Data collection method and device, terminal and network side equipment

Abstract

The invention discloses a data collection method and device, a terminal and network side equipment, and belongs to the technical field of communication, and the data collection method comprises the steps that first equipment sends first information to second equipment, the first information is used for indicating that target information of a second AI unit is obtained through reasoning based on a first AI unit, the target information comprises a virtual truth value or a virtual evaluation result; wherein the target information is used for determining a training data set of a second AI unit, the second AI unit is used for realizing a use case function, the first equipment is used for realizing a reasoning function of the second AI unit, and the second equipment is used for realizing a reasoning function of the first AI unit.

Description

Technical Field

[0001] This application belongs to the field of communication technology, specifically relating to a data collection method, apparatus, terminal, and network-side equipment. Background Technology

[0002] In mobile communication systems, an increasing number of use cases incorporate artificial intelligence (AI). Examples at the physical layer include AI-based channel state information (CSI) feedback compression, AI-based beam management, AI-based positioning, AI-based energy saving, and AI-based load balancing. However, collecting training data for AI use cases at the physical layer typically requires obtaining ground truth values; however, obtaining ground truth values ​​usually incurs significant measurement overhead. Therefore, existing technologies suffer from the problem of high overhead in collecting training data for AI use cases. Summary of the Invention

[0003] This application provides a data collection method, apparatus, terminal, and network-side device that can solve the problem of high overhead in collecting training data for AI use cases.

[0004] Firstly, a data collection method is provided, including:

[0005] The first device sends a first message to the second device, the first message being used to instruct the second AI unit to obtain target information based on the reasoning of the first AI unit, the target information including virtual truth value or virtual evaluation result;

[0006] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0007] Secondly, a data collection method is provided, including:

[0008] The second device receives the first information from the first device;

[0009] The second device obtains the target information of the second AI unit based on the first information and the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0010] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0011] Thirdly, a data collection method is provided, including:

[0012] The third device performs the target operation, which includes any one of the following:

[0013] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0014] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0015] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0016] Fourthly, a data collection method is provided, including:

[0017] The fourth device receives the ninth information from the third device. The ninth information includes the training dataset of the second AI unit, which is determined based on the first AI unit and the first information sent by the first device to the second device.

[0018] The fourth device filters the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset;

[0019] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0020] Fifthly, a data collection device is provided, applied to a first device, comprising:

[0021] A first transmission module is used to send first information to a second device. The first information is used to instruct the second AI unit to obtain target information based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0022] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0023] Sixthly, a data collection device is provided for use in a second device, comprising:

[0024] The second transmission module is used to receive first information from the first device;

[0025] The first processing module is used to obtain the target information of the second AI unit based on the first information and the reasoning of the first AI unit, wherein the target information includes virtual truth value or virtual evaluation result;

[0026] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0027] In a seventh aspect, a data collection device is provided for use in a third device, comprising:

[0028] The third transmission module is used to perform the target operation, which includes any one of the following:

[0029] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0030] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0031] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0032] Eighthly, a data collection device is provided for use in a fourth device, characterized in that it comprises:

[0033] The fourth transmission module is used to receive the ninth information from the third device, the ninth information including the training dataset of the second AI unit, the training dataset being determined based on the first information sent from the first AI unit and the first device to the second device;

[0034] The second processing module is used to filter the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset.

[0035] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0036] In a ninth aspect, a data collection apparatus is provided, the apparatus being configured to perform the steps of the method described in the first aspect, or to implement the steps of the method described in the second aspect, or to implement the steps of the method described in the third aspect, or to implement the steps of the method described in the fourth aspect.

[0037] In a tenth aspect, a terminal is provided, the terminal including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the first aspect, or the steps of the method as described in the second aspect, or the steps of the method as described in the third aspect, or the steps of the method as described in the fourth aspect.

[0038] Eleventhly, a terminal is provided, including a processor and a communication interface, wherein,

[0039] When the terminal is a first device, the communication interface is used to send first information to a second device. The first information is used to instruct the second AI unit to obtain target information of the second AI unit based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0040] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0041] When the terminal is a second device, the communication interface is used to receive first information from the first device;

[0042] A processor is configured to obtain target information of the second AI unit based on the first information and reasoning of the first AI unit, wherein the target information includes virtual truth value or virtual evaluation result;

[0043] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0044] When the terminal is a third device, the communication interface is used to perform a target operation, which includes any one of the following:

[0045] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0046] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0047] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0048] In the case that the terminal is a fourth device, the communication interface receives a ninth message from the third device. The ninth message includes a training dataset of the second AI unit, which is determined based on the first AI unit and the first message sent from the first device to the second device.

[0049] The processor is used to filter the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset;

[0050] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0051] In a twelfth aspect, a network-side device is provided, comprising a processor and a memory, the memory storing a program or instructions executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the method as described in the first aspect, or the steps of the method as described in the second aspect, or the steps of the method as described in the third aspect, or the steps of the method as described in the fourth aspect.

[0052] In a thirteenth aspect, a network-side device is provided, including a processor and a communication interface, wherein...

[0053] When the network-side device is the first device, the communication interface is used to send first information to the second device. The first information is used to instruct the second AI unit to obtain target information based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0054] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0055] When the network-side device is a second device, the communication interface is used to receive first information from the first device;

[0056] A processor is configured to obtain target information of the second AI unit based on the first information and reasoning of the first AI unit, wherein the target information includes virtual truth value or virtual evaluation result;

[0057] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0058] When the network-side device is a third device, the communication interface is used to perform a target operation, which includes any of the following:

[0059] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0060] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0061] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0062] When the network-side device is a fourth device, the communication interface receives ninth information from the third device. The ninth information includes the training dataset of the second AI unit, which is determined based on the first AI unit and the first information sent by the first device to the second device.

[0063] The processor is used to filter the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset;

[0064] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0065] In a fourteenth aspect, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect, or the steps of the method described in the second aspect, or the steps of the method described in the third aspect, or the steps of the method described in the fourth aspect.

[0066] In a fifteenth aspect, a wireless communication system is provided, comprising: a first device, a second device, a third device, or a fourth device, wherein the first device is configured to perform the steps of the method as described in the first aspect, the second device is configured to perform the steps of the method as described in the second aspect, the third device is configured to perform the steps of the method as described in the third aspect, and the fourth device is configured to perform the steps of the method as described in the fourth aspect.

[0067] In a sixteenth aspect, a chip is provided, the chip including a processor and a communication interface coupled to the processor, the processor being configured to run a program or instructions to implement the method as described in the first aspect, or the method as described in the second aspect, or the steps of the method as described in the third aspect, or the steps of the method as described in the fourth aspect.

[0068] In a seventeenth aspect, a computer program / program product is provided, the computer program / program product being stored in a storage medium, the computer program / program product being executed by at least one processor to implement the steps of the method as described in the first aspect, or the steps of the method as described in the second aspect, or the steps of the method as described in the third aspect, or the steps of the method as described in the fourth aspect.

[0069] This application embodiment sends first information to a second device via a first device, thereby obtaining target information for the second AI unit based on the first information and inference from the first AI unit. The target information includes virtual truth values ​​or virtual evaluation results. The target information is used to determine the training dataset for the second AI unit, which implements use case functions. The first device implements the inference function of the second AI unit, and the second device implements the inference function of the first AI unit. Thus, since the training dataset for the second AI unit is determined using the virtual truth values ​​or virtual evaluation results obtained through inference from the first AI unit, there is no need to obtain the truth values, avoiding the problems of unobtainable truth values ​​and the high overhead associated with obtaining them. Attached Figure Description

[0070] Figure 1 This is a block diagram of a wireless communication system applicable to embodiments of this application;

[0071] Figure 2 This is one of the flowcharts illustrating a data collection method provided in an embodiment of this application;

[0072] Figures 2a to 2c These are example diagrams illustrating solutions applicable to embodiments of this application;

[0073] Figure 3 This is a second schematic flowchart of a data collection method provided in an embodiment of this application;

[0074] Figures 3a to 3h yes Figure 3 A detailed flowchart illustrating the intermediate steps;

[0075] Figures 4a to 4h This is a flowchart illustrating a data collection method provided in an embodiment of this application;

[0076] Figure 5This is a third schematic flowchart of a data collection method provided in an embodiment of this application;

[0077] Figure 6 This is a fourth flowchart illustrating a data collection method provided in an embodiment of this application;

[0078] Figure 7 This is the fifth flowchart illustrating a data collection method provided in an embodiment of this application;

[0079] Figure 8 This is one of the structural schematic diagrams of a data collection device provided in the embodiments of this application;

[0080] Figure 9 This is a second schematic diagram of the structure of a data collection device provided in an embodiment of this application;

[0081] Figure 10 This is the third schematic diagram of the structure of a data collection device provided in the embodiments of this application;

[0082] Figure 11 This is the fourth schematic diagram of a data collection device provided in the embodiments of this application;

[0083] Figure 12 This is a schematic diagram of the structure of a communication device provided in an embodiment of this application;

[0084] Figure 13 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application;

[0085] Figure 14 This is a schematic diagram of the structure of a network-side device provided in an embodiment of this application;

[0086] Figure 15 This is a schematic diagram of another network-side device provided in an embodiment of this application. Detailed Implementation

[0087] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, the scope of protection for "A or B" covers at least three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. In addition, the terms "A and / or B," "at least one of A and B," and "at least one of A or B" also cover at least the above three scenarios. The character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0088] The term "instruction" in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction). A direct instruction can be understood as one in which the sender explicitly informs the receiver of specific information, the operation to be performed, or the requested result, etc., in the instruction sent. An indirect instruction can be understood as one in which the receiver determines the corresponding information based on the instruction sent by the sender, or makes a judgment and determines the operation to be performed or the requested result, etc., based on the judgment result.

[0089] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA), or other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used with the systems and radio technologies mentioned above, as well as with other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and the term NR is used in most of the following description; however, these technologies can also be applied to systems other than NR systems, such as 6th generation (6G) radio systems. th Generation 6G communication system.

[0090] Figure 1This diagram illustrates a block diagram of a wireless communication system applicable to embodiments of this application. The wireless communication system includes a terminal 11 and a network-side device 12. The terminal 11 can be a mobile phone, tablet computer, laptop computer, notebook computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR), virtual reality (VR) device, robot, wearable device, flight vehicle, vehicle user equipment (VUE), shipboard equipment, pedestrian user equipment (PUE), smart home devices (home appliances with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game consoles, personal computers (PCs), ATMs, or self-service machines, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among these, in-vehicle devices can also be referred to as in-vehicle terminals, in-vehicle controllers, in-vehicle modules, in-vehicle components, in-vehicle chips, or in-vehicle units, etc. It should be noted that the specific type of terminal 11 is not limited in this application embodiment. Network-side equipment 12 may include access network equipment or core network equipment, wherein access network equipment may also be referred to as Radio Access Network (RAN) equipment, radio access network function, or radio access network unit. Access network equipment may include base stations, Wireless Local Area Network (WLAN) access points (AS), or Wireless Fidelity (WiFi) nodes, etc.The term "base station" can be referred to as Node B (NB), Evolved Node B (eNB), Next Generation Node B (gNB), New Radio Node B (NR Node B), Access Point, Relay Base Station (RBS), Serving Base Station (SBS), Base Transceiver Station (BTS), Radio Base Station, Radio Transceiver, Basic Service Set (BSS), Extended Service Set (ESS), Home Node B (HNB), Home Evolved Node B, Transmit / Receive Point (TRP), or any other suitable term in the relevant field, as long as the same technical effect is achieved. The term "base station" is not limited to any specific technical terminology. It should be noted that this application embodiment only uses a base station in an NR system as an example for description and does not limit the specific type of base station.

[0091] Core network equipment, also known as core network nodes, core network functions, or core network elements, includes, but is not limited to, at least one of the following: Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), Home Subscriber Server (HSS), Centralized network configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (or L-NEF), and Binding Support Function. Support Functions (BSF), Application Functions (AF), Location Management Functions (LMF), Gateway Mobile Location Centres (GMLC), and Network Data Analytics Functions (NWDAF), etc. It should be noted that this application embodiment only uses core network equipment in the NR system as an example and does not limit the specific type of core network equipment. If the name of the core network equipment mentioned in this application embodiment changes in subsequent protocol versions (e.g., 6G), it will still be within the scope of protection of this application.

[0092] Optionally, the core network equipment can be implemented by one or more functional modules in a single device, or by multiple devices working together; this application does not specifically limit this. It is understood that the aforementioned functional modules can be network elements in hardware devices, software functional modules running on dedicated hardware, or virtualized functional modules instantiated on a platform (e.g., a cloud platform).

[0093] For ease of understanding, the following describes some aspects of the embodiments of this application:

[0094] I. Collection of training data for AI-based positioning.

[0095] In the AI-based positioning use cases discussed in 5G, a sample includes model input and labels during training data collection.

[0096] The model inputs include time-domain channel and positioning reference signal measurement results. For example, information such as the channel impulse response from at least one base station to the target UE, and channel quality (e.g., the real and imaginary parts of the channel at each time-domain sampling point of the daily line).

[0097] The true value or tag is the actual measured distance relative to the base station, or the time of arrival (TOA).

[0098] II. Training data collection for AI-based beam prediction.

[0099] In the AI-based beam prediction use cases discussed in 5G, a sample includes model input and label during training data collection.

[0100] The model inputs include (historical) partial or complete beam quality, beam identifier, measurement resource identifier, measurement association timestamp, etc.

[0101] The truth value or label is the quality of all beams (at future moments), or the strongest beam identifier, the timestamp associated with the prediction, etc.

[0102] III. Collection of training data for AI-based CSI prediction.

[0103] In the AI-based CSI prediction use cases discussed in 5G, a sample includes model input and label during training data collection.

[0104] The model inputs include historical channel matrices, precoding matrix indicators (PMIs), channel feature vectors or eigenvalues, and timestamps associated with measurements.

[0105] The truth value or label is the (future time) channel matrix, PMI, channel eigenvectors or eigenvalues, prediction association timestamps, etc.

[0106] For training data collection for AI-based positioning, ordinary users cannot provide ground truth values. Therefore, it is necessary to rely on certain specific users to provide ground truth values, which limits the number of data samples.

[0107] For AI-based beam prediction training data collection, compared to the inference process, training data collection requires additional measurement resources and terminal measurement power consumption in order to obtain the true value.

[0108] For AI-based CSI prediction, compared to the inference process, training data collection requires configuring measurement resources in a shorter period to obtain the true value. Therefore, it also consumes additional measurement resources and terminal measurement power consumption.

[0109] Therefore, the data collection method of this application is proposed. The data collection method provided by the embodiments of this application will be described in detail below with reference to the accompanying drawings and through some embodiments and application scenarios.

[0110] Reference Figure 2 This application provides a data collection method, such as... Figure 2 As shown, the data collection method includes:

[0111] Step 201: The first device sends first information to the second device. The first information is used to instruct the second AI unit to obtain target information based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0112] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0113] In this embodiment, since the training dataset of the second AI unit is determined by using the virtual truth value or virtual evaluation result obtained by the first AI unit inference, there is no need to obtain the truth value, thus avoiding the problems of not being able to obtain the truth value and the high cost of obtaining the truth value.

[0114] It should be noted that after the virtual truth value or virtual evaluation result determines the training dataset of the second AI unit, the parameters of the second AI unit can be trained and updated based on supervised learning, reinforcement learning or other learning methods to improve the inference performance of the second AI unit, thereby enhancing the gains of AI use cases.

[0115] Optionally, the above use case functions may include, but are not limited to, at least one of the following: positioning function, beam prediction function, and CSI prediction function.

[0116] Optionally, the first AI unit mentioned above can be understood or replaced as an evaluation model or a reward model. The virtual evaluation result mentioned above can be understood or replaced as the reward, feedback, or environmental feedback associated with the second AI unit. This virtual evaluation result can be understood as the deviation between the inference performance and the performance corresponding to the true value, or related information.

[0117] Alternatively, the aforementioned second device can also be understood as a device that generates labels for the second AI unit, generates rewards for the second AI unit, or is used to obtain the measurement truth value.

[0118] This application embodiment sends first information to a second device via a first device, thereby obtaining target information for the second AI unit based on the first information and inference from the first AI unit. The target information includes virtual truth values ​​or virtual evaluation results. The target information is used to determine the training dataset for the second AI unit, which implements use case functions. The first device implements the inference function of the second AI unit, and the second device implements the inference function of the first AI unit. Thus, since the training dataset for the second AI unit is determined using the virtual truth values ​​or virtual evaluation results obtained through inference from the first AI unit, there is no need to obtain the truth values, avoiding the problems of unobtainable truth values ​​and the high overhead associated with obtaining them.

[0119] Optionally, in some embodiments, the first information includes at least one of the following:

[0120] The target is the first sample data;

[0121] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0122] The measurement quantity associated with the target information;

[0123] The second device uses the first AI unit;

[0124] The second AI unit used by the first device;

[0125] The target first sample data includes at least one of the input data of the second AI unit and the output data of the second AI unit.

[0126] Optionally, the aforementioned input data can be understood or replaced with first descriptive information related to the input parameters of the AI ​​unit. This first descriptive information includes a defaultable identifier for each input parameter. For example, if each input parameter is associated with a defaultable identifier, and the input parameter is defaultable, then when the input parameter cannot be obtained, it does not need to be input. The aforementioned output data of the second AI unit can be understood or replaced with second descriptive information related to the output parameters of the AI ​​unit. The second descriptive information can describe the output parameters; for example, when the first output parameter output value is A, it represents a first meaning; when the first output parameter output value is B, it represents a second meaning.

[0127] Optionally, the indication methods for the above-mentioned input data and output data may include, but are not limited to, at least one of the following: measurement quantity; format information, such as dimension, number and order; quantization precision.

[0128] In this embodiment of the application, the measurement quantity associated with the target information can be understood or replaced as a parameter item indication or a processing method indication of the first AI unit.

[0129] Optionally, in some embodiments, before the first device sends the first information to the second device, the method further includes:

[0130] The first device and the third device transmit second information, which is used to indicate the collection of relevant information for the training dataset;

[0131] The third device is used to implement at least one of the control function and collection function for collecting training datasets, and the second information includes at least one of the following:

[0132] First indication information, used to indicate at least one candidate first sample data that is supported or expected;

[0133] The second instruction information is used to indicate at least one candidate purpose that supports or is expected to collect data;

[0134] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0135] In this embodiment, the aforementioned transmission may include at least one of sending and receiving. The second information sent by the terminal (first device) to the network-side device (third device) can be understood as request information; the second information sent by the network-side device (third device) to the terminal (first device) can be understood as configuration information. That is, the aforementioned second information may include information 1 (i.e., configuration information) sent by the network-side device to the terminal and information 2 (i.e., request information) sent by the terminal to the network-side device. The candidate information in the configuration information (such as candidate first sample data and at least one candidate destination) can be understood as candidate information supported by the network-side device, or configurable candidate information of the network-side device, or information configurable by the network-side device. The candidate information in the request information (such as a registration request) (such as candidate first sample data and at least one candidate destination) can be understood as candidate information expected by the terminal, or candidate information requested for configuration by the terminal, or information expected by the terminal, or information supported by the terminal, or candidate information supported by the terminal.

[0136] Optionally, the aforementioned candidate first sample data may specifically include, but is not limited to, at least one of the following:

[0137] The parameters of the input and output data include, for example, Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Signal to Interference plus Noise Ratio (SINR), Received Signal Strength Indication (RSSI), PMI, Channel Quality Indicator (CQI), and Rank Indication (RI).

[0138] Data format, such as dimensions and number; for example, dimensions may include time dimension, frequency dimension, Doppler dimension, angle dimension, beam dimension, etc.

[0139] Quantization precision. For example, whether the value is quantized using 8 bits or 16 bits.

[0140] Optionally, the above-mentioned candidate objectives are associated with the above-mentioned target information. In some embodiments, the candidate objectives include at least one of the following:

[0141] Determine the evaluation result of the second AI unit;

[0142] Determine the virtual truth value associated with the second AI unit.

[0143] Optionally, in some embodiments, after the first device and the third device transmit the second information, the method further includes:

[0144] The first device and the third device transmit third information, which is used to trigger the collection of the training dataset. The third information includes at least one of the following:

[0145] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0146] The fourth indication information is used to indicate the target first sample data among the at least one candidate first sample data;

[0147] The fifth instruction information is used to indicate the target objective among the at least one candidate objective;

[0148] Instructions used to direct monitoring.

[0149] In this embodiment of the application, the above transmission may include at least one of sending and receiving. The third information sent by the terminal (first device) to the network-side device (third device) can be understood as request information, used to request triggering the collection of the training dataset. The third information sent by the network-side device (third device) to the terminal (terminal) can be understood as configuration information, or trigger information, used to trigger the terminal to collect the training dataset. That is, the above third information may include information 3 (i.e., configuration information) sent by the network-side device to the terminal and information 4 (i.e., request information) sent by the terminal to the network-side device.

[0150] Optionally, the aforementioned target first sample data can be understood as the first sample data actually used, or an instance of the first sample data. Specifically, the target first sample data can be explicitly or implicitly indicated in the third information, and the first sample data can be one of the aforementioned candidate first sample data. The target purpose can be understood as the actual purpose of use. Specifically, the target purpose can be explicitly or implicitly indicated in the third information, and the purpose can be one of the aforementioned candidate purposes.

[0151] For example, the aforementioned third information can be based on predefined content or on content indicated in the second information, such as indicating one or more candidate first sample data in the second information. For instance, it can be indicated using log2(K), or in bitmap form, or in other ways. Here, K is the number of candidate information items.

[0152] It should be understood that when there is only one candidate information (candidate first sample data) in the second information, then the candidate information (candidate first sample data) is the target candidate information (target first sample data). In this case, the indication of the target candidate information corresponding to the candidate information can be omitted in the third information, or the indication of the third information can be omitted.

[0153] Optionally, in some embodiments, where the first device is further configured to implement at least one of the control function and the collection function for collecting the training dataset, the method further includes:

[0154] The first device transmits fourth information to the second device, the fourth information being used to indicate the collection of relevant information for the training dataset;

[0155] The fourth piece of information includes at least one of the following:

[0156] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0157] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0158] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0159] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0160] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0161] The candidate third AI unit is used to filter the training dataset.

[0162] In this embodiment of the application, the above transmission may include at least one of sending and receiving, wherein the fourth information sent by the terminal (first device) to the network-side device (second device) can be understood as request information; the second information sent by the network-side device (second device) to the terminal (first device) can be understood as configuration information. That is to say, the above fourth information may include information 5 (i.e., configuration information) sent by the network-side device to the terminal and information 6 (i.e., request information) sent by the terminal to the network-side device.

[0163] Optionally, the device used to implement at least one of the control function and the collection function for collecting the training dataset can be referred to as the third device. If the first device is used to implement at least one of the control function and the collection function for collecting the training dataset, it can be understood that the first device and the third device are the same device; otherwise, the first device and the third device are different devices.

[0164] It should be noted that the device used to implement the control function for collecting the training dataset and the device used to implement the collection function for collecting the training dataset can be the same device or different devices. The third device may include a fifth device and a sixth device, wherein the fifth device is used to implement the control function for collecting the training dataset, and the sixth device is used to implement the collection function for collecting the training dataset.

[0165] Optionally, the output data of the first AI unit can be understood or replaced with the target information. The input data of the first AI unit may include the first sample data. For example, in some embodiments, the input data of the first AI unit may include the input of the second AI unit and the output of the second AI unit, or the input data of the first AI unit may include the input of the second AI unit.

[0166] Optionally, the aforementioned candidate second sample data may specifically include, but is not limited to, at least one of the following:

[0167] The parameters or measurements of the input data, such as RSRP, RSRQ, SINR, RSSI, PMI, CQI, RI, etc.

[0168] The output data includes parameters or measurements. If the output is a virtual evaluation result, it could be, for example, prediction accuracy, prediction error, squared cosine similarity, positioning error, RSRP prediction error, etc. If the output is a virtual true value, it could be, for example, RSRP, RSRQ, SINR, RSSI, PMI, CQI, RI, distance to base station, TOA, etc.

[0169] Data format, such as dimensions and number; for example, dimensions may include time dimension, frequency dimension, Doppler dimension, angle dimension, beam dimension, etc.

[0170] Quantization precision. For example, whether the value is quantized using 8 bits or 16 bits.

[0171] Optionally, in some embodiments, after the first device and the second device transmit the fourth information, the method further includes:

[0172] The first device transmits fifth information to the second device, the fifth information being used to trigger the collection of the training dataset;

[0173] The fifth piece of information includes at least one of the following:

[0174] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0175] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0176] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0177] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0178] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0179] Instructions used to direct monitoring.

[0180] In this embodiment of the application, the above transmission may include at least one of sending and receiving. The fifth information sent by the terminal (first device) to the network-side device (second device) can be understood as request information, used to request triggering the collection of the training dataset. The fifth information sent by the network-side device (second device) to the terminal (first device) can be understood as configuration information, or trigger information, used to trigger the terminal to collect the training dataset. That is, the fifth information may include information 7 (i.e., configuration information) sent by the network-side device to the terminal and information 8 (i.e., request information) sent by the terminal to the network-side device.

[0181] Optionally, the aforementioned fifth information can be used to indicate the target candidate information or a use case of the candidate information actually used from the candidate information in the fourth information. For example, when multiple candidate second sample data are indicated in the fourth information, the target second sample data that is activated or actually used can be indicated by the fifth information. The target second sample data can be one of the multiple candidate second sample data.

[0182] It should be understood that when there is only one candidate information (candidate second sample data) in the fourth information, then the candidate information (candidate second sample data) is the target candidate information (target second sample data). In this case, the indication of the target candidate information corresponding to the candidate information can be omitted in the fifth information, or the indication of the fifth information can be omitted.

[0183] Optionally, in some embodiments, where the first device is further configured to implement at least one of the control function and the collection function for collecting the training dataset, the method further includes:

[0184] The first device sends a sixth message to the fourth device, the sixth message being used to instruct the collection of relevant information for the training dataset;

[0185] The sixth piece of information includes at least one of the following:

[0186] The first device supports or expects at least one candidate filtering rule;

[0187] The first device supports or expects at least one candidate sample classification category information;

[0188] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0189] The first device supports or expects at least one proportion of candidate samples;

[0190] The number of at least one candidate sample classifications supported or desired by the first device;

[0191] The first device supports or expects at least one proportion of candidate samples;

[0192] The first device supports or expects at least one candidate input data for the third AI unit;

[0193] At least one candidate output data of the third AI unit supported or desired by the first device;

[0194] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0195] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0196] In this embodiment of the application, the above transmission may include at least one of sending and receiving, wherein the sixth information sent by the terminal (first device) to the network-side device (fourth device) can be understood as request information; the second information sent by the network-side device (fourth device) to the terminal (first device) can be understood as configuration information. That is to say, the above sixth information may include information 9 (i.e., configuration information) sent by the network-side device to the terminal and information 10 (i.e., request information) sent by the terminal to the network-side device.

[0197] Optionally, in some embodiments, after the first device and the fourth device transmit the sixth information, the method further includes:

[0198] The first device sends a seventh message to the fourth device, the seventh message being used to trigger the collection of the training dataset;

[0199] The seventh piece of information includes at least one of the following:

[0200] The target filtering rule in the at least one candidate filtering rule;

[0201] The target classification sample category information in the at least one candidate sample classification category information;

[0202] The target threshold information in the at least one candidate threshold information;

[0203] The target sample ratio in the at least one candidate sample ratio;

[0204] The target sample classification number information in the at least one candidate sample classification number information;

[0205] At least the target sample proportion among the at least one candidate sample proportion;

[0206] The target input data is among at least one candidate input data of the third AI unit;

[0207] The target output data in at least one candidate output data of the third AI unit;

[0208] The sixteenth instruction indicates whether data filtering should be performed.

[0209] In this embodiment of the application, the above transmission may include at least one of sending and receiving. The seventh information sent by the terminal (first device) to the network-side device (fourth device) can be understood as request information, used to request triggering the collection of the training dataset. The seventh information sent by the network-side device (fourth device) to the terminal (first device) can be understood as configuration information, or trigger information, used to trigger the terminal to collect the training dataset. That is, the seventh information may include information 11 (i.e., configuration information) sent by the network-side device to the terminal and information 12 (i.e., request information) sent by the terminal to the network-side device.

[0210] Optionally, in the configuration information, the aforementioned sixteenth instruction information can be understood as configuring whether the fourth device performs data filtering; in the request information, the aforementioned sixteenth instruction information can be understood as requesting whether the fourth device performs data filtering.

[0211] Optionally, in some embodiments, where the first device is also used to implement the inference function of the third AI unit, the method further includes:

[0212] The first device transmits sixth information to the third device, the sixth information being used to indicate the collection of relevant information for the training dataset;

[0213] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the sixth information includes at least one of the following:

[0214] The first device supports or expects at least one candidate filtering rule;

[0215] The first device supports or expects at least one candidate sample classification category information;

[0216] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0217] The first device supports or expects at least one proportion of candidate samples;

[0218] The number of at least one candidate sample classifications supported or desired by the first device;

[0219] The first device supports or expects at least one proportion of candidate samples;

[0220] The first device supports or expects at least one candidate input data for the third AI unit;

[0221] At least one candidate output data of the third AI unit supported or desired by the first device;

[0222] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering.

[0223] In this application embodiment, the device used to implement the reasoning function of the third AI unit can be understood as the fourth device. When the first device is used to implement the reasoning function of the third AI unit, it can be understood that the first device and the fourth device are the same device; otherwise, the first device and the fourth device are different devices.

[0224] Optionally, in some embodiments, after the first device and the third device transmit the sixth information, the method further includes:

[0225] The first device receives seventh information from the third device, the seventh information being used to trigger the collection of the training dataset;

[0226] The seventh piece of information includes at least one of the following:

[0227] The target filtering rule in the at least one candidate filtering rule;

[0228] The target classification sample category information in the at least one candidate sample classification category information;

[0229] The target threshold information in the at least one candidate threshold information;

[0230] The target sample ratio in the at least one candidate sample ratio;

[0231] The target sample classification number information in the at least one candidate sample classification number information;

[0232] The target sample ratio in the at least one candidate sample ratio;

[0233] The target input data is among at least one candidate input data of the third AI unit;

[0234] The target output data in at least one candidate output data of the third AI unit;

[0235] The sixteenth instruction indicates whether data filtering should be performed.

[0236] Optionally, in some embodiments, the first device is a terminal, a base station, or a core network device; the second device is a terminal, a base station, or a core network device; the third device is a terminal, a base station, or a core network device; and the fourth device is a terminal, a base station, or a core network device.

[0237] To better understand this application, some examples are provided below.

[0238] Option 1, such as Figure 2a As shown, the first AI unit provides virtual evaluation results to the second AI unit, generating a training dataset for reinforcement learning. The process includes the following steps:

[0239] Step 11: Collect the input data of the second AI unit, perform forward propagation or model inference of the second AI unit, and obtain the prediction results or behavior.

[0240] Step 12: Use the input data (or the state of reinforcement learning) of the second AI unit and the output data (or the behavior of reinforcement learning) of the second AI unit as the input data of the first AI unit to perform model inference of the first AI unit and obtain the virtual evaluation result, or the reward or environmental feedback of the second AI unit.

[0241] Step 13: Use the input data, output data, and virtual evaluation results of the second AI unit as training data samples for the second AI unit. Subsequently, reinforcement learning can be used to train and update the parameters of the first model.

[0242] In this scheme, the input data of the first AI unit includes the input data and output data of the second AI unit. The reasoning result of the first AI unit is a virtual evaluation result, such as the deviation between the reasoning performance and the true value correspondence performance, or related information.

[0243] The training dataset for the second AI unit includes: the input data of the second AI unit, the output data of the second AI unit, and the virtual evaluation results. The virtual evaluation results can be used as a reward during reinforcement learning.

[0244] Option 2, such as Figure 2b As shown, the first AI unit provides the labels for the training data of the second AI unit, and the third AI unit filters the training data to obtain the target dataset.

[0245] Step 21: Collect input data for the second AI unit.

[0246] Step 22: Use the input data of the second AI unit as the input data of the first AI unit to perform inference in the first AI unit and obtain the virtual truth value. Generate the training dataset for the second AI unit.

[0247] Step 23: Based on the third AI unit, perform data filtering on the training dataset to obtain the target dataset.

[0248] In this scheme, the input data of the first AI unit includes the input data of the second AI unit, and the reasoning result of the first AI unit is a virtual truth value.

[0249] The input data of the third AI unit includes the input data of the second AI unit and the virtual ground truth, and the output data of the third AI unit is the filtered target dataset.

[0250] The training dataset for the second AI unit includes: the input data of the second AI unit and the virtual ground truth of the second AI unit. Specifically, a training sample for the second AI unit includes the input data of the second AI unit and the associated virtual ground truth.

[0251] Option 3, such as Figure 2c As shown, the training data for the second AI unit comes from ground truth UEs or additional measurements, but the dataset is filtered by the third AI unit. This includes the following process:

[0252] Step 31: Collect the input data and measurement ground truth of the second AI unit to obtain the training dataset.

[0253] Step 32: Based on the third AI unit, perform data filtering on the training dataset to obtain the target dataset.

[0254] In this scheme, the input data of the third AI unit includes the input data of the second AI unit and the measurement true value, and the output data of the third AI unit is the filtered target dataset.

[0255] The training dataset for the second AI unit includes: the input data for the second AI unit and the ground truth measurements for the second AI unit.

[0256] One training sample for the second AI unit includes the input data of the second AI unit and the associated ground truth measurement.

[0257] The main process of this application is as follows: Figure 3 As shown, it includes the following steps:

[0258] Step 1: Collect the training dataset of the second AI unit and related configurations or registrations. The devices involved may include the first device, the second device, the fourth device, the fifth device, and the sixth device.

[0259] Step 2: The training dataset of the second AI unit collects relevant triggers or requests, which may involve devices including the first device, the second device, the fourth device, the fifth device, and the sixth device.

[0260] Step 3: Collection and screening of training datasets for the second AI unit.

[0261] Step 1 is a pre-negotiation process, while steps 2 and 3 are the execution processes for collecting and filtering training datasets.

[0262] For step 1 above, the following methods are included:

[0263] Method 1, Base Station or Core Network Configuration Process (The fifth device is a network-side device), such as... Figure 3a As shown, the process includes the following steps.

[0264] Step 1a: The fifth device sends information 1 (i.e., second information) to the first device, said information 1 being used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0265] First indication information, used to indicate at least one candidate first sample data supported;

[0266] The second instruction information is used to indicate at least one candidate purpose that supports the collection of the training dataset (corresponding to the above scheme 1 (i.e., the evaluation result or reward of the second AI unit), scheme 2 (i.e., virtual truth value) and scheme 3 (i.e., data filtering)).

[0267] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0268] Step 1b: The fifth device sends information 5 (i.e., the fourth information) to the second device. Information 5 is used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0269] The sixth indication information is used to indicate whether the first device supports generating target information based on the first AI unit;

[0270] The seventh indication information is used to indicate at least one candidate second sample data supported by the first device, wherein the candidate second sample data includes at least one of the input data of the first AI unit and the output data of the first AI unit:

[0271] The eighth indication information is used to indicate at least one candidate second AI unit supported by the first device;

[0272] The ninth indication information is used to indicate at least one candidate third AI unit supported by the first device;

[0273] The tenth instruction information is used to indicate that the first device supports at least one candidate purpose for data collection.

[0274] Step 1c: The fifth device sends information 9 (i.e., the sixth information) to the fourth device. Information 9 is used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0275] The first device supports at least one candidate filtering rule;

[0276] The first device supports at least one candidate sample classification category information;

[0277] The first device supports at least one candidate threshold information, which is used to determine the sample classification category;

[0278] The proportion of at least one candidate sample supported by the first device;

[0279] The number of at least one candidate sample classifications supported by the first device;

[0280] The proportion of at least one candidate sample supported by the first device;

[0281] At least one candidate input data of the third AI unit supported by the first device;

[0282] At least one candidate output data of the third AI unit supported by the first device;

[0283] The fifteenth instruction is used to indicate whether the first device supports data filtering.

[0284] It should be understood that steps 1a, 1b, and 1c above are not sequential. Depending on the scheme, not all of the above steps are required. For example, step 1c is not required for scheme 1; for scheme 3, it does not involve information related to the first AI unit.

[0285] Method 2: Network configuration process (the fifth device is the terminal), such as Figure 3b As shown, the process includes the following steps.

[0286] Step 1-2a: The first device sends information 1 (i.e., the second information) to the fifth device.

[0287] Step 1-2b: The second device sends information 5 (i.e., the fourth information) to the fifth device.

[0288] Step 1-2c: The fourth device sends information 9 (i.e., the sixth information) to the fifth device.

[0289] In the sixth information, instructions or displays can be made based on predefined information.

[0290] The specific contents of information 1, information 5 and information 9 are as described in method 1 in step 1, and will not be repeated here.

[0291] Depending on the specific scheme, not all of the above steps are required. For example, for scheme 1, steps 1-2c are not needed; for scheme 3, information related to the first AI unit is not involved.

[0292] Method 3: UE registration process (the fifth device is the UE), such as Figure 3c As shown, the process includes the following steps.

[0293] Step 1-1a: The fifth device sends information 2 (i.e., the second information) to the first device. Information 2 is used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0294] First indication information, used to indicate at least one candidate first sample data as desired;

[0295] The second indication information is used to indicate at least one candidate purpose for the expected collection of the training dataset;

[0296] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0297] Step 1-2a: The first device sends information 1 to the fifth device. Information 1 may be based on the content of information 2, or based on predefined information, or may be a display instruction. No further limitations are made here.

[0298] Step 1-1b: The fifth device sends information 6 (i.e., the fourth information) to the second device. Information 6 is used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0299] The sixth indication information is used to indicate whether the first device expects to generate target information based on the first AI unit;

[0300] The seventh indication information is used to indicate at least one candidate second sample data that the first device expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0301] The eighth indication information is used to indicate at least one candidate second AI unit desired by the first device;

[0302] The ninth indication information is used to indicate at least one candidate third AI unit desired by the first device;

[0303] The tenth instruction information is used to indicate at least one candidate purpose for the first device to collect data.

[0304] Step 1-2b: The second device sends information 5 to the fifth device. Information 5 may be based on the content of information 6, or based on predefined information, or may be a display instruction. No further limitations are made here.

[0305] Step 1-1c: The fifth device sends information 10 (i.e., the sixth information) to the fourth device. Information 10 is used to instruct the collection of relevant information for the training dataset, including at least one of the following:

[0306] The first device expects at least one candidate filtering rule;

[0307] The first device expects at least one candidate sample classification category information;

[0308] The first device expects at least one candidate threshold information, which is used to determine the sample classification category;

[0309] The proportion of at least one candidate sample desired by the first device;

[0310] The number of candidate sample classifications expected by the first device;

[0311] The proportion of at least one candidate sample desired by the first device;

[0312] The first device expects at least one candidate input data from the third AI unit;

[0313] The first device expects at least one candidate output data from the third AI unit;

[0314] The fifteenth instruction is used to indicate whether the first device supports data filtering.

[0315] Step 1-2c: The fourth device sends information 9 to the fifth device. Information 9 may be based on the content of information 10, or based on predefined information, or may be a display instruction. No further limitations are made here.

[0316] The specific contents of information 1, information 5 and information 9 are as described in method 1 in step 1, and will not be repeated here.

[0317] Depending on the specific scheme, not all of the above steps are required. For example, for scheme 1, steps 1-1c and 1-2c are not needed; for scheme 3, information related to the first AI unit is not involved.

[0318] For step 2 above, the following methods are included:

[0319] Method 1, Base Station or Core Network Configuration Process (The fifth device is a network-side device), such as... Figure 3d As shown, the process includes the following steps.

[0320] Step 2a: The fifth device sends third information to the first device, the third information being used to trigger the collection of the training dataset, the third information including at least one of the following:

[0321] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0322] The fourth indication information is used to indicate the target first sample data among the at least one candidate first sample data;

[0323] The fifth instruction information is used to indicate the target objective among the at least one candidate objective;

[0324] Instructions used to direct monitoring.

[0325] Step 2b: The fifth device sends a fifth message to the second device, the fifth message being used to trigger the collection of the training dataset;

[0326] The fifth piece of information includes at least one of the following:

[0327] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0328] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0329] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0330] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0331] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0332] Instructions used to direct monitoring.

[0333] Step 2c: The fifth device sends a seventh message to the fourth device, the seventh message being used to trigger the collection of the training dataset;

[0334] The seventh piece of information includes at least one of the following:

[0335] The target filtering rule in the at least one candidate filtering rule;

[0336] The target classification sample category information in the at least one candidate sample classification category information;

[0337] The target threshold information in the at least one candidate threshold information;

[0338] The target sample ratio in the at least one candidate sample ratio;

[0339] The target sample classification number information in the at least one candidate sample classification number information;

[0340] At least the target sample proportion among the at least one candidate sample proportion;

[0341] The target input data is among at least one candidate input data of the third AI unit;

[0342] The target output data in at least one candidate output data of the third AI unit;

[0343] The sixteenth instruction indicates whether data filtering should be performed.

[0344] Optionally, in this embodiment, the content indicated in the third, fifth, and seventh information can be indicated based on predefined content or based on content that has already been interacted with, such as indicating the identifiers of the corresponding second, fourth, and sixth information.

[0345] Method 2, UE request triggering process (the fifth device is the UE), such as Figure 3e As shown, the specific process includes the following steps.

[0346] Step 2-1a: The fifth device sends information 4 (i.e., the third information) to the first device. Information 4 is used to request the triggering of the collection of the training dataset.

[0347] Step 2-2a: The first device sends information 3 (i.e., the third information) to the fifth device. Information 3 is used to trigger the collection of the training dataset.

[0348] Step 2-1b: The fifth device sends information 8 (i.e., the fifth information) to the second device. Information 8 is used to request the triggering of the collection of the training dataset.

[0349] Step 2-2b: The second device sends information 7 (i.e., the fifth information) to the fifth device. Information 7 is used to trigger the collection of the training dataset.

[0350] The indication in information 7 can be based on information 8 or a fourth piece of information. For example, the method of indicating the second sample data based on information 8 could be that information 8 contains multiple identifiers, each corresponding to a predefined variety of second sample data. For example, there could be four types. Then, the method of indicating the target second sample data based on information 8 could be 01, indicating that the second type of second sample data among the four types is indicated.

[0351] Step 2-1c: The fifth device sends information 12 (i.e., the seventh information) to the fourth device. Information 12 is used to request the triggering of the collection of the training dataset.

[0352] Step 2-2c: The fourth device sends information 11 (i.e., the seventh information) to the fifth device. Information 11 is used to trigger the collection of the training dataset.

[0353] The content contained in information 11 can be based on the content contained in information 12 or the sixth information, or predefined content. For example, information 11 can indicate the identifier of the corresponding information in information 11.

[0354] For step 3 above, different data collection processes correspond to different schemes:

[0355] For option 1, such as Figure 3f As shown, the data collection process for the input data, output data, and reward of the second AI unit includes the following steps:

[0356] Step 3-1: The first device sends first information to the second device, the first information including at least one of the following:

[0357] The target first sample data includes the input data of the second AI unit and the output data of the second AI unit;

[0358] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0359] The measurement quantity associated with the target information;

[0360] The second device uses the first AI unit;

[0361] The second AI unit used by the first device.

[0362] Step 3-2: The second device sends an eighth message to the sixth device, the eighth message including at least one of the following:

[0363] The training dataset includes the input data of the second AI unit, the output data of the second AI unit, and the virtual evaluation result (i.e., virtual reward) of the second AI unit.

[0364] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0365] The training dataset is based on the instruction information determined by the first AI unit;

[0366] The measurement quantity associated with the target information;

[0367] The second device uses the first AI unit;

[0368] The second AI unit used by the first device.

[0369] For scheme 2, the data collection and filtering of input data and virtual truth values ​​for the second AI unit. For example... Figure 3g As shown, the process includes the following:

[0370] Step 3-1: The first device sends first information to the second device, the first information including at least one of the following:

[0371] The target first sample data includes the input data of the second AI unit;

[0372] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0373] Measurements related to virtual truth values;

[0374] The second device uses the first AI unit;

[0375] The second AI unit used by the first device.

[0376] Step 3-3: The second device sends the ninth message to the fourth device. The ninth message includes at least one of the following:

[0377] The training dataset includes: input data of the second AI unit and virtual ground truth of the second AI unit;

[0378] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0379] Measurements related to virtual truth values;

[0380] The second device uses the first AI unit;

[0381] The second AI unit used by the first device;

[0382] The third AI unit used by the fourth device.

[0383] Steps 3-4: The fourth device sends the tenth message to the sixth device. The tenth message includes:

[0384] The training dataset includes: input data of the second AI unit and virtual ground truth of the second AI unit;

[0385] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0386] The third AI unit used by the fourth device.

[0387] For scheme 3, the data collection and filtering of input data and measurement truth values ​​for the second AI unit, such as... Figure 3h As shown, the process includes the following:

[0388] Step 3-1: The first device sends first information to the second device, the first information including at least one of the following:

[0389] The target first sample data includes the input data of the second AI unit;

[0390] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0391] The measurement indication of the second device is used to acquire and process the measurement true value.

[0392] Step 3-3: The second device sends an eleventh message to the fourth device. The eleventh message includes at least one of the following:

[0393] The training dataset includes: input data of the second AI unit and the measurement ground truth of the second AI unit;

[0394] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0395] Measurement quantities that are correlated with the truth value;

[0396] The third AI unit used by the fourth device.

[0397] Steps 3-4: The fourth device sends the twelfth message to the sixth device. The twelfth message includes at least one of the following:

[0398] The training dataset includes: input data of the second AI unit and the measurement ground truth of the second AI unit;

[0399] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0400] The third AI unit used by the fourth device.

[0401] The following examples illustrate different combined installation scenarios for the first, second, fifth, sixth, and fourth devices.

[0402] Example 1, Co-location Scenario 1: The first device, the fifth device, and the sixth device are the same device, i.e., the terminal; the second device and the fourth device are the same device, i.e., the base station. The above AI use cases can be AI beam prediction, CSI prediction, or AI positioning. Since the first device, the fifth device, and the sixth device are the same device, the interaction between the first device, the fifth device, and the sixth device is an internal interaction, thereby reducing the overhead of signaling interaction.

[0403] Example 1-1a: Step 1 uses method 2 (network configuration process, the fifth device is the terminal) from step 1 above; Step 2 uses method 2 (UE request triggering process, the fifth device is the UE) from step 2 above; Step 3 adopts scheme 1 (including input data of the second AI unit, output data of the second AI unit, and data collection of the second AI unit). The corresponding process is as follows: Figure 4a As shown, the process includes the following steps.

[0404] Step 1-2b: The base station (the second and fourth devices are combined) sends information 5 (i.e., the fourth information) to the terminal (the first, fifth, and sixth devices are combined);

[0405] Step 2-1b (optional): The terminal sends information 8 (i.e., the fifth information) to the base station;

[0406] Step 2-2b (optional): The base station sends information 7 (i.e., the fifth information) to the terminal;

[0407] If the configuration information in steps 1-2b is periodic, or only includes one candidate data collection method, then steps 2-1b and 2-2b are not required.

[0408] Step 3-1: The terminal sends the first information to the base station;

[0409] Step 3-2: The base station sends the eighth information to the terminal.

[0410] Before step 3-2, the base station, as the second device, obtains the eighth information based on the first information and the first AI unit.

[0411] Step 4: The terminal obtains the target dataset based on the eighth information.

[0412] Example 1-1b: Step 1 uses method 3 (UE registration process, the fifth device is the UE) from step 1 above; Step 2 uses method 2 (UE request triggering process, the fifth device is the UE) from step 2 above; Step 3 adopts scheme 1 (including the collection of input data of the second AI unit, output data of the second AI unit, and reward data of the second AI unit). The corresponding process is as follows: Figure 4b As shown, the process includes the following steps.

[0413] Step 1-1b: The terminal (combined with the first device, the fifth device, and the sixth device) sends information 6 (i.e., the fourth information) to the base station (combined with the second device and the fourth device);

[0414] Step 1-2b: The base station sends information 5 (i.e., the fourth information) to the terminal;

[0415] Step 2-1b (optional): The terminal sends information 8 (i.e., the fifth information) to the base station;

[0416] Step 2-2b (optional): The base station sends information 7 (i.e., the fifth information) to the terminal;

[0417] If the configuration information in steps 1-2b is periodic, or only includes one candidate data collection method, then steps 2-1b and 2-2b are not required.

[0418] Step 3-1: The terminal sends the first information to the base station;

[0419] Step 3-2: The base station sends the eighth information to the terminal.

[0420] Before step 3-2, the base station, as the second device, obtains the eighth information based on the first information and the first AI unit.

[0421] Step 4: The terminal obtains the target dataset based on the eighth information.

[0422] In Examples 1-2, step 1 uses method 2 from step 1 above (network configuration process, the fifth device is the terminal), step 2 uses method 2 from step 2 above (UE request triggering process, the fifth device is the UE), and step 3 adopts scheme 2 (including the collection and filtering of input data and virtual truth values ​​from the second AI unit). The corresponding process is as follows: Figure 4c As shown, the process includes the following steps.

[0423] Step 1-2b: The base station (the second and fourth devices are combined) sends information 5 (i.e., the fourth information) to the terminal (the first, fifth, and sixth devices are combined);

[0424] Step 1-2c: The base station sends information 9 (i.e., the sixth information) to the terminal;

[0425] Among them, 1-2b and 1-2c can also be sent together in a single message.

[0426] Step 2-1b (optional): The terminal sends information 8 (i.e., the fifth information) to the base station;

[0427] Step 2-2b (optional): The base station sends information 7 (i.e., the fifth information) to the terminal;

[0428] Step 2-1c (optional): The terminal sends information 12 (i.e., the seventh information) to the base station;

[0429] Step 2-2c (optional): The base station sends information 11 (i.e., the seventh information) to the terminal;

[0430] In this case, steps 2-1b and 2-1c can be combined in one message and sent by the terminal to the base station; steps 2-2b and 2-2c can also be combined in one message and sent by the base station to the terminal. Furthermore, if the configuration information in step 1-2b is periodic, or only includes one candidate data collection method, then steps 2-1b and 2-2b are not required; if the configuration information in step 1-2c is periodic, or only includes one candidate data collection method, then steps 2-1c and 2-2c are not required.

[0431] Step 3-1: The terminal sends the first information to the base station;

[0432] Steps 3-4: The base station sends the tenth information to the terminal.

[0433] Before steps 3-4, the base station, as the second device, determines the training dataset based on the first information and the first AI unit, and then, as the fourth device, filters the training dataset based on the third AI unit to obtain the tenth information.

[0434] Step 4: The terminal obtains the target dataset based on the tenth information.

[0435] Examples 1-3: Step 1 uses method 2 from step 1 above (network configuration process, the fifth device is the terminal); Step 2 uses method 2 from step 2 above (UE request triggering process, the fifth device is the UE); Step 3 adopts scheme 3 (including data collection and filtering of input data and measurement truth values ​​of the second AI unit). The corresponding process is as follows: Figure 4d As shown, the process includes the following steps.

[0436] Step 1-2b: The base station (the second and fourth devices are combined) sends information 5 (i.e., the fourth information) to the terminal (the first, fifth, and sixth devices are combined);

[0437] Step 1-2c: The base station sends information 9 (i.e., the sixth information) to the terminal;

[0438] Among them, 1-2b and 1-2c can also be sent together in a single message.

[0439] Step 2-1b (optional): The terminal sends information 8 (i.e., the fifth information) to the base station;

[0440] Step 2-2b (optional): The base station sends information 7 (i.e., the fifth information) to the terminal;

[0441] Step 2-1c (optional): The terminal sends information 12 (i.e., the seventh information) to the base station;

[0442] Step 2-2c (optional): The base station sends information 11 (i.e., the seventh information) to the terminal;

[0443] In this case, steps 2-1b and 2-1c can be combined into a single message and sent from the terminal to the base station; similarly, steps 2-2b and 2-2c can be combined into a single message and sent from the base station to the terminal. Furthermore, if the configuration information in step 1-2b is periodic, or only includes one candidate data collection method, steps 2-1b and 2-2b are unnecessary; similarly, if the configuration information in step 1-2c is periodic, or only includes one candidate data collection method, steps 2-1c and 2-2c are unnecessary.

[0444] Step 3-1: The terminal sends the first information to the base station;

[0445] Step 3-3: The base station sends the eleventh message to the terminal;

[0446] Steps 3-4: The base station sends the twelfth message to the terminal.

[0447] Before steps 3-4, the base station, as the fourth device, filters the training dataset based on the third AI unit to obtain the twelfth information.

[0448] Step 4: The terminal obtains the target dataset based on the twelfth piece of information.

[0449] Example 2: Co-location scenario 1: The first, fourth, fifth, and sixth devices are the same device, i.e., a base station; the second device is a terminal. The above AI use case can be AI beam prediction or CSI prediction. Since the first, fourth, fifth, and sixth devices are the same device, the interaction between the first, fourth, fifth, and sixth devices is an internal interaction, thereby reducing the overhead of signaling interaction.

[0450] Example 2-1: Step 1 uses Method 1 from Step 1 above (base station or core network configuration process, the fifth device is a network-side device); Step 2 uses Method 1 from Step 2 above (UE request triggering process, the fifth device is a UE); Step 3 adopts Scheme 1 (including the collection of input data of the second AI unit, output data of the second AI unit, and reward data of the second AI unit). The corresponding process is as follows: Figure 4e As shown, the process includes the following steps.

[0451] Step 1b: The base station (the first device, the fourth device, the fifth device, and the sixth device are the same device) sends information 5 (i.e., the fourth information) to the terminal (the second device);

[0452] Step 2b (optional): The base station sends the fifth information to the terminal;

[0453] If the fourth message contains periodic information, or only one candidate acquisition method is configured, then the fifth message need not be sent.

[0454] Step 3-1: The base station sends the first information to the terminal;

[0455] Step 3-2: The terminal sends the eighth information to the base station.

[0456] Before step 3-2, the terminal, as the second device, determines the training dataset based on the first information and the first AI unit, which is included in the eighth information.

[0457] Step 4: The base station obtains the target dataset based on the eighth information.

[0458] Example 2-2: Step 1 uses Method 1 from Step 1 above (base station or core network configuration process, the fifth device is a network-side device); Step 2 uses Method 1 from Step 2 above (base station or core network configuration process, the fifth device is a network-side device); Step 3 adopts Scheme 2 (including the collection and filtering of input data and virtual truth values ​​from the second AI unit). The corresponding process is as follows: Figure 4e As shown, the process includes the following steps.

[0459] Step 1b: The base station (the first device, the fourth device, the fifth device, and the sixth device are the same device) sends information 5 (i.e., the fourth information) to the terminal (the second device);

[0460] Step 2b (optional): The base station sends the fifth information to the terminal;

[0461] If the fourth message contains periodic information, or only one candidate acquisition method is configured, then the fifth message need not be sent.

[0462] Step 3-1: The base station sends the first information to the terminal;

[0463] Step 3-3: The terminal sends the ninth information to the base station.

[0464] Before step 3-3, the terminal, as the second device, obtains the ninth information based on the first information and the first AI unit.

[0465] Step 4: The base station, acting as the fourth device, filters the ninth piece of information and obtains the target dataset.

[0466] Example 3: Co-location scenario 3: The first device is a terminal, the second device is a base station, the fifth and sixth devices are the same device, i.e., LMF, and the fourth device is an NWDAF or a newly added core network element. The above AI use case can be used for AI positioning. In this example, since the first and second devices are set as independent devices, only one type of AI unit needs to be supported on the same device, thereby reducing the requirements for the device.

[0467] Example 3-1: Step 1 uses Method 1 from Step 1 above (base station or core network configuration process, the fifth device is a network-side device); Step 2 uses Method 1 from Step 2 above (base station or core network configuration process, the fifth device is a network-side device); Step 3 adopts Scheme 1 (including the collection of input data of the second AI unit, output data of the second AI unit, and reward data of the second AI unit). The corresponding process is as follows: Figure 4f As shown, the process includes the following steps.

[0468] Step 1a: The LMF (the fifth and sixth devices are the same device) sends information 1 (i.e., the second information) to the terminal (the first device);

[0469] Step 1b: The LMF sends the fourth message to the base station (second device);

[0470] Step 2a (optional): The LMF sends third information to the terminal;

[0471] Step 2b (optional): The LMF sends the fifth message to the base station;

[0472] If the configuration information in step 1a is periodic, or only includes one candidate data acquisition method, then step 2a is not required; if the configuration information in step 1b is periodic, or only includes one candidate data acquisition method, then step 2b is not required.

[0473] Step 3-1: The terminal sends the first information to the base station;

[0474] Step 3-2: The base station sends the eighth information to the LMF.

[0475] Before step 3-2, the base station, as the second device, determines the training dataset based on the first information and the first AI unit, which is included in the eighth information.

[0476] Step 4: LMF obtains the target dataset based on the eighth information.

[0477] Example 3-2: Step 1 uses Method 1 from Step 1 above (base station or core network configuration process, the fifth device is a network-side device); Step 2 uses Method 1 from Step 2 above (base station or core network configuration process, the fifth device is a network-side device); Step 3 adopts Scheme 2 (including the collection and filtering of input data and virtual truth values ​​from the second AI unit). The corresponding process is as follows: Figure 4g As shown, the process includes the following steps.

[0478] Step 1a: The LMF (the fifth and sixth devices are combined) sends information 1 (i.e., the second information) to the terminal (the first device);

[0479] Step 1b: The LMF sends the fourth message to the base station (second device);

[0480] Step 1c: The LMF sends the sixth message to the NWDAF or other core network elements (the fourth device);

[0481] Step 2a (optional): The LMF sends third information to the terminal;

[0482] Step 2b (optional): The LMF sends the fifth message to the base station;

[0483] Step 2c (optional): The LMF sends the seventh message to the NWDAF or other core network elements;

[0484] If the configuration information in step 1a is periodic, or only includes one candidate data acquisition method, then step 2a is not required; if the configuration information in step 1b is periodic, or only includes one candidate data acquisition method, then step 2b is not required; if the configuration information in step 1c is periodic, or only includes one candidate data acquisition method, then step 2c is not required.

[0485] Step 3-1: The terminal sends the first information to the base station;

[0486] Step 3-3: The base station sends the ninth information to the NWDAF or other core network elements;

[0487] Before step 3-3, the base station, as the second device, determines the training dataset based on the first information and the first AI unit, which is included in the ninth information.

[0488] Steps 3-4: NWDAF or other core network elements send the tenth message to LMF.

[0489] Before steps 3-4, NWDAF or other core network elements, as the fourth device, filter the training dataset based on the third AI unit to obtain the tenth information.

[0490] Step 4: LMF obtains the target dataset based on the tenth information.

[0491] Example 3-3: Step 1 uses Method 1 from Step 1 above (base station or core network configuration process, the fifth device is a network-side device); Step 2 uses Method 1 from Step 2 above (base station or core network configuration process, the fifth device is a network-side device); Step 3 adopts Scheme 3 (including data collection and filtering of input data and measurement truth values ​​for the second AI unit). The corresponding process is as follows: Figure 4h As shown, the process includes the following steps.

[0492] Step 1a: The LMF (the fifth and sixth devices are combined) sends information 1 (i.e., the second information) to the terminal (the first device);

[0493] Step 1b: The LMF sends the fourth message to the base station (second device);

[0494] Step 1c: LMF sends the sixth information to NWDAF or other core network elements (fourth device).

[0495] Step 2a (optional): The LMF sends third information to the terminal;

[0496] Step 2b (optional): The LMF sends the fifth message to the terminal;

[0497] Step 2c (optional): The LMF sends seven pieces of information to the NWDAF or the newly added core network element;

[0498] If the configuration information in step 1a is periodic, or only includes one candidate data acquisition method, then step 2a is not required; if the configuration information in step 1b is periodic, or only includes one candidate data acquisition method, then step 2b is not required; if the configuration information in step 1c is periodic, or only includes one candidate data acquisition method, then step 2c is not required.

[0499] Step 3-1: The terminal sends the first information to the base station;

[0500] Step 3-3: The base station sends the eleventh message to the NWDAF or the newly added core network element;

[0501] Before step 3-3, the base station, as the second device, collects the truth value or label data associated with the model input data of the second AI unit, which is included in the eleventh information.

[0502] Steps 3-4: The NWDAF or the newly added core network element sends the twelfth message to the LMF.

[0503] Before steps 3-4, NWDAF or other core network elements, as the fourth device, filter the training dataset based on the third AI unit to obtain the twelfth information.

[0504] Step 4: LMF obtains the target dataset based on the twelfth information.

[0505] It should be noted that the definitions of each piece of information in Embodiments 1 to 3 of this application can be referred to the above. Figure 3 The definitions of each piece of information in the main process shown are not repeated here.

[0506] Reference Figure 5 This application also provides a data collection method, such as... Figure 5 As shown, the data collection method includes:

[0507] Step 501: The second device receives the first information from the first device;

[0508] Step 502: The second device obtains the target information of the second AI unit based on the first information and the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0509] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0510] Optionally, the first information includes at least one of the following:

[0511] The target is the first sample data;

[0512] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0513] The measurement quantity associated with the target information;

[0514] The second device uses the first AI unit;

[0515] The second AI unit used by the first device;

[0516] The target first sample data includes at least one of the input data of the second AI unit and the output data of the second AI unit.

[0517] Optionally, after the second device obtains the target information of the second AI unit based on the first information and the reasoning of the first AI unit, the method further includes:

[0518] The second device sends an eighth message to the third device, the eighth message including the training dataset;

[0519] The third device is used to implement at least one of the control function and collection function for collecting training datasets.

[0520] Optionally, the eighth information may further include at least one of the following:

[0521] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0522] The training dataset is based on the instruction information determined by the first AI unit;

[0523] The measurement quantity associated with the target information;

[0524] The second device uses the first AI unit;

[0525] The second AI unit used by the first device.

[0526] Optionally, after the second device obtains the target information of the second AI unit based on the first information and the reasoning of the first AI unit, the method further includes:

[0527] The second device sends a ninth message to the fourth device, the ninth message including the training dataset;

[0528] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0529] Optionally, the ninth information further includes at least one of the following:

[0530] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0531] The measurement quantity associated with the target information;

[0532] The second device uses the first AI unit;

[0533] The second AI unit used by the first device;

[0534] The third AI unit used by the fourth device.

[0535] Optionally, if the second device is also used to implement the inference function of the third AI unit, the method further includes:

[0536] The second device transmits a sixth message to the third device, the sixth message being used to indicate information for collecting relevant information for the training dataset;

[0537] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the sixth information includes at least one of the following:

[0538] The first device supports or expects at least one candidate filtering rule;

[0539] The first device supports or expects at least one candidate sample classification category information;

[0540] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0541] The first device supports or expects at least one proportion of candidate samples;

[0542] The number of at least one candidate sample classifications supported or desired by the first device;

[0543] The first device supports or expects at least one proportion of candidate samples;

[0544] The first device supports or expects at least one candidate input data for the third AI unit;

[0545] At least one candidate output data of the third AI unit supported or desired by the first device;

[0546] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering.

[0547] Optionally, after the second device and the third device transmit the sixth information, the method further includes:

[0548] The second device transmits seventh information to the third device, the seventh information being used to trigger the collection of the training dataset;

[0549] The seventh piece of information includes at least one of the following:

[0550] The target filtering rule in the at least one candidate filtering rule;

[0551] The target classification sample category information in the at least one candidate sample classification category information;

[0552] The target threshold information in the at least one candidate threshold information;

[0553] The target sample ratio in the at least one candidate sample ratio;

[0554] The target sample classification number information in the at least one candidate sample classification number information;

[0555] The target sample ratio in the at least one candidate sample ratio;

[0556] The target input data is among at least one candidate input data of the third AI unit;

[0557] The target output data in at least one candidate output data of the third AI unit;

[0558] The sixteenth instruction indicates whether data filtering should be performed.

[0559] Optionally, the method further includes:

[0560] The second device transmits fourth information to the third device, the fourth information being used to indicate information for collecting relevant information for the training dataset;

[0561] The fourth piece of information includes at least one of the following:

[0562] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0563] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0564] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0565] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0566] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0567] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the candidate third AI unit is used to filter the training dataset.

[0568] Optionally, after the second device and the third device transmit the fourth information, the method further includes:

[0569] The second device transmits fifth information to the third device, the fifth information being used to trigger the collection of the training dataset;

[0570] The fifth piece of information includes at least one of the following:

[0571] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0572] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0573] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0574] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0575] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0576] Instructions used to direct monitoring.

[0577] Optionally, if the second device is further configured to implement at least one of the control function and collection function for collecting the training dataset, the method further includes:

[0578] The second device transmits sixth information to the fourth device, the sixth information being used to indicate the collection of relevant information for the training dataset;

[0579] The sixth piece of information includes at least one of the following:

[0580] The first device supports or expects at least one candidate filtering rule;

[0581] The first device supports or expects at least one candidate sample classification category information;

[0582] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0583] The first device supports or expects at least one proportion of candidate samples;

[0584] The number of at least one candidate sample classifications supported or desired by the first device;

[0585] The first device supports or expects at least one proportion of candidate samples;

[0586] The first device supports or expects at least one candidate input data for the third AI unit;

[0587] At least one candidate output data of the third AI unit supported or desired by the first device;

[0588] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0589] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0590] Optionally, after the second device and the fourth device transmit the sixth information, the method further includes:

[0591] The second device transmits seventh information to the second device, the seventh information being used to trigger the collection of the training dataset;

[0592] The seventh piece of information includes at least one of the following:

[0593] The target filtering rule in the at least one candidate filtering rule;

[0594] The target classification sample category information in the at least one candidate sample classification category information;

[0595] The target threshold information in the at least one candidate threshold information;

[0596] The target sample ratio in the at least one candidate sample ratio;

[0597] The target sample classification number information in the at least one candidate sample classification number information;

[0598] At least the target sample proportion among the at least one candidate sample proportion;

[0599] The target input data is among at least one candidate input data of the third AI unit;

[0600] The target output data in at least one candidate output data of the third AI unit;

[0601] The sixteenth instruction indicates whether data filtering should be performed.

[0602] Optionally, the candidate objectives include at least one of the following:

[0603] Determine the evaluation result of the second AI unit;

[0604] Determine the virtual truth value associated with the second AI unit.

[0605] Optionally, if the second device is also used to implement the inference function of the third AI unit, the method further includes:

[0606] The second device sends tenth information to the third device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit has been filtered based on the training dataset of the third AI unit.

[0607] Optionally, the tenth information may further include at least one of the following:

[0608] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[0609] The third AI unit used by the second device.

[0610] Optionally, the second device obtains the target information of the second AI unit based on the first information and the reasoning of the first AI unit, including at least one of the following:

[0611] The second device uses the input and output data of the second AI unit in the first information as the input of the first AI unit to obtain the virtual evaluation result of the second AI unit;

[0612] The second device uses the input data of the second AI unit in the first information as the input of the first AI unit to obtain the virtual truth value of the second AI unit.

[0613] Reference Figure 6 This application also provides a data collection method, such as... Figure 6 As shown, the data collection method includes:

[0614] Step 601, the third device performs the target operation, which includes any one of the following:

[0615] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0616] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0617] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0618] Optionally, the eighth information may further include at least one of the following:

[0619] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0620] The training dataset is based on the instruction information determined by the first AI unit;

[0621] The measurement quantity associated with the target information;

[0622] The second device uses the first AI unit;

[0623] The second AI unit used by the first device.

[0624] Optionally, the tenth information may further include at least one of the following:

[0625] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[0626] The third AI unit used by the fourth device.

[0627] Optionally, the method further includes:

[0628] The third device transmits second information to the first device, the second information being used to indicate the collection of relevant information for the training dataset;

[0629] Wherein, the first device is used to implement the reasoning function of the first AI unit, and the second information includes at least one of the following:

[0630] First indication information, used to indicate at least one candidate first sample data that is supported or expected;

[0631] The second instruction information is used to indicate at least one candidate purpose that supports or is expected to collect data;

[0632] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0633] Optionally, after the third device sends the second information to the first device, the method further includes:

[0634] The third device transmits third information to the first device, the third information being used to trigger the collection of the training dataset, the third information including at least one of the following:

[0635] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0636] The fourth indication information is used to indicate the target first sample data among the at least one candidate first sample data;

[0637] The fifth instruction information is used to indicate the target objective among the at least one candidate objective;

[0638] Instructions used to direct monitoring.

[0639] Optionally, the method further includes:

[0640] The third device transmits fourth information to the second device, the fourth information being used to indicate the collection of relevant information for the training dataset;

[0641] The fourth piece of information includes at least one of the following:

[0642] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0643] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0644] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0645] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0646] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0647] The candidate third AI unit is used to filter the training dataset.

[0648] Optionally, after the third device sends the fourth information to the second device, the method further includes:

[0649] The third device transmits fifth information to the second device, and the fifth information is used to trigger the collection of the training dataset;

[0650] The fifth piece of information includes at least one of the following:

[0651] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0652] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0653] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0654] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0655] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0656] Instructions used to direct monitoring.

[0657] Optionally, the method further includes:

[0658] The third device and the fourth device transmit sixth information, which is used to indicate the collection of relevant information for the training dataset.

[0659] The sixth piece of information includes at least one of the following:

[0660] The first device supports or expects at least one candidate filtering rule;

[0661] The first device supports or expects at least one candidate sample classification category information;

[0662] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0663] The first device supports or expects at least one proportion of candidate samples;

[0664] The number of at least one candidate sample classifications supported or desired by the first device;

[0665] The first device supports or expects at least one proportion of candidate samples;

[0666] The first device supports or expects at least one candidate input data for the third AI unit;

[0667] At least one candidate output data of the third AI unit supported or desired by the first device;

[0668] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0669] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0670] Optionally, after the third device sends the sixth information to the fourth device, the method further includes:

[0671] The third device transmits seventh information to the second device, and the seventh information is used to trigger the collection of the training dataset;

[0672] The seventh piece of information includes at least one of the following:

[0673] The target filtering rule in the at least one candidate filtering rule;

[0674] The target classification sample category information in the at least one candidate sample classification category information;

[0675] The target threshold information in the at least one candidate threshold information;

[0676] The target sample ratio in the at least one candidate sample ratio;

[0677] The target sample classification number information in the at least one candidate sample classification number information;

[0678] At least the target sample proportion among the at least one candidate sample proportion;

[0679] The target input data is among at least one candidate input data of the third AI unit;

[0680] The target output data in at least one candidate output data of the third AI unit;

[0681] The sixteenth instruction indicates whether data filtering should be performed.

[0682] Optionally, the method further includes:

[0683] The third device sends a ninth message to the fourth device, the ninth message including the training dataset;

[0684] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0685] Optionally, the ninth information further includes at least one of the following:

[0686] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0687] The measurement quantity associated with the target information;

[0688] The second device uses the first AI unit;

[0689] The second AI unit used by the first device;

[0690] The third AI unit used by the fourth device.

[0691] Reference Figure 7 This application also provides a data collection method, such as... Figure 7 As shown, the data collection method includes:

[0692] Step 701: The fourth device receives the ninth information from the third device. The ninth information includes the training dataset of the second AI unit. The training dataset is determined based on the first AI unit and the first information sent by the first device to the second device.

[0693] Step 702: The fourth device filters the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset;

[0694] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0695] Optionally, the ninth information further includes at least one of the following:

[0696] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0697] The measurement quantity associated with the target information;

[0698] The second device uses the first AI unit;

[0699] The second AI unit used by the first device;

[0700] The third AI unit used by the fourth device.

[0701] Optionally, the method further includes:

[0702] The fourth device sends tenth information to the third device, the tenth information including the target training dataset.

[0703] Optionally, the tenth information may further include at least one of the following:

[0704] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[0705] The third AI unit used by the fourth device.

[0706] Optionally, the method further includes:

[0707] The fourth device transmits sixth information to the third device, the sixth information being used to indicate the collection of relevant information for the training dataset;

[0708] The sixth piece of information includes at least one of the following:

[0709] The first device supports or expects at least one candidate filtering rule;

[0710] The first device supports or expects at least one candidate sample classification category information;

[0711] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0712] The first device supports or expects at least one proportion of candidate samples;

[0713] The number of at least one candidate sample classifications supported or desired by the first device;

[0714] The first device supports or expects at least one proportion of candidate samples;

[0715] The first device supports or expects at least one candidate input data for the third AI unit;

[0716] At least one candidate output data of the third AI unit supported or desired by the first device;

[0717] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0718] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0719] Optionally, the method further includes:

[0720] The fourth device transmits seventh information to the third device, and the seventh information is used to trigger the collection of the training dataset;

[0721] The seventh piece of information includes at least one of the following:

[0722] The target filtering rule in the at least one candidate filtering rule;

[0723] The target classification sample category information in the at least one candidate sample classification category information;

[0724] The target threshold information in the at least one candidate threshold information;

[0725] The target sample ratio in the at least one candidate sample ratio;

[0726] The target sample classification number information in the at least one candidate sample classification number information;

[0727] The target sample ratio in the at least one candidate sample ratio;

[0728] The target input data is among at least one candidate input data of the third AI unit;

[0729] The target output data in at least one candidate output data of the third AI unit;

[0730] The sixteenth instruction indicates whether data filtering should be performed.

[0731] The data collection method provided in this application can be executed by a data collection device. This application uses an example of a data collection device executing the data collection method to illustrate the data collection device provided in this application.

[0732] This application provides a data collection device. As an example, the data collection device may be a communication device or a component within a communication device, such as a chip. The communication device may be a terminal, a network-side device, or a server, etc. Exemplarily, the terminal may include, but is not limited to, the type of terminal 11 listed above, and the network-side device may include, but is not limited to, the type of network-side device 12 listed above. This application does not impose specific limitations.

[0733] The data collection device includes a receiving module, a transmitting module, and a processing module. These modules can be implemented in software or hardware. When implemented in hardware, the processing module can be implemented by a processor. For example, the processor can include general-purpose processors, special-purpose processors, such as a Central Processing Unit (CPU), microprocessor, Digital Signal Processor (DSP), Artificial Intelligence (AI) processor, Graphics Processing Unit (GPU), Application Specific Integrated Circuit (ASIC), Network Processor (NP), Field Programmable Gate Array (FPGA), or other programmable logic devices, gate circuits, transistors, discrete hardware components, etc. The receiving and transmitting modules can be implemented by a communication interface, which can include one or more of the following: transceiver, pins, circuits, bus, radio frequency unit, etc.

[0734] For details, see Figure 8 The data collection device 800 includes:

[0735] The first transmission module 801 is used to send first information to the second device. The first information is used to instruct the second AI unit to obtain target information of the second AI unit based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[0736] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0737] Optionally, the first information includes at least one of the following:

[0738] The target is the first sample data;

[0739] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0740] The measurement quantity associated with the target information;

[0741] The second device uses the first AI unit;

[0742] The second AI unit used by the first device;

[0743] The target first sample data includes at least one of the input data of the second AI unit and the output data of the second AI unit.

[0744] Optionally, the first transmission module 801 is further configured to transmit second information to the third device, the second information being used to indicate information related to the collection of training datasets;

[0745] The third device is used to implement at least one of the control function and collection function for collecting training datasets, and the second information includes at least one of the following:

[0746] First indication information, used to indicate at least one candidate first sample data that is supported or expected;

[0747] The second instruction information is used to indicate at least one candidate purpose that supports or is expected to collect data;

[0748] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0749] Optionally, the first transmission module 801 is further configured to transmit third information to the third device, the third information being used to trigger the collection of the training dataset, the third information including at least one of the following:

[0750] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0751] The fourth indication information is used to indicate the target first sample data among the at least one candidate first sample data;

[0752] The fifth instruction information is used to indicate the target objective among the at least one candidate objective;

[0753] Instructions used to direct monitoring.

[0754] Optionally, the first transmission module 801 is further configured to transmit fourth information to the second device when the first device is further configured to implement at least one of the control function and the collection function for collecting training datasets, the fourth information being used to indicate information related to the collection of training datasets;

[0755] The fourth piece of information includes at least one of the following:

[0756] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0757] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0758] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0759] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0760] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0761] The candidate third AI unit is used to filter the training dataset.

[0762] Optionally, the first transmission module 801 is further configured to transmit fifth information to the second device, the fifth information being used to trigger the collection of the training dataset;

[0763] The fifth piece of information includes at least one of the following:

[0764] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0765] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0766] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0767] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0768] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0769] Instructions used to direct monitoring.

[0770] Optionally, the first transmission module 801 is further configured to transmit sixth information to the fourth device when the first device is further configured to implement at least one of the control function and the collection function for collecting training datasets, the sixth information being used to indicate information related to the collection of training datasets;

[0771] The sixth piece of information includes at least one of the following:

[0772] The first device supports or expects at least one candidate filtering rule;

[0773] The first device supports or expects at least one candidate sample classification category information;

[0774] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0775] The first device supports or expects at least one proportion of candidate samples;

[0776] The number of at least one candidate sample classifications supported or desired by the first device;

[0777] The first device supports or expects at least one proportion of candidate samples;

[0778] The first device supports or expects at least one candidate input data for the third AI unit;

[0779] At least one candidate output data of the third AI unit supported or desired by the first device;

[0780] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0781] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0782] Optionally, the first transmission module 801 is further configured to transmit seventh information to the fourth device, the seventh information being used to trigger the collection of the training dataset;

[0783] The seventh piece of information includes at least one of the following:

[0784] The target filtering rule in the at least one candidate filtering rule;

[0785] The target classification sample category information in the at least one candidate sample classification category information;

[0786] The target threshold information in the at least one candidate threshold information;

[0787] The target sample ratio in the at least one candidate sample ratio;

[0788] The target sample classification number information in the at least one candidate sample classification number information;

[0789] At least the target sample proportion among the at least one candidate sample proportion;

[0790] The target input data is among at least one candidate input data of the third AI unit;

[0791] The target output data in at least one candidate output data of the third AI unit;

[0792] The sixteenth instruction indicates whether data filtering should be performed.

[0793] Optionally, the first transmission module 801 is further configured to transmit sixth information to the third device when the first device is also configured to implement the inference function of the third AI unit, the sixth information being used to indicate information for collecting relevant information for training datasets;

[0794] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the sixth information includes at least one of the following:

[0795] The first device supports or expects at least one candidate filtering rule;

[0796] The first device supports or expects at least one candidate sample classification category information;

[0797] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0798] The first device supports or expects at least one proportion of candidate samples;

[0799] The number of at least one candidate sample classifications supported or desired by the first device;

[0800] The first device supports or expects at least one proportion of candidate samples;

[0801] The first device supports or expects at least one candidate input data for the third AI unit;

[0802] At least one candidate output data of the third AI unit supported or desired by the first device;

[0803] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering.

[0804] Optionally, the first transmission module 801 is further configured to transmit seventh information to the third device, the seventh information being used to trigger the collection of the training dataset;

[0805] The seventh piece of information includes at least one of the following:

[0806] The target filtering rule in the at least one candidate filtering rule;

[0807] The target classification sample category information in the at least one candidate sample classification category information;

[0808] The target threshold information in the at least one candidate threshold information;

[0809] The target sample ratio in the at least one candidate sample ratio;

[0810] The target sample classification number information in the at least one candidate sample classification number information;

[0811] The target sample ratio in the at least one candidate sample ratio;

[0812] The target input data is among at least one candidate input data of the third AI unit;

[0813] The target output data in at least one candidate output data of the third AI unit;

[0814] The sixteenth instruction indicates whether data filtering should be performed.

[0815] Optionally, the candidate objectives include at least one of the following:

[0816] Determine the evaluation result of the second AI unit;

[0817] Determine the virtual truth value associated with the second AI unit.

[0818] For details, see Figure 9 The data collection device 900 includes

[0819] The second transmission module 901 is used to receive first information from the first device;

[0820] The first processing module 902 is used to obtain the target information of the second AI unit based on the first information and the reasoning of the first AI unit, wherein the target information includes virtual truth value or virtual evaluation result;

[0821] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0822] Optionally, the first information includes at least one of the following:

[0823] The target is the first sample data;

[0824] The sample identifier corresponding to the first sample data of the target or the sample set identifier corresponding to the first sample data of the target;

[0825] The measurement quantity associated with the target information;

[0826] The second device uses the first AI unit;

[0827] The second AI unit used by the first device;

[0828] The target first sample data includes at least one of the input data of the second AI unit and the output data of the second AI unit.

[0829] Optionally, the second transmission module 901 is further configured to send eighth information to the third device, the eighth information including the training dataset;

[0830] The third device is used to implement at least one of the control function and collection function for collecting training datasets.

[0831] Optionally, the eighth information may further include at least one of the following:

[0832] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0833] The training dataset is based on the instruction information determined by the first AI unit;

[0834] The measurement quantity associated with the target information;

[0835] The second device uses the first AI unit;

[0836] The second AI unit used by the first device.

[0837] Optionally, the second transmission module 901 is further configured to send ninth information to the fourth device, the ninth information including the training dataset;

[0838] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0839] Optionally, the ninth information further includes at least one of the following:

[0840] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0841] The measurement quantity associated with the target information;

[0842] The second device uses the first AI unit;

[0843] The second AI unit used by the first device;

[0844] The third AI unit used by the fourth device.

[0845] Optionally, the second transmission module 901 is further configured to transmit sixth information to the third device when the second device is also configured to implement the inference function of the third AI unit, the sixth information being used to indicate the collection of relevant information for training datasets;

[0846] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the sixth information includes at least one of the following:

[0847] The first device supports or expects at least one candidate filtering rule;

[0848] The first device supports or expects at least one candidate sample classification category information;

[0849] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0850] The first device supports or expects at least one proportion of candidate samples;

[0851] The number of at least one candidate sample classifications supported or desired by the first device;

[0852] The first device supports or expects at least one proportion of candidate samples;

[0853] The first device supports or expects at least one candidate input data for the third AI unit;

[0854] At least one candidate output data of the third AI unit supported or desired by the first device;

[0855] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering.

[0856] Optionally, the second transmission module 901 is further configured to transmit seventh information to the third device, the seventh information being used to trigger the collection of the training dataset;

[0857] The seventh piece of information includes at least one of the following:

[0858] The target filtering rule in the at least one candidate filtering rule;

[0859] The target classification sample category information in the at least one candidate sample classification category information;

[0860] The target threshold information in the at least one candidate threshold information;

[0861] The target sample ratio in the at least one candidate sample ratio;

[0862] The target sample classification number information in the at least one candidate sample classification number information;

[0863] The target sample ratio in the at least one candidate sample ratio;

[0864] The target input data is among at least one candidate input data of the third AI unit;

[0865] The target output data in at least one candidate output data of the third AI unit;

[0866] The sixteenth instruction indicates whether data filtering should be performed.

[0867] Optionally, the second transmission module 901 is further configured to transmit fourth information to the third device, the fourth information being used to indicate information for collecting relevant information for training datasets;

[0868] The fourth piece of information includes at least one of the following:

[0869] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0870] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0871] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0872] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0873] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0874] The third device is used to implement at least one of the control function and collection function for collecting the training dataset, and the candidate third AI unit is used to filter the training dataset.

[0875] Optionally, the second transmission module 901 is further configured to transmit fifth information to the third device, the fifth information being used to trigger the collection of the training dataset;

[0876] The fifth piece of information includes at least one of the following:

[0877] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0878] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0879] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0880] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0881] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0882] Instructions used to direct monitoring.

[0883] Optionally, the second transmission module 901 is further configured to transmit sixth information to the fourth device when the second device is further configured to implement at least one of the control function and collection function for collecting training datasets, the sixth information being used to indicate information related to the collection of training datasets;

[0884] The sixth piece of information includes at least one of the following:

[0885] The first device supports or expects at least one candidate filtering rule;

[0886] The first device supports or expects at least one candidate sample classification category information;

[0887] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0888] The first device supports or expects at least one proportion of candidate samples;

[0889] The number of at least one candidate sample classifications supported or desired by the first device;

[0890] The first device supports or expects at least one proportion of candidate samples;

[0891] The first device supports or expects at least one candidate input data for the third AI unit;

[0892] At least one candidate output data of the third AI unit supported or desired by the first device;

[0893] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0894] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0895] Optionally, the second transmission module 901 is further configured to transmit seventh information to the second device, the seventh information being used to trigger the collection of the training dataset;

[0896] The seventh piece of information includes at least one of the following:

[0897] The target filtering rule in the at least one candidate filtering rule;

[0898] The target classification sample category information in the at least one candidate sample classification category information;

[0899] The target threshold information in the at least one candidate threshold information;

[0900] The target sample ratio in the at least one candidate sample ratio;

[0901] The target sample classification number information in the at least one candidate sample classification number information;

[0902] At least the target sample proportion among the at least one candidate sample proportion;

[0903] The target input data is among at least one candidate input data of the third AI unit;

[0904] The target output data in at least one candidate output data of the third AI unit;

[0905] The sixteenth instruction indicates whether data filtering should be performed.

[0906] Optionally, the candidate objectives include at least one of the following:

[0907] Determine the evaluation result of the second AI unit;

[0908] Determine the virtual truth value associated with the second AI unit.

[0909] Optionally, the second transmission module 901 is further configured to send tenth information to the third device when the second device is also configured to implement the inference function of the third AI unit. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0910] Optionally, the tenth information may further include at least one of the following:

[0911] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[0912] The third AI unit used by the second device.

[0913] Optionally, the first processing module 902 is specifically configured to perform at least one of the following:

[0914] The input and output data of the second AI unit in the first information are used as the input of the first AI unit to obtain the virtual evaluation result of the second AI unit;

[0915] The input data of the second AI unit in the first information is used as the input of the first AI unit to obtain the virtual truth value of the second AI unit.

[0916] For details, see Figure 10 The data collection device 1000 includes:

[0917] The third transmission module 1001 is used to perform a target operation, the target operation including any one of the following:

[0918] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[0919] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[0920] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[0921] Optionally, the eighth information may further include at least one of the following:

[0922] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0923] The training dataset is based on the instruction information determined by the first AI unit;

[0924] The measurement quantity associated with the target information;

[0925] The second device uses the first AI unit;

[0926] The second AI unit used by the first device.

[0927] Optionally, the tenth information may further include at least one of the following:

[0928] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[0929] The third AI unit used by the fourth device.

[0930] Optionally, the third transmission module 1001 is further configured to transmit second information to the first device, the second information being used to indicate information related to the collection of training datasets;

[0931] Wherein, the first device is used to implement the reasoning function of the first AI unit, and the second information includes at least one of the following:

[0932] First indication information, used to indicate at least one candidate first sample data that is supported or expected;

[0933] The second instruction information is used to indicate at least one candidate purpose that supports or is expected to collect data;

[0934] The third indication information is used to indicate whether target information is generated based on the first AI unit.

[0935] Optionally, the third transmission module 1001 is further configured to transmit third information to the first device, the third information being used to trigger the collection of the training dataset, the third information including at least one of the following:

[0936] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0937] The fourth indication information is used to indicate the target first sample data among the at least one candidate first sample data;

[0938] The fifth instruction information is used to indicate the target objective among the at least one candidate objective;

[0939] Instructions used to direct monitoring.

[0940] Optionally, the third transmission module 1001 is further configured to transmit fourth information to the second device, the fourth information being used to indicate information for collecting relevant information for training datasets;

[0941] The fourth piece of information includes at least one of the following:

[0942] The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit;

[0943] The seventh indication information is used to indicate at least one candidate second sample data that the first device supports or expects, the candidate second sample data including at least one of the input data of the first AI unit and the output data of the first AI unit:

[0944] The eighth indication information is used to indicate at least one candidate second AI unit supported or desired by the first device;

[0945] The ninth indication information is used to indicate at least one candidate third AI unit supported or desired by the first device;

[0946] The tenth instruction information is used to indicate at least one candidate purpose for which the first device supports or expects data collection;

[0947] The candidate third AI unit is used to filter the training dataset.

[0948] Optionally, the third transmission module 1001 is further configured to transmit fifth information to the second device, the fifth information being used to trigger the collection of the training dataset;

[0949] The fifth piece of information includes at least one of the following:

[0950] The third indication information is used to indicate whether target information is generated based on the first AI unit;

[0951] The eleventh indication information is used to indicate the target second sample data in at least one candidate second sample data:

[0952] The twelfth instruction information is used to indicate the target second AI unit among the at least one candidate second AI units;

[0953] The thirteenth instruction information is used to indicate the target third AI unit among the at least one candidate third AI units;

[0954] The fourteenth instruction information is used to indicate a target objective among the at least one candidate objective;

[0955] Instructions used to direct monitoring.

[0956] Optionally, the third transmission module 1001 is further configured to transmit sixth information to the fourth device, the sixth information being used to indicate information for collecting relevant information for training datasets;

[0957] The sixth piece of information includes at least one of the following:

[0958] The first device supports or expects at least one candidate filtering rule;

[0959] The first device supports or expects at least one candidate sample classification category information;

[0960] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[0961] The first device supports or expects at least one proportion of candidate samples;

[0962] The number of at least one candidate sample classifications supported or desired by the first device;

[0963] The first device supports or expects at least one proportion of candidate samples;

[0964] The first device supports or expects at least one candidate input data for the third AI unit;

[0965] At least one candidate output data of the third AI unit supported or desired by the first device;

[0966] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[0967] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0968] Optionally, the third transmission module 1001 is further configured to transmit seventh information to the second device, the seventh information being used to trigger the collection of the training dataset;

[0969] The seventh piece of information includes at least one of the following:

[0970] The target filtering rule in the at least one candidate filtering rule;

[0971] The target classification sample category information in the at least one candidate sample classification category information;

[0972] The target threshold information in the at least one candidate threshold information;

[0973] The target sample ratio in the at least one candidate sample ratio;

[0974] The target sample classification number information in the at least one candidate sample classification number information;

[0975] At least the target sample proportion among the at least one candidate sample proportion;

[0976] The target input data is among at least one candidate input data of the third AI unit;

[0977] The target output data in at least one candidate output data of the third AI unit;

[0978] The sixteenth instruction indicates whether data filtering should be performed.

[0979] Optionally, the third transmission module 1001 is further configured to send ninth information to the fourth device, the ninth information including the training dataset;

[0980] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[0981] Optionally, the ninth information further includes at least one of the following:

[0982] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0983] The measurement quantity associated with the target information;

[0984] The second device uses the first AI unit;

[0985] The second AI unit used by the first device;

[0986] The third AI unit used by the fourth device.

[0987] For details, see Figure 11 The data collection device 1100 includes:

[0988] The fourth transmission module 1101 is used to receive ninth information from the third device, the ninth information including the training dataset of the second AI unit, the training dataset being determined based on the first information sent from the first AI unit and the first device to the second device;

[0989] The second processing module 1102 is used to filter the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset.

[0990] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[0991] Optionally, the ninth information further includes at least one of the following:

[0992] The sample identifier associated with the training dataset or the sample set identifier associated with the training dataset;

[0993] The measurement quantity associated with the target information;

[0994] The second device uses the first AI unit;

[0995] The second AI unit used by the first device;

[0996] The third AI unit used by the fourth device.

[0997] Optionally, the fourth transmission module 1101 is further configured to send tenth information to the third device, the tenth information including the target training dataset.

[0998] Optionally, the tenth information may further include at least one of the following:

[0999] The sample identifier associated with the target training dataset or the sample set identifier associated with the target training dataset;

[1000] The third AI unit used by the fourth device.

[1001] Optionally, the fourth transmission module 1101 is further configured to transmit sixth information to the third device, the sixth information being used to indicate information related to the collection of training datasets;

[1002] The sixth piece of information includes at least one of the following:

[1003] The first device supports or expects at least one candidate filtering rule;

[1004] The first device supports or expects at least one candidate sample classification category information;

[1005] The first device supports or expects at least one candidate threshold information, which is used to determine the sample classification category;

[1006] The first device supports or expects at least one proportion of candidate samples;

[1007] The number of at least one candidate sample classifications supported or desired by the first device;

[1008] The first device supports or expects at least one proportion of candidate samples;

[1009] The first device supports or expects at least one candidate input data for the third AI unit;

[1010] At least one candidate output data of the third AI unit supported or desired by the first device;

[1011] The fifteenth instruction information is used to indicate whether the first device supports or expects to perform data filtering;

[1012] The fourth device is used to implement the reasoning function of the third AI unit, which is used to filter the training dataset.

[1013] Optionally, the fourth transmission module 1101 is further configured to transmit seventh information to the third device, the seventh information being used to trigger the collection of the training dataset;

[1014] The seventh piece of information includes at least one of the following:

[1015] The target filtering rule in the at least one candidate filtering rule;

[1016] The target classification sample category information in the at least one candidate sample classification category information;

[1017] The target threshold information in the at least one candidate threshold information;

[1018] The target sample ratio in the at least one candidate sample ratio;

[1019] The target sample classification number information in the at least one candidate sample classification number information;

[1020] The target sample ratio in the at least one candidate sample ratio;

[1021] The target input data is among at least one candidate input data of the third AI unit;

[1022] The target output data in at least one candidate output data of the third AI unit;

[1023] The sixteenth instruction indicates whether data filtering should be performed.

[1024] like Figure 12 As shown, this application embodiment also provides a communication device 1200, including a processor 1201 and a memory 1202. The memory 1202 stores a program or instructions that can run on the processor 1201. When the program or instructions are executed by the processor 1201, they implement the various steps of the above-described data collection method embodiment and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[1025] This application embodiment also provides a terminal, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement, for example... Figure 2 , 5 The steps in the method embodiments shown in 6 or 7. This terminal embodiment corresponds to the above-described terminal-side method embodiments. All implementation processes and methods of the above-described method embodiments can be applied to this terminal embodiment and achieve the same technical effect. The terminal can be... Figure 8 , 9 The data collection device shown in 10 or 11. Specifically, Figure 13 A schematic diagram of the hardware structure of a terminal to implement an embodiment of this application.

[1026] The terminal 1300 includes, but is not limited to, at least some of the following components: radio frequency unit 1301, network module 1302, audio output unit 1303, input unit 1304, sensor 1305, display unit 1306, user input unit 1307, interface unit 1308, memory 1309, and processor 1310.

[1027] Those skilled in the art will understand that the terminal 1300 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 1310 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 13 The terminal structure shown does not constitute a limitation on the terminal. The terminal may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[1028] It should be understood that, in this embodiment, the input unit 1304 may include a graphics processor 13041 and a microphone 13042. The graphics processor 13041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 1306 may include a display panel 13061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like. The user input unit 1307 includes a touch panel 13071 and at least one of other input devices 13072. The touch panel 13071 is also called a touch screen. The touch panel 13071 may include a touch detection device and a touch controller. Other input devices 13072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here.

[1029] In this embodiment, after receiving downlink data from the network-side device, the radio frequency unit 1301 can transmit it to the processor 1310 for processing; in addition, the radio frequency unit 1301 can send uplink data to the network-side device. Typically, the radio frequency unit 1301 includes, but is not limited to, antennas, amplifiers, transceivers, couplers, low-noise amplifiers, duplexers, etc.

[1030] The memory 1309 can be used to store software programs or instructions, as well as various data. The memory 1309 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 1309 may include volatile memory or non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 1309 in this embodiment includes, but is not limited to, these and any other suitable types of memory.

[1031] Processor 1310 may include one or more processing units; optionally, processor 1310 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 1310.

[1032] When the terminal is a first device, the radio frequency unit 1301 is used to send first information to a second device. The first information is used to instruct the second AI unit to obtain target information of the second AI unit based on the reasoning of the first AI unit. The target information includes virtual truth value or virtual evaluation result.

[1033] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[1034] When the terminal is a second device, the radio frequency unit 1301 is used to receive first information from the first device;

[1035] Processor 1310 is configured to obtain target information of the second AI unit based on the first information and reasoning of the first AI unit, wherein the target information includes virtual truth value or virtual evaluation result;

[1036] Wherein, the target information is used to determine the training dataset of the second AI unit, the second AI unit is used to implement use case functions, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[1037] When the terminal is a third device, the radio frequency unit 1301 is used to perform a target operation, the target operation including any one of the following:

[1038] Receive eighth information from the second device, the eighth information including the training dataset of the second AI unit;

[1039] The tenth information is received from the fourth device. The tenth information includes a target training dataset, which is a dataset after the training dataset of the second AI unit is filtered based on the third AI unit.

[1040] Wherein, the training dataset of the second AI unit is determined based on the first information sent by the first AI unit and the first device to the second device, the first AI unit is used to infer the target information of the second AI unit, the target information includes virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, the second device is used to implement the inference function of the first AI unit, the third device is used to implement at least one of the control function and collection function of training dataset collection, and the fourth device is used to implement the inference function of the third AI unit.

[1041] When the terminal is a fourth device, the radio frequency unit 1301 receives ninth information from the third device. The ninth information includes the training dataset of the second AI unit, which is determined based on the first information sent from the first AI unit and the first device to the second device.

[1042] The processor 1310 is used to filter the training dataset of the second AI unit based on the third AI unit to obtain the target training dataset.

[1043] Wherein, the first AI unit is used to infer the target information of the second AI unit, the target information including virtual truth value or virtual evaluation result, the second AI unit is used to implement use case function, the first device is used to implement the inference function of the second AI unit, and the second device is used to implement the inference function of the first AI unit.

[1044] It is understood that the implementation process of each implementation method mentioned in this embodiment can refer to the relevant descriptions of the method embodiments on the first device, second device, third device or fourth device side, and achieve the same or corresponding technical effects. In order to avoid repetition, it will not be described again here.

[1045] This application embodiment also provides a network-side device, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement, for example... Figure 2 , Figure 5 , Figure 6 or Figure 7 The steps of the method embodiment shown are illustrated. This network-side device embodiment corresponds to the above-described network-side device method embodiment. All implementation processes and methods of the above-described method embodiments can be applied to this network-side device embodiment and can achieve the same technical effect.

[1046] Specifically, embodiments of this application also provide a network-side device, which may be... Figure 8 , Figure 9 , Figure 10 or Figure 11 The data collection device shown. (For example...) Figure 14 As shown, the network-side device 1400 includes: an antenna 1401, a radio frequency (RF) device 1402, a baseband device 1403, a processor 1404, and a memory 1405. The antenna 1401 is connected to the RF device 1402. In the uplink direction, the RF device 1402 receives information through the antenna 1401 and transmits the received information to the baseband device 1403 for processing. In the downlink direction, the baseband device 1403 processes the information to be transmitted and sends it to the RF device 1402. The RF device 1402 processes the received information and transmits it through the antenna 1401.

[1047] The method executed by the network-side device in the above embodiments can be implemented in the baseband device 1403, which includes a baseband processor.

[1048] The baseband device 1403 may include, for example, at least one baseband board on which multiple chips are disposed, such as... Figure 14 As shown, one of the chips is, for example, a baseband processor, which is connected to the memory 1405 via a bus interface to call the program in the memory 1405 and execute the network-side device operations shown in the above method embodiment.

[1049] The network-side device may also include a network interface 1406, such as a Common Public Radio Interface (CPRI).

[1050] Specifically, the network-side device 1400 in this application embodiment further includes: instructions or programs stored in memory 1405 and executable on processor 1404, wherein processor 1404 calls the instructions or programs in memory 1405 to execute. Figure 8 , Figure 9 , Figure 10 or Figure 11 The methods executed by each module shown achieve the same technical effect, and to avoid repetition, they will not be described in detail here.

[1051] Specifically, embodiments of this application also provide a network-side device. For example... Figure 15 As shown, the network-side device 1500 includes: a processor 1501, a network interface 1502, and a memory 1503. This network-side device can be... Figure 8 , Figure 9 , Figure 10 or Figure 11 The data collection device shown. The network interface 1502 is, for example, a common public radio interface (CPRI).

[1052] Specifically, the network-side device 1500 in this application embodiment further includes: instructions or programs stored in memory 1503 and executable on processor 1501, wherein processor 1501 calls the instructions or programs in memory 1503 to execute. Figure 8 , Figure 9 , Figure 10 or Figure 11 The methods executed by each module shown achieve the same technical effect, and to avoid repetition, they will not be described in detail here.

[1053] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described data collection method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[1054] The processor mentioned above is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk. In some examples, the readable storage medium may be a non-transient readable storage medium.

[1055] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above data collection method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[1056] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[1057] This application also provides a computer program / program product, which includes computer instructions. The computer program / program product is executed by at least one processor to implement the various processes of the above-described data collection method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[1058] This application also provides a wireless communication system, including: a first device, a second device, a third device, and a fourth device. The first device can be used to perform the steps of the data collection method on the first device side as described above, the second device can be used to perform the steps of the data collection method on the second device side as described above, the third device can be used to perform the steps of the data collection method on the third device side as described above, and the fourth device can be used to perform the steps of the data collection method on the fourth device side as described above.

[1059] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[1060] From the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of computer software products plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, magnetic disk, optical disk, etc.) and includes several instructions to cause the terminal or network-side device to execute the methods described in the various embodiments of this application.

[1061] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other implementations under the guidance of this application without departing from the spirit and scope of the claims. All of these implementations are within the protection scope of this application.

Claims

1. A data collection method, characterized by, Comprise: The first device sends first information to the second device, the first information is used to indicate that the target information of the second AI unit is obtained based on the inference of the first AI unit, and the target information includes virtual true value or virtual evaluation result; Wherein, the target information is used to determine the training data set of the second AI unit, the second AI unit is used to realize the use case function, the first device is used to realize the inference function of the second AI unit, and the second device is used to realize the inference function of the first AI unit.

2. The method of claim 1, wherein, The first information includes at least one of the following: Target first sample data; Sample identifier corresponding to the target first sample data or sample set identifier corresponding to the target first sample data; The measurement quantity associated with the target information; The first AI unit used by the second device; The second AI unit used by the first device; Wherein, the target first sample data includes at least one of the input data of the second AI unit and the output data of the second AI unit.

3. The method according to claim 1 or 2, characterized in that, Before the first device sends the first information to the second device, the method further comprises: The first device transmits second information to the third device, the second information is used to indicate information related to training data set collection; Wherein, the third device is used to realize at least one of the control function and the collection function of the training data set collection, and the second information includes at least one of the following: First indication information, used to indicate at least one candidate first sample data supported or expected; Second indication information, used to indicate at least one candidate purpose supported or expected data collection; Third indication information, used to indicate whether to generate target information based on the first AI unit.

4. The method of claim 3, wherein, After the first device transmits the second information to the third device, the method further comprises: The first device transmits third information to the third device, the third information is used to trigger the collection of the training data set, and the third information includes at least one of the following: Third indication information, used to indicate whether to generate target information based on the first AI unit; Fourth indication information, used to indicate target first sample data in the at least one candidate first sample data; Fifth indication information, used to indicate target purpose in the at least one candidate purpose; Indication information for monitoring.

5. The method according to claim 1 or 2, characterized in that, In the case where the first device is also used to realize at least one of the control function and the collection function of the training data set collection, the method further comprises: The first device transmits fourth information to the second device, the fourth information is used to indicate information related to training data set collection; Wherein, the fourth information includes at least one of the following: Sixth indication information, used to indicate whether the first device supports or expects to generate target information based on the first AI unit; Seventh indication information, used to indicate at least one candidate second sample data supported or expected by the first device, the candidate second sample data includes at least one of the input data of the first AI unit and the output data of the first AI unit: Eighth indication information, used to indicate at least one candidate second AI unit supported or expected by the first device; a ninth indication information, used for indicating at least one candidate third AI unit supported or expected by the first device; a tenth indication information, used for indicating at least one candidate purpose of data collection supported or expected by the first device; wherein the candidate third AI unit is used for data screening on the training data set.

6. The method of claim 5, wherein, After the first device and the second device transmit the fourth information, the method further comprises: the first device and the second device transmit fifth information, the fifth information being used for triggering the collection of the training data set; wherein the fifth information comprises at least one of: third indication information, used for indicating whether target information is generated based on the first AI unit; eleventh indication information, used for indicating target second sample data in at least one candidate second sample data: twelfth indication information, used for indicating a target second AI unit in the at least one candidate second AI unit; thirteenth indication information, used for indicating a target third AI unit in the at least one candidate third AI unit; fourteenth indication information, used for indicating a target purpose in the at least one candidate purpose; indication information used for indicating monitoring.

7. The method of claim 1, 2, 5, or 6, wherein, In a case where the first device is further used for implementing at least one of a control function and a collection function of a training data set collection, the method further comprises: the first device and a fourth device transmit sixth information, the sixth information being used for indicating information related to training data set collection; wherein the sixth information comprises at least one of: at least one candidate screening rule supported or expected by the first device; at least one candidate sample classification category information supported or expected by the first device; at least one candidate threshold information supported or expected by the first device, the candidate threshold information being used for determining a sample classification category; at least one candidate sample proportion supported or expected by the first device; at least one candidate sample classification number supported or expected by the first device; at least one candidate sample proportion supported or expected by the first device; at least one candidate input data of a third AI unit supported or expected by the first device; at least one candidate output data of a third AI unit supported or expected by the first device; fifteenth indication information, used for indicating whether data screening is supported or expected by the first device; wherein the fourth device is used for implementing an inference function of a third AI unit, the third AI unit being used for data screening on the training data set.

8. The method of claim 7, wherein, After the first device and the fourth device transmit the sixth information, the method further comprises: the first device and the fourth device transmit seventh information, the seventh information being used for triggering the collection of the training data set; wherein the seventh information comprises at least one of: a target screening rule in the at least one candidate screening rule; target classification sample category information in the at least one candidate sample classification category information; target threshold information in the at least one candidate threshold information; a target sample proportion in the at least one candidate sample proportion; Target sample classification number information in the at least one candidate sample classification number information; At least target sample proportion in the at least one candidate sample proportion; Target input data in at least one candidate input data of the third AI unit; Target output data in at least one candidate output data of the third AI unit; Sixteenth indication information for indicating whether to perform data screening.

9. The method of claim 1 or 2, wherein, In the case that the first device is also used to implement the inference function of the third AI unit, the method further comprises: The first device transmits sixth information with the third device, the sixth information being used to indicate information related to training data set collection; The third device is used to implement at least one of the control function and the collection function of the training data set collection, and the sixth information comprises at least one of the following: At least one candidate screening rule supported or expected by the first device; At least one candidate sample classification category information supported or expected by the first device; At least one candidate threshold information supported or expected by the first device, the candidate threshold information being used to determine a sample classification category; At least one candidate sample proportion supported or expected by the first device; At least one candidate sample classification number supported or expected by the first device; At least one candidate sample proportion supported or expected by the first device; At least one candidate input data of the third AI unit supported or expected by the first device; At least one candidate output data of the third AI unit supported or expected by the first device; Fifteenth indication information for indicating whether the first device supports or expects to perform data screening.

10. The method of claim 9, wherein, After the first device transmits the sixth information with the third device, the method further comprises: The first device transmits seventh information with the third device, the seventh information being used to trigger the collection of the training data set; The seventh information comprises at least one of the following: Target screening rule in the at least one candidate screening rule; Target classification sample category information in the at least one candidate sample classification category information; Target threshold information in the at least one candidate threshold information; Target sample proportion in the at least one candidate sample proportion; Target sample classification number information in the at least one candidate sample classification number information; Target sample proportion in the at least one candidate sample proportion; Target input data in at least one candidate input data of the third AI unit; Target output data in at least one candidate output data of the third AI unit; Sixteenth indication information for indicating whether to perform data screening.

11. The method according to any one of claims 3 to 6, characterized in that, The candidate purposes comprise at least one of the following: Determine the evaluation result of the second AI unit; Determine the virtual true value associated with the second AI unit.

12. A data collection method characterized by, Comprise: The second device receives first information from the first device; The second device obtains target information of the second AI unit based on the first information and the inference of the first AI unit, the target information comprising a virtual true value or a virtual evaluation result; The target information is used to determine a training data set of a second AI unit, the second AI unit is used to implement a use case function, the first device is used to implement an inference function of the second AI unit, and the second device is used to implement an inference function of the first AI unit.

13. The method of claim 12, wherein, The first information includes at least one of the following: Target first sample data; A sample identifier corresponding to the target first sample data or a sample set identifier corresponding to the target first sample data; A measurement quantity associated with the target information; A first AI unit used by the second device; A second AI unit used by the first device; The target first sample data includes at least one of input data of the second AI unit and output data of the second AI unit.

14. The method according to claim 12 or 13, characterized in that, After the second device obtains the target information of the second AI unit based on the first information and the inference of the first AI unit, the method further includes: The second device sends eighth information to a third device, and the eighth information includes the training data set; The third device is used to implement at least one of a control function and a collection function of the training data set collection.

15. The method of claim 14, wherein, The eighth information further includes at least one of the following: A sample identifier associated with the training data set or a sample set identifier associated with the training data set; Indicative information determined based on the first AI unit; A measurement quantity associated with the target information; A first AI unit used by the second device; A second AI unit used by the first device.

16. The method of claim 12 or 13, wherein, After the second device obtains the target information of the second AI unit based on the first information and the inference of the first AI unit, the method further includes: The second device sends ninth information to a fourth device, and the ninth information includes the training data set; The fourth device is used to implement an inference function of a third AI unit, and the third AI unit is used to perform data screening on the training data set.

17. The method of claim 16, wherein, The ninth information further includes at least one of the following: A sample identifier associated with the training data set or a sample set identifier associated with the training data set; A measurement quantity associated with the target information; A first AI unit used by the second device; A second AI unit used by the first device; A third AI unit used by the fourth device.

18. The method of claim 12, 13, 16, or 17, wherein, In the case where the second device is also used to implement an inference function of a third AI unit, the method further includes: The second device transmits sixth information to a third device, and the sixth information is used to indicate information related to training data set collection; The third device is used to implement at least one of a control function and a collection function of the training data set collection, and the sixth information includes at least one of the following: At least one candidate screening rule supported or expected by the first device; At least one candidate sample classification category information supported or expected by the first device; At least one candidate threshold information supported or expected by the first device, the candidate threshold information being used to determine a sample classification category; At least one candidate sample proportion supported or expected by the first device; At least one candidate sample classification number supported or expected by the first device; at least one candidate sample proportion supported or expected by the first device; at least one candidate input data of a third AI unit supported or expected by the first device; at least one candidate output data of the third AI unit supported or expected by the first device; fifteenth indication information for indicating whether data screening is supported or expected by the first device.

19. The method of claim 18, wherein, After the second device and the third device transmit the sixth information, the method further comprises: The second device and the third device transmit seventh information, and the seventh information is used to trigger the collection of the training data set; The seventh information comprises at least one of: The target screening rule in the at least one candidate screening rule; The target classification sample category information in the at least one candidate sample classification category information; The target threshold information in the at least one candidate threshold information; The target sample proportion in the at least one candidate sample proportion; The target sample classification number information in the at least one candidate sample classification number information; The target sample proportion in the at least one candidate sample proportion; The target input data in the at least one candidate input data of the third AI unit; The target output data in the at least one candidate output data of the third AI unit; The sixteenth indication information is used to indicate whether data screening is performed.

20. The method of claim 12 or 13, wherein, The method further comprises: The second device and the third device transmit fourth information, and the fourth information is used to indicate information related to the collection of the training data set; The fourth information comprises at least one of: The sixth indication information is used to indicate whether the first device supports or expects to generate target information based on the first AI unit; The seventh indication information is used to indicate at least one candidate second sample data supported or expected by the first device, and the candidate second sample data comprises at least one of input data of the first AI unit and output data of the first AI unit: The eighth indication information is used to indicate at least one candidate second AI unit supported or expected by the first device; The ninth indication information is used to indicate at least one candidate third AI unit supported or expected by the first device; The tenth indication information is used to indicate at least one candidate purpose of data collection supported or expected by the first device; The third device is used to implement at least one of a control function and a collection function of the training data set collection, and the candidate third AI unit is used to perform data screening on the training data set.

21. The method of claim 20, wherein, After the second device and the third device transmit the fourth information, the method further comprises: The second device and the third device transmit fifth information, and the fifth information is used to trigger the collection of the training data set; The fifth information comprises at least one of: The third indication information is used to indicate whether target information is generated based on the first AI unit; The eleventh indication information is used to indicate target second sample data in the at least one candidate second sample data: The twelfth indication information is used to indicate a target second AI unit in the at least one candidate second AI unit; Thirteenth indication information, used for indicating a target third AI unit in the at least one candidate third AI unit; Fourteenth indication information, used for indicating a target destination in the at least one candidate destination; Indication information used for indicating monitoring.

22. The method of claim 12 or 13, wherein, In a case where the second device is further used for implementing at least one of the control function and the collection function of the training data set collection, the method further includes: The second device transmits sixth information to the fourth device, the sixth information being used for indicating information related to training data set collection; The sixth information includes at least one of the following: At least one candidate screening rule supported or expected by the first device; At least one candidate sample classification category information supported or expected by the first device; At least one candidate threshold information supported or expected by the first device, the candidate threshold information being used for determining a sample classification category; At least one candidate sample proportion supported or expected by the first device; At least one candidate sample classification number supported or expected by the first device; At least one candidate sample proportion supported or expected by the first device; At least one candidate input data of the third AI unit supported or expected by the first device; At least one candidate output data of the third AI unit supported or expected by the first device; Fifteenth indication information, used for indicating whether the first device supports or expects data screening; The fourth device is used for implementing an inference function of a third AI unit, and the third AI unit is used for data screening on the training data set.

23. The method of claim 22, wherein, After the second device transmits the sixth information to the fourth device, the method further includes: The second device transmits seventh information to the second device, the seventh information being used for triggering collection of the training data set; The seventh information includes at least one of the following: A target screening rule in the at least one candidate screening rule; Target classification sample category information in the at least one candidate sample classification category information; Target threshold information in the at least one candidate threshold information; A target sample proportion in the at least one candidate sample proportion; Target sample classification number information in the at least one candidate sample classification number information; At least a target sample proportion in the at least one candidate sample proportion; Target input data in the at least one candidate input data of the third AI unit; Target output data in the at least one candidate output data of the third AI unit; Sixteenth indication information, used for indicating whether data screening is performed.

24. The method of claim 20 or 21, wherein, The candidate destination includes at least one of the following: Determining an evaluation result of the second AI unit; Determining a virtual true value associated with the second AI unit.

25. The method of claim 12 or 13, wherein, In a case where the second device is further used for implementing an inference function of the third AI unit, the method further includes: The second device sends tenth information to the third device, the tenth information including a target training data set, the target training data set being a data set obtained by performing data screening on the training data set of the second AI unit by the third AI unit.

26. The method of claim 25, wherein, The tenth information further includes at least one of the following: a sample identifier associated with the target training data set or a sample set identifier associated with the target training data set; a third AI unit used by the second device.

27. The method according to any one of claims 12 to 26, characterized in that, The second device obtains target information of the second AI unit based on the first information and inference of the first AI unit, and the target information includes at least one of: The second device obtains a virtual evaluation result of the second AI unit by taking input data and output data of the second AI unit in the first information as input of the first AI unit. The second device obtains a virtual true value of the second AI unit by taking input data of the second AI unit in the first information as input of the first AI unit.

28. A data collection method, characterized by, Comprising: The third device performs a target operation, and the target operation includes any one of: The eighth information includes a training data set of the second AI unit, and the training data set is determined based on data filtering of the training data set of the second AI unit by the third AI unit. The training data set of the second AI unit is determined based on a first AI unit and first information sent by a first device to the second device, the first AI unit is used to infer to obtain target information of the second AI unit, the target information includes a virtual true value or a virtual evaluation result, the second AI unit is used to realize a use case function, the first device is used to realize an inference function of the second AI unit, the second device is used to realize an inference function of the first AI unit, the third device is used to realize at least one of a control function and a collection function of training data set collection, and the fourth device is used to realize an inference function of the third AI unit. The eighth information further includes at least one of:

29. The method of claim 28, wherein, a sample identifier associated with the training data set or a sample set identifier associated with the training data set; indication information determined based on the first AI unit; a measurement quantity associated with the target information; a first AI unit used by the second device; a second AI unit used by the first device. The tenth information further includes at least one of:

30. The method of claim 28, wherein, a sample identifier associated with the target training data set or a sample set identifier associated with the target training data set; a third AI unit used by the fourth device. The method further includes:

31. The method of any one of claims 28 to 30, wherein, The third device transmits second information to the first device, and the second information is used to indicate information related to training data set collection; The first device is used to realize an inference function of the first AI unit, and the second information includes at least one of: first indication information used to indicate at least one candidate first sample data supported or expected; second indication information used to indicate at least one candidate purpose supported or expected for data collection; third indication information used to indicate whether to generate target information based on the first AI unit. After the third device sends the second information to the first device, the method further includes:

32. The method of claim 31, wherein, ​ The third device transmits third information to the first device, the third information being used to trigger the collection of the training data set, and the third information including at least one of the following: Third indication information used to indicate whether target information is to be generated based on the first AI unit; Fourth indication information used to indicate target first sample data in the at least one candidate first sample data; Fifth indication information used to indicate a target purpose in the at least one candidate purpose; Indication information used to indicate monitoring.

33. The method according to any one of claims 28 to 32, characterized in that, The method further includes: The third device transmits fourth information to the second device, the fourth information being used to indicate information related to the collection of the training data set; The fourth information includes at least one of the following: Sixth indication information used to indicate whether the first device supports or expects target information to be generated based on the first AI unit; Seventh indication information used to indicate at least one candidate second sample data supported or expected by the first device, the candidate second sample data including at least one of input data of the first AI unit and output data of the first AI unit: Eighth indication information used to indicate at least one candidate second AI unit supported or expected by the first device; Ninth indication information used to indicate at least one candidate third AI unit supported or expected by the first device; Tenth indication information used to indicate at least one candidate purpose of data collection supported or expected by the first device; The candidate third AI unit is used to perform data screening on the training data set.

34. The method of claim 33, wherein, After the third device sends the fourth information to the second device, the method further includes: The third device transmits fifth information to the second device, the fifth information being used to trigger the collection of the training data set; The fifth information includes at least one of the following: Third indication information used to indicate whether target information is to be generated based on the first AI unit; Eleventh indication information used to indicate target second sample data in the at least one candidate second sample data: Twelfth indication information used to indicate a target second AI unit in the at least one candidate second AI unit; Thirteenth indication information used to indicate a target third AI unit in the at least one candidate third AI unit; Fourteenth indication information used to indicate a target purpose in the at least one candidate purpose; Indication information used to indicate monitoring.

35. The method of any one of claims 28 to 34, wherein, The method further includes: The third device transmits sixth information to the fourth device, the sixth information being used to indicate information related to the collection of the training data set; The sixth information includes at least one of the following: At least one candidate screening rule supported or expected by the first device; At least one candidate sample classification category information supported or expected by the first device; At least one candidate threshold information supported or expected by the first device, the candidate threshold information being used to determine a sample classification category; At least one candidate sample proportion supported or expected by the first device; At least one candidate sample classification number supported or expected by the first device; at least one candidate sample proportion supported or expected by the first device; at least one candidate input data of the third AI unit supported or expected by the first device; at least one candidate output data of the third AI unit supported or expected by the first device; fifteenth indication information for indicating whether data screening is supported or expected by the first device; The fourth device is configured to implement an inference function of a third AI unit, and the third AI unit is configured to perform data screening on the training data set.

36. The method of claim 35, wherein, After the third device sends the sixth information to the fourth device, the method further includes: The third device transmits seventh information to the second device, and the seventh information is used to trigger collection of the training data set. The seventh information includes at least one of the following: a target screening rule in the at least one candidate screening rule; target classification sample category information in the at least one candidate sample classification category information; target threshold information in the at least one candidate threshold information; a target sample proportion in the at least one candidate sample proportion; target sample classification number information in the at least one candidate sample classification number information; at least one target sample proportion in the at least one candidate sample proportion; target input data in the at least one candidate input data of the third AI unit; target output data in the at least one candidate output data of the third AI unit; sixteenth indication information for indicating whether data screening is performed.

37. The method of claim 32, wherein, The method further includes: The third device sends ninth information to the fourth device, and the ninth information includes the training data set. The fourth device is configured to implement an inference function of a third AI unit, and the third AI unit is configured to perform data screening on the training data set.

38. The method of claim 37, wherein, The ninth information further includes at least one of the following: a sample identifier associated with the training data set or a sample set identifier associated with the training data set; a measurement quantity associated with the target information; a first AI unit used by the second device; a second AI unit used by the first device; a third AI unit used by the fourth device.

39. A method of data collection, comprising: The method further includes: The fourth device receives ninth information from the third device, and the ninth information includes a training data set of a second AI unit, which is determined based on a first AI unit and first information sent by a first device to a second device; The fourth device performs data screening on the training data set of the second AI unit based on a third AI unit to obtain a target training data set; The first AI unit is configured to infer to obtain target information of the second AI unit, the target information includes a virtual true value or a virtual evaluation result, the second AI unit is configured to implement a use case function, the first device is configured to implement an inference function of the second AI unit, and the second device is configured to implement an inference function of the first AI unit.

40. The method of claim 39, wherein, The ninth information further includes at least one of the following: a sample identifier associated with the training data set or a sample set identifier associated with the training data set; a measurement quantity associated with the target information; a first AI unit used by the second device; A second AI unit used by the first device; A third AI unit used by the fourth device.

41. The method of claim 39 or 40, wherein, The method further includes: The fourth device sends tenth information to the third device, and the tenth information includes the target training data set.

42. The method of claim 41, wherein, The tenth information further includes at least one of: A sample identifier associated with the target training data set or a sample set identifier associated with the target training data set; A third AI unit used by the fourth device.

43. The method of any one of claims 39 to 42, wherein, The method further includes: The fourth device transmits sixth information to the third device, and the sixth information is used to indicate information related to training data set collection; The sixth information includes at least one of: At least one candidate filtering rule supported or expected by the first device; At least one candidate sample classification category information supported or expected by the first device; At least one candidate threshold information supported or expected by the first device, the candidate threshold information being used to determine a sample classification category; At least one candidate sample proportion supported or expected by the first device; At least one candidate sample classification number supported or expected by the first device; At least one candidate sample proportion supported or expected by the first device; At least one candidate input data of the third AI unit supported or expected by the first device; At least one candidate output data of the third AI unit supported or expected by the first device; Fifteenth indication information used to indicate whether the first device supports or expects to perform data filtering; The fourth device is configured to implement an inference function of the third AI unit, and the third AI unit is configured to perform data filtering on the training data set.

44. The method of claim 43, wherein, The method further includes: The fourth device transmits seventh information to the third device, and the seventh information is used to trigger collection of the training data set; The seventh information includes at least one of: A target filtering rule in the at least one candidate filtering rule; A target classification sample category information in the at least one candidate sample classification category information; A target threshold information in the at least one candidate threshold information; A target sample proportion in the at least one candidate sample proportion; A target sample classification number information in the at least one candidate sample classification number information; A target sample proportion in the at least one candidate sample proportion; A target input data in the at least one candidate input data of the third AI unit; A target output data in the at least one candidate output data of the third AI unit; Sixteenth indication information used to indicate whether data filtering is performed.

45. A data collection apparatus applied to a first device, the apparatus comprising: The method further includes: A first transmission module configured to send first information to a second device, the first information being used to indicate that target information of a second AI unit is obtained based on an inference of a first AI unit, and the target information including virtual true values or virtual evaluation results; The target information is used to determine a training data set of the second AI unit, the second AI unit being configured to implement a use case function, the first device being configured to implement an inference function of the second AI unit, and the second device being configured to implement an inference function of the first AI unit.

46. The device of claim 45, wherein, The first transmission module is further configured to: The first device transmits second information to the third device, the second information being used to indicate information related to training data set collection; The third device is configured to implement at least one of a control function and a collection function of the training data set collection, and the second information includes at least one of: First indication information used to indicate at least one candidate first sample data supported or expected; Second indication information used to indicate at least one candidate purpose of data collection supported or expected; Third indication information used to indicate whether target information is generated based on the first AI unit.

47. A data collection apparatus applied to a second device, the apparatus comprising: Comprise: A second transmission module configured to receive first information from a first device; A first processing module configured to obtain target information of a second AI unit based on the first information and first AI unit inference, the target information including virtual true value or virtual evaluation result; The target information is used to determine a training data set of the second AI unit, the second AI unit is used to implement a use case function, the first device is used to implement an inference function of the second AI unit, and the second device is used to implement an inference function of the first AI unit.

48. The device of claim 47, wherein, The second transmission module is further configured to transmit fourth information to the third device, the fourth information being used to indicate information related to training data set collection; The fourth information includes at least one of: Sixth indication information used to indicate whether the first device supports or expects to generate target information based on the first AI unit; Seventh indication information used to indicate at least one candidate second sample data supported or expected by the first device, the candidate second sample data including at least one of input data of the first AI unit and output data of the first AI unit: Eighth indication information used to indicate at least one candidate second AI unit supported or expected by the first device; Ninth indication information used to indicate at least one candidate third AI unit supported or expected by the first device; Tenth indication information used to indicate at least one candidate purpose of data collection supported or expected by the first device; The third device is configured to implement at least one of a control function and a collection function of the training data set collection, and the candidate third AI unit is configured to perform data screening on the training data set.

49. A data collection apparatus applied to a third device, characterized by comprising: Comprise: A third transmission module configured to perform a target operation, the target operation including any one of: Receiving eighth information from a second device, the eighth information including a training data set of a second AI unit; Receiving tenth information from a fourth device, the tenth information including a target training data set, the target training data set being a data set screened based on a third AI unit on a training data set of the second AI unit; The training data set of the second AI unit is determined based on a first AI unit and first information sent by a first device to the second device, the first AI unit is configured to infer target information of the second AI unit, the target information includes virtual true value or virtual evaluation result, the second AI unit is configured to implement use case function, the first device is configured to implement inference function of the second AI unit, the second device is configured to implement inference function of the first AI unit, and the third device is configured to implement at least one of control function and collection function of training data set collection.

50. The device of claim 49, wherein, The third transmission module is further configured to transmit second information to the first device, and the second information is used to indicate information related to training data set collection. The first device is configured to implement inference function of the first AI unit, and the second information includes at least one of the following: First indication information, used to indicate at least one candidate first sample data supported or expected; Second indication information, used to indicate at least one candidate purpose of supporting or expecting data collection; Third indication information, used to indicate whether to generate target information based on the first AI unit.

51. A data collection device for use with a fourth device, the data collection device comprising: Comprise: The fourth transmission module is configured to receive ninth information from the third device, and the ninth information includes training data set of the second AI unit, and the training data set is determined based on the first information sent by the first AI unit and the first device to the second device; The second processing module is configured to perform data screening on the training data set of the second AI unit based on the third AI unit, and obtain target training data set; The first AI unit is configured to infer target information of the second AI unit, the target information includes virtual true value or virtual evaluation result, the second AI unit is configured to implement use case function, and the first device is configured to implement inference function of the second AI unit.

52. The device of claim 51, wherein, The fourth transmission module is further configured to send tenth information to the third device, and the tenth information includes the target training data set.

53. A terminal, characterized by The device comprises a processor and a memory, the memory stores programs or instructions executable on the processor, and the programs or instructions are executed by the processor to implement the steps of the data collection method in any one of claims 1 to 44.

54. A network-side device, comprising: The device comprises a processor and a memory, the memory stores programs or instructions executable on the processor, and the programs or instructions are executed by the processor to implement the steps of the data collection method in any one of claims 1 to 44.

55. A readable storage medium, characterized by, The readable storage medium stores programs or instructions, and the programs or instructions are executed by the processor to implement the steps of the data collection method in any one of claims 1 to 44.

56. A computer program product, characterised in that, The computer instructions are executed by the processor to implement the steps of the data collection method in any one of claims 1 to 44.

Images