain at least one of the region ID of the second device and the user ID of the third device and the task ID separately, and the specific application is not limited. For example, the first device obtains the task ID first, then obtains the region ID, and finally obtains the user ID. Alternatively, the first device obtains the task ID first, then obtains the user ID, and finally obtains the region ID, and the specific application is not limited.

[0200] 602, the first device generates a first model ID of the AI model according to at least one of the region ID and the user ID and the task ID.

[0201] The AI model is used to provide an AI service. The first model ID is used to identify or identify the AI model.

[0202] In one possible implementation, the first device combines at least one of the region ID and the user ID and the task ID to obtain the first model ID. For example, as shown in FIG. 6B, the first model ID is obtained by combining the task ID, the region ID, and the user ID. Figure 6B

[0203] ​In another possible implementation, the first device obtains the first model ID by performing a corresponding operation on at least one of the region ID and the user ID and the task ID. For example, the first device obtains the first model ID by performing a hash operation on at least one of the region ID and the user ID and the task ID.

[0204] Optionally, before the step 602, the fourth device sends the AI model to the first device. For example, for an AI model that has been trained, the fourth device can deploy the trained model parameters to the AI model and send the AI model to the first device. For another example, for an AI model to be trained, the model parameters in the AI model are randomly initialized, that is, the fourth device initializes the AI model and sends the AI model to the first device.

[0205] Optionally, the step 602 specifically includes that the first device further generates the first model ID according to a model version ID and / or a training round ID of the AI model. For example, the first device combines the region ID, the user ID, the task ID, the model version ID, and the training round ID to obtain the first model ID. For another example, the first device obtains the first model ID by performing a corresponding operation on the region ID, the user ID, the task ID, the model version ID, and the training round ID.

[0206] If the AI model is a model to be trained, the first device can determine the training round ID. For example, if the AI model is trained for the first time, the training round ID is an initial training round ID. If the AI model is not trained for the first time, the first device updates the training round ID corresponding to the last training to obtain a training round ID corresponding to the current training. Therefore, each training of the AI model has a corresponding training round ID, and thus the training round ID can indicate the AI model obtained by the training corresponding to the training round ID.

[0207] Optionally, Figure 6A The illustrated embodiment also includes a step 602a. The step 602a can be performed before the step 602.

[0208] 602a. The fourth device sends a model version ID of the AI model to the first device. Correspondingly, the first device receives the model version ID of the AI model from the fourth device.

[0209] For example, when the AI model inference is performed, i.e., the AI model is an AI model that has been trained, the fourth device can deploy the trained model parameter to the AI model; the fourth device assigns a corresponding model version ID to the AI model, and sends the model version ID to the first device. For another example, when the AI model training is performed, i.e., the AI model is an AI model to be trained, the model parameter in the AI model is randomly initialized, i.e., the fourth device initializes the AI model. The fourth device assigns a corresponding model version ID to the AI model, and sends the model version ID to the first device.

[0210] It should be noted that the fourth device can send the AI model and the model version ID to the first device at the same time, or can send the AI model and the model version ID separately, which is not limited in the present application.

[0211] 603、The first device sends the first model ID to the third device. Correspondingly, the third device receives the first model ID from the first device.

[0212] Specifically, after the third device receives the first model ID, the third device determines the first model ID of the AI model. Optionally, the first device sends the AI model to the third device.

[0213] Optionally, before step 603, the first device can send the AI model to the third device. Optionally, the first device can send the AI model and the first model ID to the third device at the same time, or can send the AI model and the first model ID separately, which is not limited in the present application.

[0214] Optionally, Figure 6A The embodiments shown also include step 604. Step 604 can be performed after step 602.

[0215] 604、The first device sends the first model ID to the fourth device. Correspondingly, the fourth device receives the first model ID from the first device.

[0216] Specifically, after the fourth device receives the first model ID, the fourth device determines the first model ID of the AI model.

[0217] It should be noted that there is no fixed execution order between step 603 and step 604. Step 603 can be performed first, and then step 604 can be performed; or step 604 can be performed first, and then step 603 can be performed; or step 603 and step 604 can be performed at the same time according to the situation, which is not limited in the present application.

[0218] Optionally, Figure 6A The embodiments shown also include steps 605 to 607. Steps 605 to 607 can be performed after step 603.

[0219] 605、The first device sends first indication information to the third device. The first indication information includes the first model ID. Correspondingly, the third device receives the first indication information from the first device.

[0220] The first indication information is used to indicate training of the AI model. The first indication information includes the first model ID, thereby indicating training of the AI model corresponding to the first model ID.

[0221] 606、The first device and the third device train the AI model based on the first model ID, to obtain an updated AI model.

[0222] Optionally, after step 606, the first device can update a training round ID, and the updated training round ID is used to identify the next round of training of the AI model. The first device updates the first model ID according to the updated training round ID.

[0223] 607、The first device sends the updated AI model to the fourth device. Correspondingly, the fourth device receives the updated AI model from the first device.

[0224] Optionally, after receiving the updated AI model, the fourth device can update a model version ID of the AI model, and send the updated model version ID to the first device. Thereby, the first device can update the first model ID of the AI model.

[0225] It should be noted that there is no fixed execution order between steps 605 to 607 and step 604. Step 604 can be executed first, and then steps 605 to 607 can be executed. Alternatively, steps 605 to 607 can be executed first, and then step 604 can be executed. Alternatively, steps 604 and steps 605 to 607 can be executed simultaneously according to the situation, and the specific application is not limited.

[0226] Optionally, Figure 6A The embodiments shown also include steps 608 to 609. Steps 608 to 609 can be executed after step 603.

[0227] 608、The first device generates a second model ID of the AI model according to the first model ID.

[0228] The length of the second model ID is less than the length of the first model ID. For example, the first model ID can uniquely identify the AI model in a larger area range, facilitating corresponding management operations of the AI model between different devices. However, in a smaller area range, the AI model can be uniquely identified by a relatively short second model ID.

[0229] In a possible implementation, the first device extracts part of the first model ID to obtain the second model ID. For example, the first model ID is a combination of a region ID, a user ID and a task ID. The first device removes the region ID from the first model ID and combines the remaining part to obtain the second model ID.

[0230] In another possible implementation, the first device can extract part of the first model ID and calculate the second model ID according to the extracted part. For example, the first model is a combination of a region ID, a user ID and a task ID. The first device removes the region ID and calculates the second model ID according to the remaining user ID and task ID.

[0231] Optionally, after determining the second model ID, the first device can establish a mapping relationship between the first model ID and the second model ID.

[0232] 609. The first device sends the second model ID to the third device. Correspondingly, the third device receives the second model ID from the first device.

[0233] It should be noted that, if the first model ID is a combination of a region ID, a user ID and a task ID, the first device can remove the region ID from the first model ID and combine the remaining part to obtain the second model ID. Figure 6A The embodiments shown in FIG. 6 further include step 604, and there is no fixed execution order between step 608 and step 609 and step 604. Step 604 can be performed first, and then step 608 and step 609 can be performed; or step 608 and step 609 can be performed first, and then step 604 can be performed; or step 604 and step 608 and step 609 can be performed at the same time according to the situation, which is not limited in the present application.

[0234] It should be noted that, if the first model ID is a combination of a region ID, a user ID and a task ID, the first device can remove the region ID from the first model ID and combine the remaining part to obtain the second model ID. Figure 6A The embodiments shown in FIG. 6 further include step 605 to step 607, and there is no fixed execution order between step 608 and step 609 and step 605 to step 607. Step 605 to step 607 can be performed first, and then step 608 and step 609 can be performed; or step 608 and step 609 can be performed first, and then step 605 to step 607 can be performed; or step 605 to step 607 and step 608 and step 609 can be performed at the same time according to the situation, which is not limited in the present application.

[0235] It should be noted that, if the first model ID is a combination of a region ID, a user ID and a task ID, the first device can remove the region ID from the first model ID and combine the remaining part to obtain the second model ID. Figure 6AThe illustrated embodiment also includes steps 604 to 607, and there is no fixed execution order between steps 608 to 609. Steps 604 to 607 can be executed first, and then steps 608 to 609 can be executed; or steps 608 to 609 can be executed first, and then steps 604 to 607 can be executed; or steps 604 to 607 and steps 608 to 609 can be executed simultaneously according to circumstances, and the specific application is not limited.

[0236] It should be noted that the above is an example of the first device sending the first model ID and the second model ID to the third device to introduce the technical solutions of the present application. In actual application, the first device can only send the second model ID to the third device. In this implementation mode, the above steps 603 and the above steps 605 to 607 are optional steps.

[0237] It should be noted that the above step 609 is an example of the first device sending the second model ID to the third device to introduce the technical solutions of the present application. In actual application, after the third device receives the first model ID in the above step 603, the third device can generate the second model ID of the AI model based on the first model ID. The process of generating the second model ID by the third device is similar to the process of generating the second model ID by the first device in the foregoing step 608, and details are not repeated here. In this implementation mode, the above step 609 is an optional step.

[0238] Optionally, Figure 6A The illustrated embodiment also includes steps 610 to 612. Steps 610 to 612 can be executed after step 609.

[0239] 610. The fourth device sends second indication information to the first device. Correspondingly, the first device receives the second indication information from the fourth device.

[0240] The second indication information is used to indicate that the first management operation is performed on the AI model. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training.

[0241] Optionally, the second indication information includes the first model ID, and the second indication information is used to indicate that the AI model corresponding to the first model ID is activated. For example, the second indication information includes a first field and a second field, the first field is used to indicate that the first management operation is performed on the AI model, and the second field is used to indicate the first model ID. It should be noted that the second indication information can also indicate the AI model through other ways, that is, the second indication information does not carry the first model ID, but carries other fields to indicate the AI model.

[0242] 611、The first device sends third indication information to the third device. The third indication information includes the second model ID. Correspondingly, the third device receives the third indication information from the first device.

[0243] After the first device receives the second indication information, the first device can determine the second model ID in combination with the first model ID included in the second indication information and the mapping relationship between the first model ID and the second model ID. Then, the first device sends third indication information to the third device. The third indication information includes the second model ID, thereby instructing the third device to perform the first management operation on the AI model. Since the length of the second model ID is less than the length of the first model ID, the above step 611 can reduce the indication overhead of the first device.

[0244] 612、The third device performs the first management operation on the AI model based on the second model ID.

[0245] Specifically, after the third device receives the third indication information, the third device obtains the second model ID and identifies the AI model through the second model ID. Then, the third device performs the first management operation on the AI model. For example, the first management operation includes activation, and the third device activates the AI model.

[0246] Optionally, Figure 6A The embodiments shown also include steps 613 to 614.

[0247] 613、The fifth device sends a third model ID of the AI model to the third device. Correspondingly, the third device receives the third model ID from the fifth device.

[0248] The third model ID corresponds to a second training path, and the devices on the second training path are used to perform training operations on the AI model. The third model ID is used to identify or mark the AI model. Optionally, the first model ID corresponds to a first training path, and the devices on the first training path are used to perform training operations on the AI model. In other words, the first model ID is used to identify or mark the AI model trained by the devices on the first training path, and the third model ID is used to identify or mark the AI model trained by the devices on the second training path. For example, as shown in Figure 2 The first device is a network device 203, the third device is a terminal device 201, and the fifth device is a network device 204. For the terminal device 201, the terminal device 201 can identify or mark the model 1 trained by the devices on the training path 1 through the first model ID, and identify or mark the model 1 trained by the devices on the training path 2 through the third model ID.

[0249] It should be noted that the fifth device generates the third model ID in a process similar to that of the first device generating the first model ID, and specific reference can be made to the foregoing description of the process of the first device generating the first model ID, which will not be repeated here.

[0250] 614、The third device determines that the first model ID and the third model ID are both used to identify the AI model.

[0251] For example, as shown in FIG. 13, the third device is a terminal device 201. For the AI model 1, the terminal device 201 receives the first model ID and the third model ID. The terminal device 201 identifies or recognizes the AI model 1 through the first model ID and the third model ID. That is, for the terminal device 201, the terminal device 201 can identify or recognize the AI model 1 trained by the device on the training path 1 through the first model ID, and identify or recognize the AI model 1 trained by the device on the training path 2 through the third model ID. Figure 2

[0252] It should be noted that steps 613 to 614 have no fixed order between the steps before step 613 in the embodiment shown in FIG. 12. Steps 613 to 614 can be performed first, and then the steps before step 613 in the embodiment shown in FIG. 12 can be performed; or the steps before step 613 in the embodiment shown in FIG. 12 can be performed first, and then steps 613 to 614 can be performed; or steps 613 to 614 and the steps before step 613 in the embodiment shown in FIG. 12 can be performed simultaneously according to the situation, and the specific application will not be limited. Figure 6A Figure 6A Figure 6A Figure 6A

[0253] Optionally, the embodiment shown in FIG. 13 further includes steps 615 to 617. Steps 615 to 617 can be performed after step 614. Figure 6A

[0254] 615、The fifth device sends seventh indication information to the third device. The seventh indication information includes the third model ID. Correspondingly, the third device receives the seventh indication information from the fifth device.

[0255] The seventh indication information is used to indicate that the training operation is performed on the AI model. Optionally, the seventh indication information includes a third field and a fourth field, the third field is used to indicate that the training operation is performed on the AI model, and the fourth field is used to indicate the third model ID of the AI model. It should be noted that the second indication information can also indicate the AI model through other ways, that is, the second indication information does not carry the third model ID, but indicates the AI model through other ways.

[0256] ​​​​​​616、The third device trains the AI model based on the third model ID and the fifth device to obtain an updated AI model.

[0257] For example, in federated learning, the third device trains the AI model based on the third model ID according to local data of the third device to obtain local model parameters. The third device sends the local model parameters to the fifth device. The fifth device fuses the local model parameters to obtain global model parameters, and sends the global model parameters to the third device. The third device takes the global model parameters as the local model parameters of the AI model. As shown in Figure 2 , the third device is a terminal device 201, and the fifth device is a network device 204. The terminal device 201 and the network device 204 train the AI model 1 based on the third model ID to obtain an updated AI model 1.

[0258] 617、The fifth device sends the updated AI model to the fourth device. Correspondingly, the fourth device receives the updated AI model from the fifth device.

[0259] For example, as shown in Figure 2 , the fourth device is an AI server 206, and the fifth device is a network device 204. The network device 204 sends the updated AI model 1 to the AI server 206.

[0260] Optionally, after the fourth device obtains the updated AI model, the fourth device updates the model version ID of the AI model, and sends the updated model version ID to the fifth device. Thus, the fifth device updates the third model ID of the AI model.

[0261] In the embodiment shown in Figure 6A , the first device obtains at least one of the region ID of the second device and the user ID of the third device and the task ID of the AI service. Then, the first device generates the first model ID of the AI model according to at least one of the region ID and the user ID and the task ID. The AI model is used to provide the AI task, and the first model ID is used to identify the AI model. Then, the first device sends the first model ID to the third device. As can be seen, the first device generates the first model ID of the AI model in combination with at least one of the region ID and the user ID and the task ID. The first device sends the first model ID to the third device. Thus, the AI model is identified by the first model ID.

[0262] Figure 7 Another embodiment of the model identification method of the present application is shown in the following Figure 7 . The method includes the following steps. Optionally, Figure 7In the illustrated embodiment, the first device can be an AI server, or a chip, a chip system, or a processor in the AI server, or a logic module or software that implements part or all of the functions of the AI server. The second device is a first network device, or a chip, a chip system, or a processor in the first network device, or a logic module or software that implements part or all of the functions of the first network device. The third device is a terminal device, or a chip, a chip system, or a processor in the terminal device, or a logic module or software that implements part or all of the functions of the terminal device. The fifth device is a second network device, or a chip, a chip system, or a processor in the second network device, or a logic module or software that implements part or all of the functions of the second network device.

[0263] 701. The first device obtains at least one of a region ID of the second device and a user ID of the third device, and a task ID of the AI service.

[0264] Optionally, the step 701 specifically includes that the first device receives at least one of the region ID and the user ID and the task ID from the second device. For example, the first device is a first AI server, and the second device is a first network device. For example, the region ID can be a cell ID of the first network device. The first AI server receives the cell ID from the first network device.

[0265] Optionally, the second device selects the third device according to at least one of local data information of the third device, computing power information, and communication information between the second device and the third device. Then, the second device allocates a user ID for the third device. Then, the second device sends the user ID to the first device. Optionally, the second device sends the user ID to the third device.

[0266] Optionally, Figure 7 The illustrated embodiment also includes a step 701a. The step 701a can be performed before the step 701.

[0267] 701a. The first device sends a registration request to the second device. Correspondingly, the second device receives the registration request from the first device.

[0268] The step 701a is similar to the foregoing Figure 6A The step 601a in the illustrated embodiment is similar, and for details, please refer to the foregoing Figure 6A The step 601a in the illustrated embodiment is similar, and for details, please refer to the foregoing

[0269] 702. The first device generates a first model ID of the AI model according to at least one of the region ID and the user ID, and the task ID of the AI service.

[0270] The step 702 is similar to the foregoing Figure 6AThe step 602 in the illustrated embodiment is similar, and can be specifically referred to the foregoing Figure 6A The related description of the step 602 in the illustrated embodiment is not repeated here.

[0271] Optionally, Figure 7 The illustrated embodiment further includes a step 702a. The step 702a can be performed before the step 702.

[0272] 702a, the first device determines the model version ID and / or the training round ID of the AI model.

[0273] Specifically, the process in which the first device determines the model version ID of the AI model can be referred to the foregoing Figure 6A The related description of the step 602a in the illustrated embodiment. Optionally, the first device determining the training round ID includes: the first device receiving the training round ID from the second device. Optionally, the second device determines the training round ID. The process in which the second device determines the training round ID can be referred to the foregoing Figure 6A The related description of the step 602 in the illustrated embodiment is not repeated here.

[0274] Optionally, the step 702 specifically includes: the first device further generates the first model ID according to the model version ID and / or the training round ID of the AI model. For example, the first device combines the region ID, the user ID, the task ID, the model version ID and the training round ID to obtain the first model ID. For another example, the first device obtains the first model ID according to the region ID, the user ID, the task ID, the model version ID and the training round ID through corresponding operations.

[0275] 703, the first device sends the first model ID to the third device. Correspondingly, the third device receives the first model ID from the first device.

[0276] Optionally, Figure 7 The illustrated embodiment further includes a step 704. The step 704 can be performed after the step 702.

[0277] 704, the first device sends the first model ID to the second device. Correspondingly, the second device receives the first model ID from the first device.

[0278] It should be noted that there is no fixed execution order between the step 703 and the step 704. The step 703 can be performed first, and then the step 704 is performed. Alternatively, the step 704 can be performed first, and then the step 703 is performed. Alternatively, the step 703 and the step 704 are simultaneously performed according to the situation, and the specific application is not limited.

[0279] Optionally, Figure 7The illustrated embodiment also includes steps 705 to 707. Steps 705 to 707 can be performed after step 704.

[0280] 705. The first device sends fourth indication information to the second device. The fourth indication information includes the first model ID. Correspondingly, the second device receives the fourth indication information from the first device.

[0281] The fourth indication information is used to indicate training of the AI model. The fourth indication information includes the first model ID, thereby indicating training of the AI model corresponding to the first model ID.

[0282] 706. The second device sends the fourth indication information to the third device. The fourth indication information includes the first model ID. Correspondingly, the third device receives the fourth indication information from the first device.

[0283] 707. The first device, the second device, and the third device train the AI model based on the first model ID to obtain an updated AI model.

[0284] Specifically, the second device receives the fourth indication information, and the second device identifies the AI model based on the first model ID. The second device sends the fourth indication information to the third device. Then, the third device identifies the AI model based on the first model ID in the fourth indication information. The first device, the second device, and the third device jointly complete the training task of the AI model to obtain an updated AI model. For example, in federated learning, the first device trains the AI model based on the first model ID according to local data of the first device to obtain first local model parameters. The first device sends the first local model parameters to the second device. The third device trains the AI model based on the first model ID according to local data of the third device to obtain second local model parameters. The third device sends the second local model parameters to the second device. The second device fuses the first local model parameters and the second local model parameters to obtain global model parameters, and sends the global model parameters to the first device and the third device respectively. The first device and the third device respectively take the global model parameters as local model parameters of the AI model. For example, Figure 2 As shown, the first device is an AI server 206, the second device is a network device 203, and the third device is a terminal device 201. The AI server 206, the network device 203, and the terminal device 201 train the AI model 1 based on the first model ID to obtain an updated AI model 1.

[0285] Optionally, after the first device obtains the updated AI model, the first device can update the model version ID of the AI model. Thereby, the first device can update the first model ID of the AI model based on the updated model version ID.

[0286] Optionally, after obtaining the updated AI model, the second device can update the training round ID corresponding to the AI model, and send the updated training round ID to the first device. Thus, the first device can update the first model ID of the AI model based on the updated training round ID.

[0287] Optionally, Figure 7 The embodiments shown also include steps 708-709. Steps 708-709 can be performed after step 704.

[0288] 708. The second device generates a second model ID of the AI model according to the first model ID.

[0289] 709. The second device sends the second model ID to the third device. Correspondingly, the third device receives the second model ID from the second device.

[0290] Steps 708-709 are similar to the aforementioned Figure 6A Steps 608-609 in the embodiments shown are similar, and can be referred to the aforementioned Figure 6A introduction of steps 608-609 in the embodiments shown, which will not be repeated here.

[0291] It should be noted that if Figure 7 The embodiments shown include steps 705-707, and steps 705-707 and steps 708-709 have no fixed execution order. Steps 705-707 can be executed first, and then steps 708-709 can be executed. Alternatively, steps 708-709 can be executed first, and then steps 705-707 can be executed. Alternatively, steps 705-707 and steps 708-709 can be executed simultaneously according to the situation, and the specific application is not limited.

[0292] It should be noted that the above is an example of the technical solution of the present application, in which the first device sends the first model ID to the third device, and the second device sends the second model ID to the third device. In actual application, in one possible implementation, the first device can only send the first model ID to the second device, and not to the third device. That is, step 703 is an optional step. For example, in this implementation, the second device can optionally send the first model ID and the second model ID to the third device. Alternatively, the second device can optionally send the second model ID to the third device, without sending the first model ID. In this implementation, steps 705 to 707 are optional steps. Alternatively, the second device can optionally send the first model ID to the third device, and the third device generates a third model ID of the AI model according to the first model ID. In this implementation, steps 705 to 707 are optional steps. The third device generates the third model ID according to the first model ID, which is similar to the process of generating the second model ID by the second device in step 708, and will not be described here.

[0293] Optionally, Figure 7 The embodiments shown also include steps 710 to 712. Steps 710 to 712 can be performed after step 709.

[0294] 710. The first device sends fifth indication information to the second device. The fifth indication information includes the first model ID. Correspondingly, the second device receives the fifth indication information from the first device.

[0295] The fifth indication information is used to indicate that the first management operation is performed on the AI model. For the first management operation, please refer to the related description of the embodiments shown in Figure 6A The fifth indication information is similar to the second indication information in step 610 of the embodiments shown in Figure 6A The fifth indication information is similar to the second indication information in step 610 of the embodiments shown in Figure 6A The fifth indication information is similar to the second indication information in step 610 of the embodiments shown in

[0296] 711. The second device sends sixth indication information to the third device. The sixth indication information includes the second model ID. Correspondingly, the third device receives the sixth indication information from the second device.

[0297] 712. The third device performs the first management operation on the AI model based on the second model ID.

[0298] Steps 711 to 712 are similar to steps 611 to 612 of the embodiments shown in Figure 6A Steps 711 to 712 are similar to steps 611 to 612 of the embodiments shown in Figure 6A Steps 711 to 712 are similar to steps 611 to 612 of the embodiments shown in

[0299] Optionally, Figure 7 The embodiment shown also includes steps 713 to 714.

[0300] 713. The first device sends the third model ID to the third device. Correspondingly, the third device receives the third model ID from the first device.

[0301] The third model ID corresponds to the second training path, and the devices on the second training path are used to perform training operations on the AI model. The third model ID is used to identify or mark the AI model. Optionally, the first model ID corresponds to the first training path, and the devices on the first training path are used to perform training operations on the AI model. In other words, the first model ID is used by the devices on the first training path to identify or mark the AI model. The third model ID is used by the devices on the second training path to identify or mark the AI model. For example, as shown in Figure 2 The first device is the AI server 206, the second device is the network device 203, the third device is the terminal device 201, and the fifth device is the network device 204. For the terminal device 201, the terminal device 201 can identify the model 1 trained by the devices on the training path 1 through the first model ID, and identify the model 1 trained by the devices on the training path 2 through the third model ID.

[0302] It should be noted that the process of generating the third model ID by the first device is similar to the process of generating the first model ID by the first device in the aforementioned step 702. For details, please refer to the related description of the process of generating the first model ID by the first device in the aforementioned step 702, which will not be repeated here.

[0303] 714. The third device determines that the first model ID and the third model ID are both used to identify the AI model.

[0304] Step 714 is similar to the aforementioned step 614 in the embodiment shown in Figure 6A , and for details, please refer to the related description of the step 614 in the aforementioned embodiment shown in Figure 6A , which will not be repeated here.

[0305] It should be noted that steps 713 to 714 do not have a fixed order between the steps before step 713 in the embodiment shown in Figure 7 . Steps 713 to 714 can be executed first, and then the steps before step 713 in the embodiment shown in Figure 7 . Alternatively, the steps before step 713 in the embodiment shown in Figure 7 are executed first, and then steps 713 to 714 are executed. Alternatively, steps 713 to 714 and the steps before step 713 in the embodiment shown in Figure 7The steps preceding step 713 in the illustrated embodiment are not specifically limited in this application.

[0306] Optional, Figure 7 The illustrated embodiment also includes step 715.

[0307] 715. The first device sends the third model ID to the fifth device. Correspondingly, the fifth device receives the third model ID from the first device.

[0308] For details regarding the third model ID, please refer to the relevant description in step 713 above; it will not be repeated here. For example, the first device is AI server 206, and the fifth device is network device 204. AI server 206 sends the third model ID to network device 204.

[0309] It should be noted that step 715 and Figure 7 In the illustrated embodiment, there is no fixed execution order between the steps preceding step 714 (including step 714). Step 715 can be executed first, followed by... Figure 7 In the illustrated embodiment, the steps preceding step 714 (including step 714) are executed first; or, the steps preceding step 714 are executed first. Figure 7 In the illustrated embodiment, steps preceding step 714 (including step 714) are executed before step 715; or, depending on the circumstances, steps 715 and 715 are executed simultaneously. Figure 7 The steps preceding step 714 (including step 714) in the illustrated embodiments are not specifically limited in this application.

[0310] Optional, Figure 6A The illustrated embodiment also includes steps 716 to 718. Steps 716 to 718 may be performed after step 715.

[0311] 716. The first device sends a seventh instruction message to the fifth device. The seventh instruction message includes the third model ID. Accordingly, the fifth device receives the seventh instruction message from the first device.

[0312] The seventh instruction information is the same as the aforementioned Figure 6A The seventh instruction information in step 615 of the illustrated embodiment is similar; for details, please refer to the foregoing. Figure 2 The relevant description of the seventh instruction information in step 615 of the illustrated embodiment will not be repeated here.

[0313] 717. The fifth device sends a seventh instruction message to the third device. The seventh instruction message includes the third model ID. Accordingly, the third device receives the seventh instruction message from the fifth device.

[0314] 718. The third device trains the AI ​​model based on the third model ID, the fifth device, and the first device to obtain an updated AI model.

[0315] For example, as shown in Figure 7 , the first device is an AI server 206, the third device is a terminal device 201, and the fifth device is a network device 204. The terminal device 201, the network device 204, and the AI server 206 train the AI model 1 based on the third model ID to obtain an updated AI model 1.

[0316] Optionally, the first device updates the model version ID of the AI model and updates the third model ID based on the updated model version ID.

[0317] Optionally, the fifth device updates the third model ID based on the updated training round ID according to the training round ID of the AI model.

[0318] In the embodiments shown in Figure 8 , the first device obtains at least one of the region ID of the second device and the user ID of the third device and the task ID of the AI service. Then, the first device generates a first model ID of the AI model according to at least one of the region ID and the user ID and the task ID. The AI model is used to provide the AI task, and the first model ID is used to identify the AI model. Then, the first device sends the first model ID to the third device. Therefore, the first device generates the first model ID of the AI model in combination with at least one of the region ID and the user ID and the task ID. The first device sends the first model ID to the third device. Thus, the AI model is identified through the first model ID.

[0319] The first device provided by the embodiments of the present application is described below. Please refer to Figure 8 , Figure 6A for a structural schematic diagram of the first device of the embodiments of the present application. The first device 800 can be used to execute the steps performed by the first device in the embodiments shown in Figure 7 and Figure 6A . For details, please refer to the related introduction of the above method embodiments. The first device 800 includes a transceiver module 801 and a processing module 802.

[0320] The processing module 802 is used for data processing. The transceiver module 801 can realize the corresponding communication function. The transceiver module 801 can also be called a communication interface or a communication module.

[0321] Optionally, the first device 800 can also include a storage module, which can be used to store program codes, program instructions, and / or data. The processing module 802 can read the instructions and / or data in the storage module, so that the first device 800 realizes the foregoing method embodiments.

[0322] The first device 800 can be used to perform the actions performed by the first device in the above method embodiments. The first device 800 can be a network device or a component configurable on a network device; or, the first device 800 can be an AI server or a component configurable on an AI server. The processing module 802 is used to perform processing-related operations on the first device side in the above method embodiments. The transceiver module 801 is used to perform receiving-related operations on the first device side in the above method embodiments.

[0323] Optionally, the transceiver module 801 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiments. The receiving module is used to perform the receiving operation in the above method embodiments.

[0324] It should be noted that the first device 800 may include a transmitting module but not a receiving module. Alternatively, the first device 800 may include a receiving module but not a transmitting module. Specifically, it depends on whether the above-described scheme executed by the first device 800 includes both transmitting and receiving actions. For example, the first device 800 is used to execute the above... Figure 7 and Figure 6A The actions performed by the first device in the illustrated embodiment are shown above. For details, please refer to the above. Figure 7 and Figure 9 The relevant descriptions in the illustrated embodiments are not elaborated here. For example, the first device 800 is used to execute the following scheme:

[0325] The processing module 802 is used to obtain at least one of the region ID of the second device and the user ID of the third device, and the task ID of the AI ​​service; and to generate a first model ID of the AI ​​model based on at least one of the region ID and the user ID and the task ID, wherein the AI ​​model is used to provide AI services and the first model ID is used to identify the AI ​​model.

[0326] The transceiver module 801 is used to send the first model ID to the third device.

[0327] In one possible implementation, the processing module 802 is further configured to: generate a first model ID based on at least one of the region ID and user ID, the task ID, and the model version ID and / or the training round ID of the AI ​​model.

[0328] In another possible implementation, the transceiver module 801 is also used to receive the model version ID of the AI ​​model from the fourth device.

[0329] In another possible implementation, the transceiver module 801 is also used to send the first model ID to the fourth device.

[0330] In another possible implementation, the transceiving module 801 is further configured to receive a registration request from the fourth device, the registration request being used to request registration of the AI service; and send, to the fourth device, a task ID of the AI service.

[0331] In another possible implementation, the processing module 802 is further configured to select the third device according to at least one of local data information of the third device, computing power information, and communication information between the second device and the third device; and the transceiving module 801 is further configured to send, to the third device, a user ID of the third device.

[0332] In another possible implementation, the transceiving module 801 is further configured to send, to the third device, first indication information, the first indication information being used to instruct to train the AI model, and the first indication information including a first model ID; the processing module 802 is further configured to train, with the third device, the AI model to obtain an updated AI model; and the transceiving module 801 is further configured to send, to the fourth device, the updated AI model.

[0333] In another possible implementation, the processing module 802 is further configured to generate a second model ID of the AI model according to the first model ID, a length of the second model ID being less than a length of the first model ID.

[0334] In another possible implementation, the transceiving module 801 is further configured to send, to the third device, the second model ID.

[0335] In another possible implementation, the transceiving module 801 is further configured to receive second indication information from the fourth device, the second indication information being used to instruct to perform a first management operation on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, updating, fine-tuning, inference, or training, and the second indication information including the first model ID; and send, to the third device, third indication information, the third indication information being used to instruct to perform the first management operation on the AI model, and the third indication information including the second model ID.

[0336] In another possible implementation, the transceiving module 801 is further configured to send, to the second device, the second model ID.

[0337] In another possible implementation, the processing module 802 is specifically configured to receive at least one of a region ID and a user ID from the second device and a task ID.

[0338] In another possible implementation, the transceiver module 801 is further configured to: send fourth indication information to the second device and the third device, the fourth indication information being used to instruct to train the AI model, and the fourth indication information comprising the first model ID; and the processing module 802 is further configured to: train the AI model with the second device and the third device to obtain an updated AI model; and update the model version ID and / or the training round ID of the AI model based on the updated AI model.

[0339] In another possible implementation, the transceiver module 801 is further configured to: send fifth indication information to the second device, the fifth indication information being used to instruct to perform a first management operation on the AI model, the first management operation comprising at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the fifth indication information comprising the first model ID.

[0340] In another possible implementation, the first model ID corresponds to a first training path, and devices on the first training path are configured to perform a training operation on the AI model.

[0341] It should be understood that specific processes by which the modules perform the corresponding processes described above have been described in detail in the method embodiments, and thus will not be described here for brevity.

[0342] The processing module 802 in the above embodiments can be implemented by at least one processor or processor-related circuit. The transceiver module 801 can be implemented by a transceiver or transceiver-related circuit. The transceiver module 801 can also be referred to as a communication module or a communication interface. The storage module can be implemented by at least one memory.

[0343] A structural diagram of a second device according to an embodiment of the present application is shown below. Please refer to Figure 6A , the second device can be configured to perform the processes performed by the second device in the embodiments shown in Figure 7 and Figure 6A , and specific details can be referred to the related descriptions in the foregoing method embodiments. The second device 900 comprises a transceiver module 901 and a processing module 902.

[0344] The processing module 902 is configured to perform data processing. The transceiver module 901 can implement corresponding communication functions. The transceiver module 901 can also be referred to as a communication interface or a communication module.

[0345] Optionally, the second device 900 can further comprise a storage module, which can be configured to store program codes, program instructions and / or data. The processing module 902 can read the instructions and / or data in the storage module, so that the second device 900 implements the foregoing method embodiments.

[0346] The second device 900 can be configured to perform the actions of the second device in the above method embodiments. The second device 900 can be a network device or a component configurable to a network device. The processing module 902 is configured to perform the processing-related operations of the second device side in the above method embodiments. The transceiver module 901 is configured to perform the receiving-related operations of the second device side in the above method embodiments.

[0347] Optionally, the transceiver module 901 can include a sending module and a receiving module. The sending module is configured to perform the sending operations in the above method embodiments. The receiving module is configured to perform the receiving operations in the above method embodiments.

[0348] It should be noted that the second device 900 can include a sending module and not include a receiving module. Alternatively, the second device 900 can include a receiving module and not include a sending module. Specifically, whether the second device 900 includes a sending module and a receiving module can depend on whether the second device 900 performs the sending actions and the receiving actions in the above schemes. For example, the second device 900 is configured to perform the actions of the second device in the embodiments shown in the above Figure 7 and Figure 6A . For details, reference can be made to the related descriptions in the above embodiments, which will not be repeated here. For example, the second device 900 is configured to perform the following schemes: Figure 7 Figure 10

[0349] The transceiver module 901 is configured to receive a first model ID of an AI model from a first device, the first model ID being generated according to at least one of a region ID of the second device 900 and a user ID of a third device and a task ID of an AI service, the AI model being configured to provide the AI service;

[0350] The processing module 902 is configured to determine the first model ID of the AI model.

[0351] In one possible implementation, the transceiver module 901 is further configured to receive a registration request from the first device, the registration request being configured to request registration of the AI service, the AI model being an AI model corresponding to the AI service; and send, to the first device, the task ID of the AI service.

[0352] In another possible implementation, the transceiver module 901 is further configured to send, to the first device, the region ID of the second device 900.

[0353] In another possible implementation, the processing module 902 is further configured to select the third device according to at least one of local data information of the third device, computing power information, and communication information between the second device 900 and the third device; and the transceiver module 901 is further configured to send, to the first device, the user ID of the third device.

[0354] ​​In another possible implementation, the transceiver module 901 is further configured to receive fourth indication information from the first device, the fourth indication information being used to indicate that the AI model is trained, and the fourth indication information comprising the first model ID; and the processing module 902 is further configured to train the AI model with the first device and the third device to obtain an updated AI model.

[0355] In another possible implementation, the processing module 902 is further configured to generate a second model ID of the AI model according to the first model ID, the length of the second model ID being less than the length of the first model ID.

[0356] In another possible implementation, the transceiver module 901 is further configured to send the second model ID to the third device.

[0357] In another possible implementation, the transceiver module 901 is further configured to receive fifth indication information from the first device, the fifth indication information being used to indicate that the first management operation is performed on the AI model, the first management operation comprising at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the fifth indication information comprising the first model ID; and the transceiver module 901 is further configured to send sixth indication information to the third device, the sixth indication information being used to indicate that the first management operation is performed on the AI model, and the sixth indication information comprising the second model ID.

[0358] It should be understood that the specific processes by which the modules perform the corresponding processes described above have been described in detail in the method embodiments described above, and thus will not be described here for brevity.

[0359] The processing module 902 in the above embodiments can be implemented by at least one processor or processor-related circuit. The transceiver module 901 can be implemented by a transceiver or transceiver-related circuit. The transceiver module 901 can also be referred to as a communication module or a communication interface. The storage module can be implemented by at least one memory.

[0360] A structural diagram of a third device according to an embodiment of the present application is shown below. Please refer to Figure 6A , the third device can be configured to perform the processes performed by the third device in the embodiments shown in Figure 7 and Figure 6A , and specific details can be referred to in the foregoing method embodiments. The third device 1000 comprises a transceiver module 1001 and a processing module 1002.

[0361] The processing module 1002 is configured to perform data processing. The transceiver module 1001 can implement corresponding communication functions. The transceiver module 1001 can also be referred to as a communication interface or a communication module.

[0362] Optionally, the third device 1000 may further include a storage module, which can be used to store program code, program instructions and / or data. The processing module 1002 can read the instructions and / or data in the storage module so that the third device 1000 can implement the aforementioned method embodiments.

[0363] The third device 1000 can be used to perform the actions performed by the third device in the above method embodiments. The third device 1000 can be a terminal device or a component configurable on a terminal device. The processing module 1002 is used to perform processing-related operations on the third device side in the above method embodiments. The transceiver module 1001 is used to perform receiving-related operations on the third device side in the above method embodiments.

[0364] Optionally, the transceiver module 1001 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiments. The receiving module is used to perform the receiving operation in the above method embodiments.

[0365] It should be noted that the third device 1000 may include a transmitting module but not a receiving module. Alternatively, the third device 1000 may include a receiving module but not a transmitting module. Specifically, it depends on whether the above-described scheme executed by the third device 1000 includes both transmitting and receiving actions. For example, the third device 1000 is used to execute the above... Figure 7 and Figure 6A The actions performed by the third device in the illustrated embodiment. For details, please refer to the above. Figure 7 and Figure 11 The relevant descriptions in the illustrated embodiments are not elaborated here. For example, the third device 1000 is used to execute the following scheme:

[0366] The transceiver module 1001 is used to receive a first model ID from the AI ​​model of the first device; the first model ID is generated based on at least one of the region ID of the second device and the user ID of the third device 1000 and the task ID of the AI ​​service, and the AI ​​model is used to provide AI services.

[0367] Processing module 1002 is used to determine the first model ID of the AI ​​model.

[0368] In one possible implementation, the transceiver module 1001 is further configured to: receive first instruction information from the first device, the first instruction information being used to instruct the training of the AI ​​model, the first instruction information including a first model ID; the processing module is further configured to: train the AI ​​model based on the first model ID and the second device to obtain an updated AI model.

[0369] In another possible implementation, the transceiving module 1001 is further configured to receive the second model ID of the AI model from the first device, the second model ID having a length smaller than that of the first model ID; or the processing module 1002 is further configured to generate the second model ID of the AI model according to the first model ID, the second model ID having a length smaller than that of the first model ID.

[0370] In another possible implementation, the transceiving module 1001 is further configured to receive third indication information from the first device, the third indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the third indication information including the second model ID; and the processing module 1002 is further configured to perform the first management operation on the AI model based on the second model ID.

[0371] In another possible implementation, the transceiving module 1001 is further configured to receive sixth indication information from the second device, the sixth indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the sixth indication information including the second model ID; and the processing module 1002 is further configured to perform the first management operation on the AI model based on the second model ID.

[0372] In another possible implementation, the first model ID corresponds to a first training path, and devices on the first training path are configured to perform corresponding training operations on the AI model; the transceiving module 1001 is further configured to receive a third model ID of the AI model from the fifth device or the first device, the third model ID corresponding to a second training path, and devices on the second training path are configured to perform training operations on the AI model; and the processing module 1002 is further configured to determine that the first model ID and the third model ID are both used to identify the AI model.

[0373] In another possible implementation, the transceiving module 1001 is further configured to receive seventh indication information from the fifth device, the seventh indication information being used to indicate that the AI model is trained, and the seventh indication information including the third model ID; and the processing module 1002 is further configured to train the AI model based on the third model ID and the fifth device; or the processing module 1002 is further configured to train the AI model based on the third model ID and the fifth device and the first device.

[0374] It should be understood that the specific processes by which the modules perform the corresponding processes described above have been described in detail in the method embodiments described above, and thus are not described herein again for the sake of brevity.

[0375] The processing module 1002 in the above embodiments can be implemented by at least one processor or processor-related circuit. The transceiver module 1001 can be implemented by a transceiver or transceiver-related circuit. The transceiver module 1001 can also be referred to as a communication module or a communication interface. The storage module can be implemented by at least one memory.

[0376] A structural schematic diagram of a fourth device in an embodiment of the present application is shown below. Please refer to Figure 6A , the fourth device can be used to execute the processes performed by the fourth device in the embodiments shown in Figure 7 and Figure 6A . For details, please refer to the related descriptions in the foregoing method embodiments. The fourth device 1100 includes a transceiver module 1101 and a processing module 1102.

[0377] The processing module 1102 is configured to perform data processing. The transceiver module 1101 can implement corresponding communication functions. The transceiver module 1101 can also be referred to as a communication interface or a communication module.

[0378] Optionally, the fourth device 1100 can further include a storage module, which can be configured to store program codes, program instructions and / or data. The processing module 1102 can read the instructions and / or data in the storage module, so that the fourth device 1100 implements the foregoing method embodiments.

[0379] The fourth device 1100 can be used to execute the actions performed by the fourth device in the foregoing method embodiments. The fourth device 1100 can be an AI server or a component configurable to an AI server. The processing module 1102 is configured to perform processing-related operations on the fourth device side in the foregoing method embodiments. The transceiver module 1101 is configured to perform receiving-related operations on the fourth device side in the foregoing method embodiments.

[0380] Optionally, the transceiver module 1101 can include a sending module and a receiving module. The sending module is configured to perform the sending operations in the foregoing method embodiments. The receiving module is configured to perform the receiving operations in the foregoing method embodiments.

[0381] It should be noted that the fourth device 1100 can include a sending module but not a receiving module. Alternatively, the fourth device 1100 can include a receiving module but not a sending module. Specifically, whether the fourth device 1100 includes a sending module and a receiving module can depend on whether the fourth device 1100 performs the sending actions and the receiving actions in the foregoing schemes. For example, the fourth device 1100 is configured to perform the actions performed by the fourth device in the embodiments shown in Figure 7 and Figure 6A . For details, please refer to the related descriptions in the foregoing embodiments Figure 7 and Figure 12 . Here, no detailed description is given. For example, the fourth device 1100 is configured to perform the following scheme:

[0382] The transceiver module 1101 is configured to receive a first model ID of an AI model from the first device, the first model ID being generated according to at least one of a region ID of the second device and a user ID of the third device and a task ID of the AI service, the AI model being used to provide the AI service.

[0383] The processing module 1102 is configured to determine the first model ID of the AI model.

[0384] In a possible implementation, the first model ID is further generated according to a model version ID of the AI model. The transceiver module 1101 is further configured to send the model version ID of the AI model to the first device.

[0385] In another possible implementation, the transceiver module 1101 is further configured to send, to the first device, second indication information, the second indication information being used to indicate that the first management operation is performed on the AI model, the first management operation including at least one of the following: activation, deactivation, switching, selection, fallback, update, fine-tuning, inference, or training, and the second indication information including the first model ID.

[0386] It should be understood that the specific processes in which the modules perform the corresponding processes described above have been described in detail in the method embodiments described above, and thus will not be described here again for the sake of brevity.

[0387] The processing module 1102 in the above embodiments can be implemented by at least one processor or processor-related circuit. The transceiver module 1101 can be implemented by a transceiver or transceiver-related circuit. The transceiver module 1101 can also be referred to as a communication module or a communication interface. The storage module can be implemented by at least one memory.

[0388] The embodiments of the present application also provide a device 1200. Please refer to Figure 12 The device 1200 includes a processor 1210 and a memory 1220. The memory 1220 is configured to store computer programs or instructions and / or data. The processor 1210 is configured to execute the computer programs or instructions and / or data stored in the memory 1220, so that the methods in the above method embodiments are performed. The device 1200 is configured to implement the operations performed by the first device, the second device, the third device, or the fourth device in the above method embodiments.

[0389] Optionally, the processor 1210 included in the device 1200 is one or more.

[0390] Optionally, as shown in Figure 12 The device 1200 can further include the memory 1220.

[0391] Optionally, the memory 1220 included in the device 1200 can be one or more.

[0392] Optionally, the memory 1220 can be integrated with the processor 1210 or be separately arranged.

[0393] Optionally, as shown in FIG. 12B, the apparatus 1200 can further include a transceiver 1230 for receiving and / or sending signals. For example, the processor 1210 is configured to control the transceiver 1230 to receive and / or send signals. Figure 13

[0394] The present application further provides an apparatus 1300, which can be a terminal device, a processor in a terminal device, or a chip. The apparatus 1300 can be configured to perform operations performed by the third apparatus in the above method embodiments.

[0395] When the apparatus 1300 is a terminal device, Figure 13 FIG. 13A shows a simplified structural schematic diagram of a terminal device. As shown in FIG. 13A, the terminal device includes a processor, a memory, and a transceiver. The memory can store computer program codes, and the transceiver includes a transmitter 1331, a receiver 1332, a radio frequency circuit (not shown in the figure), an antenna 1333, and an input / output device (not shown in the figure). Figure 13

[0396] The processor is mainly configured to process communication protocols and communication data, control the terminal device, execute software programs and process data of the software programs, and the like.

[0397] The memory is mainly configured to store software programs and data.

[0398] The radio frequency circuit is mainly configured to convert baseband signals and radio frequency signals and process radio frequency signals.

[0399] The antenna is mainly configured to transceive radio frequency signals in the form of electromagnetic waves.

[0400] The input / output device can include a touch screen, a display screen, a keyboard, or the like. The input / output device is mainly configured to receive data input by a user and output data to the user. It should be noted that some types of terminal devices can not have an input / output device.

[0401] When data needs to be sent, the processor performs baseband processing on the data to be sent, and outputs the baseband signal to the radio frequency circuit. Then, the radio frequency circuit performs radio frequency processing on the baseband signal, and transmits the radio frequency signal in the form of electromagnetic waves through the antenna. When data is sent to the terminal device, the radio frequency circuit receives the radio frequency signal through the antenna. The radio frequency circuit converts the radio frequency signal into a baseband signal, and outputs the baseband signal to the processor. The processor converts the baseband signal into data and processes the data. For the sake of brevity, Figure 13 ​​Only one memory, one processor and one transceiver are shown in the terminal device. In actual products of terminal devices, one or more processors and one or more memories can exist. The memory can also be referred to as a storage medium or a storage device, etc. The memory can be arranged independently of the processor or integrated with the processor, and the embodiments of the present application do not limit this.

[0402] In the embodiments of the present application, the antenna and the radio frequency circuit with transceiving functions can be regarded as a transceiving module of the terminal device, and the processor with processing functions can be regarded as a processing module of the terminal device.

[0403] As shown in Figure 6A , the terminal device includes a processor 1310, a memory 1320 and a transceiver 1330. The processor 1310 can also be referred to as a processing unit, a processing board, a processing module, or a processing device, etc. The transceiver 1330 can also be referred to as a transceiving unit, a transceiver, or a transceiving device, etc.

[0404] Optionally, the devices for implementing the receiving function in the transceiver 1330 are regarded as a receiving module, and the devices for implementing the sending function in the transceiver 1330 are regarded as a sending module, that is, the transceiver 1330 includes a receiver and a transmitter. The transceiver can also be referred to as a transceiver, a transceiving module, or a transceiving circuit, etc. The receiver can also be referred to as a receiver, a receiving module, or a receiving circuit, etc. The transmitter can also be referred to as a transmitter, a transmitting module, or a transmitting circuit, etc.

[0405] The processor 1310 is configured to perform the processing actions of the third device side in the embodiments shown in Figure 7 and Figure 6A . The transceiver 1330 is configured to perform the transceiving actions of the third device side in the embodiments shown in Figure 7 and Figure 13 .

[0406] It should be understood that Figure 10 the above-mentioned terminal device including a transceiving module and a processing module can not depend on the structure shown in Figure 13 or Figure 6A .

[0407] When the device 1300 is a chip, the chip includes a processor, a memory and a transceiver. The transceiver can be an input / output circuit or a communication interface. The processor can be a processing module integrated on the chip or a microprocessor or an integrated circuit. The sending operation of the first device in the above-mentioned method embodiments can be understood as the output of the chip, and the receiving operation of the first device in the above-mentioned method embodiments can be understood as the input of the chip.

[0408] The application further provides a device 1400, which can be a network device or a chip. The device 1400 can be used to perform the operations of the first device or the second device in the above-mentioned embodiments. Figure 7 and Figure 14 The device 1400 can be used to perform the operations of the first device or the second device in the above-mentioned embodiments.

[0409] When the device 1400 is a network device, for example, a base station. Figure 6A A simplified structure diagram of a base station is shown. The base station includes a 1410 part, a 1420 part, and a 1430 part.

[0410] The 1410 part is mainly used for baseband processing, controlling the base station, etc. The 1410 part is usually the control center of the base station, which can be referred to as a processor, and is used to control the base station to perform the processing operations of the first device or the second device in the above-mentioned method embodiments.

[0411] The 1420 part is mainly used for storing computer program codes and data.

[0412] The 1430 part is mainly used for transceiving radio frequency signals and converting radio frequency signals and baseband signals. The 1430 part can be referred to as a transceiving module, a transceiver, a transceiving circuit, or a transceiver, etc. The transceiving module of the 1430 part, which can also be referred to as a transceiver or a transceiver, includes an antenna 1433 and a radio frequency circuit (not shown in the figure), wherein the radio frequency circuit is mainly used for radio frequency processing. Optionally, the devices in the 1430 part used to realize the receiving function can be regarded as a receiver, and the devices used to realize the sending function can be regarded as a transmitter, that is, the 1430 part includes a receiver 1432 and a transmitter 1431. The receiver can also be referred to as a receiving module, a receiver, or a receiving circuit, etc. The transmitter can be referred to as a transmitting module, a transmitter, or a transmitting circuit, etc.

[0413] The 1410 part and the 1420 part can include one or more single boards, and each single board can include one or more processors and one or more memories. The processor is used to read and execute the program in the memory to realize the baseband processing function and control the base station. If there are multiple single boards, the single boards can be interconnected to enhance the processing capability. As an optional implementation, the multiple single boards can share one or more processors, or the multiple single boards can share one or more memories, or the multiple single boards can share one or more processors at the same time.

[0414] For example, in an implementation, the transceiving module of the 1430 part is used to perform the transceiving-related processes performed by the first device or the second device in the embodiments shown in Figure 7 and Figure 6A The processor of the 1410 part is used to perform the processes performed by the first device or the second device in the embodiments shown in Figure 7 and Figure 14The process related to the processing performed by the first device or the second device in the illustrated embodiment.

[0415] It should be understood that Figure 8 For example, but not limited to, the network device including the processor, the memory and the transceiver can not depend on Figure 9 、 Figure 14 or Figure 6A the structure illustrated.

[0416] When the device 1400 is a chip, the chip includes a transceiver, a memory and a processor. The transceiver can be an input and output circuit, a communication interface; the processor is a processor integrated on the chip, or a microprocessor, or an integrated circuit. The sending operation of the first device or the second device in the above method embodiment can be understood as the output of the chip, and the receiving operation of the first device or the second device in the above method embodiment can be understood as the input of the chip.

[0417] The present application also provides a computer readable storage medium having stored thereon computer instructions for implementing the method performed by the first device, the second device, the third device or the fourth device in the above method embodiment.

[0418] For example, the computer program is executed by a computer, so that the computer can implement the method performed by the first device, the second device, the third device or the fourth device in the above method embodiment.

[0419] The present application also provides a computer program product containing instructions, which are executed by a computer to make the computer implement the method performed by the first device, the second device, the third device or the fourth device in the above method embodiment.

[0420] The present application also provides a communication system, which includes a first device and a third device, the first device is used to perform part or all of the operations performed by the first device in the embodiments as Figure 7 and Figure 6A The third device is used to perform part or all of the operations performed by the third device in the embodiments as Figure 7 and Figure 6A .

[0421] Optionally, the communication system further includes a second device, the second device is used to perform part or all of the steps performed by the second device in the embodiments as Figure 7 .

[0422] Optionally, the communication system further includes a fourth device, the fourth device is used to perform part or all of the steps performed by the fourth device in the embodiments as Figure 6A .

[0423] Optionally, the communication system further includes a fifth device, the fifth device is used to perform part or all of the steps performed by the fifth device in the embodiments asFigure 7 and Figure 6A some or all of the steps performed by the fifth apparatus in the embodiments shown in

[0424] The embodiments of the present application also provide a chip apparatus, comprising a processor, configured to invoke computer programs or computer instructions stored in the memory, so that the processor executes the method provided in any one of the embodiments shown in Figure 7 and Figure 6A the embodiments shown in

[0425] In a possible implementation, the input of the chip apparatus corresponds to the receiving operation in any one of the embodiments shown in Figure 7 and Figure 6A the embodiments shown in Figure 7 and Figure 6A the embodiments shown in

[0426] Optionally, the processor is coupled with the memory through an interface.

[0427] Optionally, the chip apparatus further comprises a memory, and the memory stores computer programs or computer instructions.

[0428] The processor mentioned in any one of the above embodiments can be a general central processing unit, a microprocessor, an application-specific integrated circuit (ASIC), or an integrated circuit for program execution of the method provided in any one of the embodiments shown in Figure 7 and ​ The memory mentioned in any one of the above embodiments can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM), etc.

[0429] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the explanations and beneficial effects of the related contents in any one of the above provided apparatuses can refer to the corresponding method embodiments provided above, and will not be repeated here.

[0430] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. The division of the units is merely logical function division. There can be other division manners in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0431] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0432] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0433] When the integrated unit is realized in the form of 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, the technical solutions of the present application essentially make contributions to the part or the whole or part of the technical solutions. The computer software product can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk, and various program codes that can be stored in the storage medium.

[0434] The above embodiments are merely used to describe the technical solutions of the present application, rather than limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalent replacements; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A model recognition method, characterized in that, The method includes: The first device obtains at least one of the area identifier ID of the second device and the user ID of the third device, as well as the task ID of the artificial intelligence (AI) service; The first device generates a first model ID for the AI ​​model based on at least one of the region ID and the user ID and the task ID. The AI ​​model is used to provide the AI ​​service, and the first model ID is used to identify the AI ​​model. The first device sends the first model ID to the third device.

2. The method according to claim 1, characterized in that, The first device generates a first model ID for the AI ​​model based on at least one of the region ID and the user ID, and the task ID, including: The first device generates the first model ID based on at least one of the region ID and the user ID, the task ID, and the model version ID and / or the training round ID of the AI ​​model.

3. The method according to claim 2, characterized in that, Before the first device generates the first model ID based on at least one of the region ID and the user ID, the task ID, and the model version ID and / or the training epoch ID of the AI ​​model, the method further includes: The first device receives the model version ID of the AI ​​model from the fourth device.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: The first device sends the first model ID to the fourth device.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: The first device selects the third device based on at least one of the local data information of the third device, computing power information, and communication information between the second device and the third device; The first device sends the user ID of the third device to the third device.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: The first device sends a first instruction message to the third device, the first instruction message being used to instruct the AI ​​model to be trained, the first instruction message including the first model ID; The first device and the third device train the AI ​​model to obtain an updated AI model; The first device sends the updated AI model to the fourth device.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: The first device generates a second model ID for the AI ​​model based on the first model ID, wherein the length of the second model ID is less than the length of the first model ID.

8. The method according to claim 7, characterized in that, The method further includes: The first device sends the second model ID to the third device.

9. The method according to claim 7 or 8, characterized in that, The method further includes: The first device receives second instruction information from the fourth device. The second instruction information is used to instruct the AI ​​model to perform a first management operation. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training. The second instruction information includes the first model ID. The first device sends a third instruction message to the third device, the third instruction message being used to instruct the AI ​​model to perform a first management operation, the third instruction message including the second model ID.

10. The method according to claim 1 or 2, characterized in that, The method further includes: The first device sends the first model ID to the second device.

11. The method according to claim 1, 2, or 10, characterized in that, The first device acquires at least one of the region identifier ID of the second device and the user ID of the third device, as well as the task ID of the artificial intelligence (AI) service, including: The first device receives at least one of the region ID and the user ID from the second device, as well as the task ID.

12. The method according to claim 1, 2, 10 or 11, characterized in that, The method further includes: The first device sends a fourth instruction message to the second device and the third device respectively. The fourth instruction message is used to instruct the AI ​​model to be trained. The fourth instruction message includes the first model ID. The first device, the second device, and the third device train the AI ​​model to obtain an updated AI model; The first device updates the model version ID of the AI ​​model.

13. The method according to any one of claims 1, 2, 10 to 12, characterized in that, The method further includes: The first device sends a fifth instruction message to the second device. The fifth instruction message is used to instruct the AI ​​model to perform a first management operation. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training. The fifth instruction message includes the first model ID.

14. The method according to any one of claims 1 to 13, characterized in that, The first model ID corresponds to the first training path, and the device on the first training path is used to perform training operations on the AI ​​model.

15. A model recognition method, characterized in that, The method includes: The second device receives a first model identifier ID from the first device's artificial intelligence (AI) model. The first model ID is generated based on at least one of the second device's region ID and the third device's user ID, as well as the AI ​​service's task ID. The AI ​​model is used to provide the AI ​​service. The second device determines the first model ID of the AI ​​model.

16. The method according to claim 15, characterized in that, Before the second device receives the first model identifier ID from the artificial intelligence (AI) model of the first device, the method further includes: The second device sends its region ID to the first device.

17. The method according to claim 15 or 16, characterized in that, Before the second device receives the first model identifier ID from the artificial intelligence (AI) model of the first device, the method further includes: The second device selects the third device based on at least one of the following: local data information of the third device, computing power information, and communication information between the second device and the third device; The second device sends the user ID of the third device to the first device.

18. The method according to any one of claims 15 to 17, characterized in that, The method further includes: The second device receives a fourth instruction from the first device, the fourth instruction being used to instruct the AI ​​model to be trained, the fourth instruction including the first model ID; The second device trains the AI ​​model based on the first model ID, the first device, and the third device to obtain an updated AI model.

19. The method according to any one of claims 15 to 18, characterized in that, The method further includes: The second device generates a second model ID for the AI ​​model based on the first model ID, wherein the length of the second model ID is less than the length of the first model ID.

20. The method according to claim 19, characterized in that, The method further includes: The second device sends the second model ID to the third device.

21. The method according to claim 19 or 20, characterized in that, The method further includes: The second device receives a fifth instruction from the first device, the fifth instruction being used to instruct the AI ​​model to perform a first management operation, the first management operation including at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training, the fifth instruction including the first model ID; The second device sends a sixth instruction message to the third device. The sixth instruction message is used to instruct the AI ​​model to perform a first management operation. The sixth instruction message includes the second model ID.

22. A model recognition method, characterized in that, The method includes: The third device receives a first model identifier ID from the first device's artificial intelligence (AI) model. The first model ID is generated based on at least one of the region ID of the second device and the user ID of the third device, as well as the task ID of the AI ​​service. The AI ​​model is used to provide the AI ​​service. The third device determines the first model ID of the AI ​​model.

23. The method according to claim 22, characterized in that, The method further includes: The third device receives first instruction information from the first device, the first instruction information being used to instruct the AI ​​model to be trained, the first instruction information including the first model ID; The third device trains the AI ​​model based on the first model ID and the second device to obtain an updated AI model.

24. The method according to claim 22 or 23, characterized in that, The method further includes: The third device receives a second model ID from the AI ​​model of the first device, wherein the length of the second model ID is less than the length of the first model ID; or... The third device generates a second model ID for the AI ​​model based on the first model ID, wherein the length of the second model ID is less than the length of the first model ID.

25. The method according to claim 24, characterized in that, The method further includes: The third device receives third instruction information from the first device. The third instruction information is used to instruct the AI ​​model to perform a first management operation. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training. The third instruction information includes the second model ID. The third device performs the first management operation on the AI ​​model based on the second model ID.

26. The method according to claim 24, characterized in that, The method further includes: The third device receives a sixth instruction information from the second device. The sixth instruction information is used to instruct the AI ​​model to perform a first management operation. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training. The sixth instruction information includes the second model ID. The third device performs the first management operation on the AI ​​model based on the second model ID.

27. The method according to any one of claims 22 to 26, characterized in that, The first model ID corresponds to the first training path, and the device on the first training path is used to perform corresponding training operations on the AI ​​model. The method further includes: The third device receives a third model ID of the AI ​​model from the fifth device or the first device. The third model ID corresponds to a second training path, and the device on the second training path is used to perform training operations on the AI ​​model. The third device determines that both the first model ID and the third model ID are used to identify the AI ​​model.

28. The method according to claim 27, characterized in that, The method further includes: The third device receives a seventh instruction information from the fifth device, the seventh instruction information being used to instruct the AI ​​model to be trained, the seventh instruction information including the third model ID; The third device trains the AI ​​model based on the third model ID and the fifth device; or, the third device trains the AI ​​model based on the third model ID, the fifth device, and the first device.

29. A model recognition method, characterized in that, The method includes: The fourth device receives a first model identifier ID from the first device's artificial intelligence (AI) model. The first model ID is generated based on at least one of the region ID of the second device and the user ID of the third device, as well as the task ID of the AI ​​service. The AI ​​model is used to provide the AI ​​service. The fourth device determines the first model ID of the AI ​​model.

30. The method according to claim 29, characterized in that, Before the fourth device receives the first model ID from the artificial intelligence (AI) model of the first device, the method further includes: The fourth device sends the model version ID of the AI ​​model to the first device.

31. The method according to claim 29 or 30, characterized in that, The method further includes: The fourth device sends a second instruction to the first device. The second instruction is used to instruct the AI ​​model to perform a first management operation. The first management operation includes at least one of the following: activation, deactivation, switching, selection, rollback, update, fine-tuning, inference, or training. The second instruction includes the first model ID.

32. An apparatus, characterized in that, The device includes a transceiver module and a processing module; The transceiver module is configured to perform the transceiver operation of the method as described in any one of claims 1 to 14, and the processing module is configured to perform the processing operation of the method as described in any one of claims 1 to 14; or, The transceiver module is configured to perform the transceiver operation of the method as described in any one of claims 15 to 21, and the processing module is configured to perform the processing operation of the method as described in any one of claims 15 to 21; or, The transceiver module is configured to perform the transceiver operation of the method as described in any one of claims 22 to 28, and the processing module is configured to perform the processing operation of the method as described in any one of claims 22 to 28; or, The transceiver module is used to perform the transceiver operation of the method as described in any one of claims 29 to 31, and the processing module is used to perform the processing operation of the method as described in any one of claims 29 to 31.

33. An apparatus, characterized in that, The apparatus includes a processor; the processor is configured to execute a computer program or computer instructions stored in a memory to perform the method as described in any one of claims 1 to 14; or... The processor is configured to execute a computer program or computer instructions stored in memory to perform the method as described in any one of claims 15 to 21; or... The processor is configured to execute a computer program or computer instructions stored in memory to perform the method as described in any one of claims 22 to 28; or... The processor is configured to execute computer programs or computer instructions in memory to perform the method as described in any one of claims 29 to 31.

34. The apparatus according to claim 33, characterized in that, The device also includes the memory.

35. A computer-readable storage medium, characterized in that, It stores a computer program thereon, which, when executed by the device, causes the device to perform the method as described in any one of claims 1 to 31.

36. A computer program product, characterized in that, When the computer program product is run on a computer, it causes the computer to perform the method as described in any one of claims 1 to 31.

Images