[go: up one dir, main page]

WO2024169797A1 - Ai model indication method and communication device - Google Patents

Ai model indication method and communication device Download PDF

Info

Publication number
WO2024169797A1
WO2024169797A1 PCT/CN2024/076413 CN2024076413W WO2024169797A1 WO 2024169797 A1 WO2024169797 A1 WO 2024169797A1 CN 2024076413 W CN2024076413 W CN 2024076413W WO 2024169797 A1 WO2024169797 A1 WO 2024169797A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
models
association relationship
same
indication information
Prior art date
Application number
PCT/CN2024/076413
Other languages
French (fr)
Chinese (zh)
Inventor
贾承璐
杨昂
Original Assignee
维沃移动通信有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 维沃移动通信有限公司 filed Critical 维沃移动通信有限公司
Publication of WO2024169797A1 publication Critical patent/WO2024169797A1/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/16Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/04Inference or reasoning models
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14Network analysis or design
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14Network analysis or design
    • H04L41/145Network analysis or design involving simulating, designing, planning or modelling of a network

Definitions

  • the present application belongs to the field of communication technology, and specifically relates to an AI model indication method and communication equipment.
  • AI models are usually introduced to perform some tasks to improve network throughput, latency, and user capacity.
  • a positioning model can be used to predict the location information of a terminal
  • a channel measurement model can be used to perform channel estimation.
  • other communication devices can indicate or transmit the AI model to the communication device, and when indicating or transmitting the AI model, all the information of the AI model will be indicated or transmitted so that the communication device can obtain the AI model.
  • this method usually consumes more transmission resources, resulting in a waste of transmission resources.
  • the embodiments of the present application provide an AI model indication method and a communication device, which can solve the problem that a large amount of transmission resources are consumed when transmitting or indicating an AI model, resulting in a waste of transmission resources.
  • an AI model indication method which is performed by a communication device, and the method includes:
  • the communication device sends or receives indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  • an AI model indication device comprising:
  • a communication module is used to send or receive indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  • a communication device comprising a processor and a memory, wherein the memory stores a program or instruction that can be executed on the processor, and when the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.
  • a communication device comprising a processor and a communication interface, wherein the communication interface is used to send or receive indication information, and the indication information is used to indicate an association relationship between multiple artificial intelligence AI models.
  • a readable storage medium on which a program or instruction is stored.
  • the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.
  • a chip comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method described in the first aspect.
  • a computer program/program product is provided, wherein the computer program/program product is stored in a storage medium, and the program/program product is executed by at least one processor to implement the steps of the AI model indication method as described in the first aspect.
  • the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model.
  • the traditional model indication since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
  • FIG1 is a schematic diagram of a wireless communication system according to an embodiment of the present application.
  • FIG2 is a schematic flow chart of an AI model indication method according to an embodiment of the present application.
  • FIG3 is a schematic diagram of the structure of an AI model indication device according to an embodiment of the present application.
  • FIG4 is a schematic diagram of the structure of a communication device according to an embodiment of the present application.
  • FIG5 is a schematic diagram of the structure of a communication device according to an embodiment of the present application.
  • FIG6 is a schematic diagram of the structure of a communication device according to an embodiment of the present application.
  • first, second, etc. of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by “first” and “second” are generally of one type, and the number of objects is not limited, for example, the first object can be one or more.
  • “or” in the present application represents at least one of the connected objects.
  • “A or B” covers three schemes, namely, Scheme 1: including A but not including B; Scheme 2: including B but not including A; Scheme 3: including both A and B.
  • the character "/" generally indicates that the objects associated with each other are in an "or” relationship.
  • indication in this application can be a direct indication (or explicit indication) or an indirect indication (or implicit indication).
  • a direct indication can be understood as the sender explicitly informing the receiver of specific information, operations to be performed, or request results in the sent indication;
  • an indirect indication can be understood as the receiver determining the corresponding information according to the indication sent by the sender, or making a judgment and determining the operation to be performed or the request result according to the judgment result.
  • LTE Long Term Evolution
  • LTE-A Long Term Evolution-Advanced
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • SC-FDMA Single-carrier Frequency-Division Multiple Access
  • NR New Radio
  • 6G 6th Generation
  • FIG1 shows a block diagram of a wireless communication system applicable to an embodiment of the present application.
  • the wireless communication system includes a terminal 11 and a network side device 12 .
  • the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a handheld computer, a netbook, an ultra-mobile personal computer (Ultra-mobile Personal Computer, UMPC), a mobile Internet device (Mobile Internet Device, MID), an augmented reality (Augmented Reality, AR), a virtual reality (Virtual Reality, VR) device, a robot, a wearable device (Wearable Device), a flight vehicle (flight vehicle), a vehicle user equipment (VUE), a shipborne equipment, a pedestrian terminal (Pedestrian User Equipment, PUE), a smart home (home appliances with wireless communication functions, such as refrigerators, televisions, washing machines or furniture, etc.), a game console, a personal computer (
  • Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc.
  • the vehicle-mounted device can also be called a vehicle-mounted terminal, a vehicle-mounted controller, a vehicle-mounted module, a vehicle-mounted component, a vehicle-mounted chip or a vehicle-mounted unit, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application.
  • the network side device 12 may include an access network device or a core network device, wherein the access network device may also be called a radio access network (Radio Access Network, RAN) device, a radio access network function or a radio access network unit.
  • the access network device may include a base station, a wireless local area network (Wireless Local Area Network, WLAN) access point (Access Point, AS) or a wireless fidelity (Wireless Fidelity, WiFi) node, etc.
  • WLAN wireless Local Area Network
  • AS Access Point
  • WiFi wireless Fidelity
  • the base station can be called Node B (Node B, NB), Evolved Node B (Evolved Node B, eNB), the next generation Node B (the next generation Node B, gNB), New Radio Node B (New Radio Node B, NR Node B), access point, Relay Base Station (Relay Base Station, RBS), Serving Base Station (Serving Base Station, SBS), Base Transceiver Station (Base Transceiver Station, BTS), radio base station, radio transceiver, base
  • the base station is not limited to specific technical terms as long as the same technical effect is achieved. It should be noted that in the embodiments of the present application, only the base station in the NR system is taken as an example for introduction, and the specific type of the base station is not limited.
  • the embodiment of the present application provides an AI model indication method and communication device.
  • the association relationship between multiple AI models can be indicated through indication information.
  • the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model.
  • transmission resources can be reduced and waste of transmission resources can be avoided.
  • AI artificial intelligence
  • the artificial intelligence (AI) models in the embodiments of the present application include but are not limited to neural networks, decision trees, support vector machines, Bayesian classifiers, etc.
  • an embodiment of the present application provides an AI model indication method 200, which can be executed by a communication device.
  • the AI model indication method can be executed by software or hardware installed in the communication device.
  • the AI model indication method includes the following steps.
  • the communication device sends or receives indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  • the communication device can send indication information to other communication devices to indicate the association relationship between multiple AI models to other communication devices, and then the other communication devices determine the model information indicated by the communication device based on the association relationship.
  • the communication device can receive indication information sent by other communication devices to determine the association relationship between multiple AI models based on the indication information, and then determine the model information indicated by other communication devices based on the association relationship.
  • “multiple" refers to two or more.
  • the communication device may be a terminal or a network side device.
  • the communication device sending the indication information may be sending the indication information to other terminals, and the communication device receiving the indication information may be receiving the indication information from other terminals or network side devices.
  • the communication device sending the indication information may be sending the indication information to a terminal or other network side device, and the communication device receiving the indication information may be receiving the indication information from other network side devices.
  • the multiple AI models may be models to be indicated, or some may have been indicated (or pre-configured) and others may not have been indicated, or all may have been indicated (or pre-configured).
  • By indicating the association relationship between the multiple AI models it is convenient for the communication device that receives the indication information to determine (or construct) the multiple AI models, or some of the multiple AI models, or other models related to the multiple AI models according to the association relationship.
  • multiple AI models may be models that the communication device needs to indicate to other communication devices.
  • the same model information can be avoided from being repeatedly transmitted when the multiple AI models are subsequently indicated, thereby reducing transmission resources.
  • Other communication devices Multiple AI models can be constructed according to the indication information; or, multiple AI models may be that some of the AI models have been indicated to other communication devices (or have been pre-configured in other communication devices), and the other AI models have not yet been indicated.
  • the same model information can be avoided from being repeatedly transmitted when the other part of the AI model is subsequently transmitted, thereby reducing transmission resources.
  • the other communication devices can construct another part of the AI model according to the indication information and the indicated (or configured) model; or, multiple AI models may be that all of them have been indicated to other communication devices (or have been pre-configured in other communication devices).
  • the other communication devices can construct a new model according to the indication information and the multiple AI models that have been indicated (or configured). In this way, the communication device does not need to indicate the new model, thereby reducing transmission resources.
  • the association relationship between multiple AI models can be the association relationship between the model features of multiple AI models.
  • association relationship between AI models it can be known which features between AI models are associated.
  • association relationship between AI models it is also possible to indicate the association relationship between the tasks/functions associated with the AI models.
  • association relationship between the multiple AI models may include at least one of the following ten association relationships:
  • a first association relationship where the first association relationship indicates that the model inputs of multiple AI models are the same.
  • the same model input may include the same type of model input and/or the same format of model input.
  • the type of model input may be the specific information of the model input, such as a time domain channel impulse response.
  • the format of the model input may be the arrangement of the model input, etc.
  • the format of the model input may be the number of carrier to interference ratio (CIR) sampling points associated with each transmission and reception point (TRP), the arrangement of CIR, the arrangement of time domain channel impulse responses of different TRP IDs, the number of sampling points of the time domain channel impulse response, etc.
  • CIR carrier to interference ratio
  • Model A is a positioning model
  • the input is the channel impulse response of multiple TRPs
  • the output is the UE position.
  • Model B is used to supervise the effectiveness of Model A
  • the input is also the channel impulse response of multiple TRPs
  • the output is the confidence that Model A is valid under the current input. Therefore, the model inputs of Model A and Model B are the same.
  • the model inputs of other AI models can be determined based on the model inputs of one of the AI models. In this way, there is no need to indicate the model inputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A and model B with the same model input to device 2, then device 1 can indicate model A to device 2 and indicate that the model inputs of model A and model B are the same.
  • Device 2 can determine the model input of model A based on model B. In this way, device 1 does not need to indicate the model input of model B, thereby avoiding repeated transmission of the same model input, reducing transmission resources, and avoiding waste of transmission resources.
  • the indication information when the indication information indicates the first association relationship, that is, when indicating that the model inputs of the multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model inputs, so that the model information of one of the AI models and the model information of the multiple AI models can be used to distinguish the differences between the model information of the multiple AI models.
  • the distinction between the model information determines the model information of the remaining AI models.
  • the same model output may include the same type of model output and/or the same format of model output.
  • the type of model output may be what information the model output specifically is, such as position coordinates, etc.
  • the format of model output may be the dimension of model output, etc. For example, when the model output is a coordinate position, the format of the model output may be that the first coordinate position represents the first dimension, and the second coordinate position represents the second dimension.
  • model A is the positioning model of scene A
  • model B is the positioning model of scene B
  • the input of model A is the CIR information of N TRPs in scene A, and the output is the position.
  • the input of model B is the CIR information of M TRPs in scene B, and the output is also the position. Therefore, the model outputs of model A and model B are the same.
  • the model outputs of other AI models can be determined based on the model output of one of the AI models. In this way, there is no need to indicate the model outputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • model B has been pre-configured in device 2, and device 1 wants to indicate model A to device 2, and the model output of model A is the same as the model output of model B. Then, when device 1 indicates model A to device 2, it can indicate that the model outputs of model A and model B are the same, and device 2 can determine the model output of model A based on model B. In this way, device 1 does not need to indicate the model output of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information when the indication information indicates a second association relationship, that is, indicating that the model outputs of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model outputs, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
  • the third association relationship represents that the model input and model output of multiple AI models are the same.
  • the same model inputs may include the same model input types and/or the same model input formats.
  • the same model outputs may include the same model output types and/or the same model output formats.
  • Model A and Model B are two positioning models of different complexity.
  • the structures of the middle layers of the two models are different, but the input is the CIR information of N TRPs, and the output is the location information. Therefore, the model input and model output of Model A and Model B are the same.
  • the model inputs and outputs of other AI models can be determined based on the model inputs and outputs of one of the AI models. In this way, there is no need to indicate the model inputs and outputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A and model B with the same model input and output to device 2, then device 1 can indicate model A to device 2 and indicate that the model input and output of model A and model B are the same.
  • Device 2 can determine the model input and output of model A based on model B. In this way, device 1 can indicate model A to device 2.
  • model A there is no need to indicate the model input and output of model B, thereby avoiding repeated transmission of the same model input and output, reducing transmission resources, and avoiding waste of transmission resources.
  • the indication information when the indication information indicates a third association relationship, that is, indicating that the model inputs and outputs of multiple AI models are the same, the indication information may further indicate the differences between other model information of the multiple AI models except for the model inputs and outputs. In this way, the model information of the remaining AI models can be determined based on the model information of one of the AI models and the differences between the model information of multiple AI models.
  • the same model structure includes but is not limited to the same number of model layers, and/or the same number of elements included in the model layers, and/or the same type of model layers, and/or the same type of activation functions (if any) in the model layers.
  • the same model structure of multiple AI models may include at least one of the following:
  • the number of layers of neurons is the same;
  • the number of neurons included in each layer is the same;
  • the type of activation function included in each layer of neurons is the same;
  • the type of neurons in each layer (e.g., batch normalization layer, layer normalization layer, convolutional layer, etc.) is the same.
  • Model A and Model B are both models in the model pool, and their functions are positioning. They contain the same number of convolutional layers, but Model A and Model B correspond to different usage scenarios and have different parameters of the convolutional layers, such as different positioning reference signal (PRS) configurations. Then Model A and Model B have the same structure.
  • PRS positioning reference signal
  • the structures of other AI models can be determined based on the model structure of one of the AI models. In this way, there is no need to indicate the model structures of other AI models. That is, when transmitting the model, there is no need to transmit the model structures of other AI models, but only the parameters corresponding to the model structure. This can reduce the transmission resources consumed when indicating the model and avoid waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2.
  • Model A and model B have the same model structure.
  • device 1 indicates model A, it can indicate that the model structures of model A and model B are the same.
  • Device 2 can determine the model structure of model A based on model B. In this way, device 1 does not need to transmit the model structure information of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
  • device 1 sends a model pool to device 2.
  • the model pool contains multiple models. These models have the same model structure but different model parameters. If the model configured in device 2 fails, device 1 can update the failed model by only configuring the parameters.
  • the indication information when the indication information indicates a fourth association relationship, that is, indicating that the model structures of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model structure, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
  • the fifth association relationship represents that some model structures of multiple AI models are the same.
  • model structures of multiple AI models have both common and different parts.
  • the common partial structures of model A and model B can be:
  • the structure of the first N1 model layers of model A is the same as the structure of the first N1 model layers of model B, and N1 is less than the total number of layers M1 of model A and less than the total number of layers M2 of model B; or,
  • the number of model layers of model A is the same as the number of model layers of model B, but the number of neurons in each model layer of model A is smaller than the number of neurons in each model layer of model B.
  • Model A and Model B may be the same in other cases besides the above three cases, which will not be explained one by one here.
  • model A is a positioning model
  • the input is the channel impulse response of multiple TRPs
  • the output is the UE position.
  • Model B is used to supervise the effectiveness of model A
  • the input is also the channel impulse response of multiple TRPs
  • the output is the confidence that model A is effective under the current input.
  • Only the last layer structure is different between model B and model A.
  • the number of neurons in the last layer of model B is changed from 2 dimensions of model A to 1 dimension, and the activation function is changed from linear to sigmoid.
  • some structures of model A and model B are the same.
  • model A and model B also have the above-mentioned first association relationship and second association relationship.
  • the same partial structures of other AI models can be determined based on the model structure of one of the AI models. In this way, there is no need to indicate the same model structure parts of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2.
  • Model A and model B have some identical model structures.
  • device 1 when device 1 indicates model A, it can indicate that model A and model B have some identical model structures.
  • Device 2 can determine the partial model structure of model A based on model B. In this way, device 1 does not need to transmit the model structure information in model A that is identical to that in model B, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information when the indication information indicates the fifth association relationship, that is, indicating that some model structures of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except for the partial model structure, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
  • the indication information may also include the differences in the model structures of the multiple AI models.
  • the model structures of the remaining models may be determined based on the model structure of one of the AI models and the same and different parts of the structures of the multiple models indicated by the indication information.
  • the indication information can also indicate the difference between the last (M1-N1) model layers of model A and the last (M2-N1) model layers of model B.
  • the last (M1-N1) model layers of model A each contain 1 neuron
  • the last (M2-N1) model layers of model B each contain 2 neurons.
  • the sixth association relationship represents that some model structures and corresponding parameters of multiple AI models are the same.
  • the sixth association relationship not only indicates that the partial structures of multiple AI models are the same, but also further indicates that the parameters corresponding to the same partial structures are the same.
  • the parameters here can be model parameters corresponding to the model structure.
  • the parameters corresponding to the model structure can be the weights and biases of the model structure.
  • an AI model trained in one scenario can be migrated to another similar scenario.
  • the model structure and parameters of the first N layers of the original model can be kept unchanged, and only the structure and corresponding parameters of the last M layers of the original model are changed.
  • the first N layer structures and corresponding parameters between the original model and the retrained new model are the same.
  • the same partial structures and corresponding parameters of other AI models can be determined based on the model structure and corresponding parameters of one of the AI models. In this way, there is no need to indicate the same model structure parts and corresponding parameters of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. Some model structures and corresponding parameters of model A and model B are the same. Then, when device 1 indicates model A, it can indicate that some model structures and corresponding parameters of model A and model B are the same. Device 2 can determine some model structures and corresponding parameters of model A based on model B. In this way, device 1 does not need to transmit the information of model structure and corresponding parameters in model A that are the same as those in model B, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information when the indication information indicates the sixth association relationship, that is, indicating that some model structures and corresponding parameters of multiple AI models are the same, the indication information may further indicate the difference between other model information of multiple AI models except for the partial model structure and corresponding parameters. In this way, the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
  • the indication information may also include the differences in the model structures of the multiple AI models.
  • the model structures of the remaining models may be determined based on the model structure of one of the AI models and the same and different parts of the structures of the multiple models indicated by the indication information.
  • the seventh relationship represents that the model structures and corresponding parameters of multiple AI models are the same.
  • the multiple AI models can be considered to be the same model.
  • the models corresponding to multiple UEs in the same scene may be exactly the same.
  • the structures and corresponding parameters of other AI models can be determined based on the model structure and corresponding parameters of one AI model, so that there is no need to The structure and corresponding parameters of other AI models can be indicated, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2.
  • the model structures and corresponding parameters of model A and model B are the same.
  • device 1 when device 1 indicates model A, it can indicate that the model structures and corresponding parameters of model A and model B are the same.
  • Device 2 can determine the model structure and corresponding parameters of model A based on model B. In this way, device 1 does not need to transmit the information of the model structure and corresponding parameters of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information may further indicate the difference between other model information of the multiple AI models except the model structure and corresponding parameters, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
  • An eighth association relationship represents that the first model among the multiple AI models is a sub-model of the second model among the multiple AI models.
  • the first model is a sub-model of the second model, which can be considered as a part of the second model.
  • model A can be the first N3 model layers of model B, and the structures and corresponding parameters of the first N3 model layers are the same.
  • N3 is less than the total number of model layers of model B.
  • model A is its first N+1 layers of encoder part or the last M+1 layers of decoder part, then model A is a submodel of model B.
  • the first model can be determined according to the second model, so there is no need to indicate the first model.
  • some structures and parameters in the second model can be determined according to the first model, so there is no need to indicate the parts of the second model that are the same as the first model, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2, and model A is a sub-model of model B. Then, when device 1 indicates model A, it can indicate that model A is a sub-model of model B. Device 2 can determine model A based on model B. In this way, device 1 does not need to transmit model A, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information may also include the first part in the second model, the structure of the first part is the same as the structure of the first model, and the parameters of the first part are the same as the parameters of the first model. That is to say, when the indication information indicates the eighth association relationship, it may also indicate which part of the second model the first model is, and the part is the first part.
  • the first model is the first N layers of the second model, the encoder, or the last M layers of the second model, the decoder, etc. In this way, the specific structure and specific parameters of the first model can be determined according to the second model, or the structure and corresponding parameters of the part of the second model that is the same as the first model can be determined according to the first model.
  • the indication information may further include a model portion of the second model that is different from the first model, and the remaining portion is the same as the first model.
  • the specific structure and specific parameters of the first model may also be determined based on the second model, or the second model may be determined based on the first model.
  • a ninth association relationship wherein the ninth association relationship represents that the structure of the third model among the multiple AI models is a substructure of the fourth model among the multiple AI models.
  • the third model is a substructure of the fourth model. It can be considered that the third model has the same structure as a part of the fourth model, but the corresponding parameters are different.
  • the structure of model A is the same as the structure of the first N3 model layers of model B, but the parameters of model A are different from the parameters of the first N3 model layers of model B.
  • N3 is less than the total number of model layers of model B.
  • a possible application scenario may be to select a part of the complex model B structure as the structure of model A to reduce the complexity of the model.
  • the structure of the third model can be determined based on the structure of the fourth model, so that there is no need to indicate the structure of the third model.
  • part of the structure in the fourth model can be determined based on the third model, so that there is no need to indicate the part of the fourth model that has the same structure as the third model, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
  • device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2.
  • the structure of model A is a substructure of model B.
  • device 1 indicates model A, it can indicate that model A is a substructure of model B.
  • Device 2 determines the structure of model A based on the structure of model B. In this way, device 1 does not need to transmit the structural information of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information can also be used for the second part in the fourth model, the structure of the second part is the same as the structure of the third model, and the parameters of the second part are different from the parameters of the third model. That is to say, when the indication information indicates the ninth association relationship, it can also indicate which part of the fourth model the third model is, and this part is the second part.
  • the third model is the first N layers, encoder of the fourth model, or the last M layers, decoder, etc. of the fourth model, but the corresponding parameters are different. In this way, the specific structure of the third model can be determined according to the fourth model, or the structure of the part of the model in the fourth model that has the same structure as the third model can be determined according to the structure of the third model.
  • the indication information may further include structural parts of the fourth model that are different from the third model, and the remaining structural parts are the same as the third model.
  • the specific structure of the third model may be determined based on the fourth model, or the structure of the fourth model may be determined based on the structure of the third model.
  • the tenth association relationship represents that the output of the fifth model among the multiple models is the input of the sixth model among the multiple models.
  • the fifth model is the input of the sixth model. It can be considered that the fifth model and the sixth model are in a cascade relationship, and the output of the fifth model can be used as the input of the sixth model.
  • the original CSI can be first predicted by model A and then input to model B for compression.
  • model A and model B are Model B is a cascade relationship, and the output of model A is the input of model B.
  • device 1 wants to indicate model C to device 2.
  • Model A and model B are pre-configured in device 2.
  • Model C is a combination of model A and model B, and the model output of model A is the model input of model B.
  • device 1 indicates model C, it can indicate that the model output of model A is the model input of model B.
  • Device 2 can determine model C based on the instruction of device 1 and models A and B. In this way, device 1 does not need to transmit model C, thereby reducing transmission resources and avoiding waste of transmission resources.
  • the indication information may indicate any one or more combinations thereof, which may be specifically determined according to the actual application scenario and is not specifically limited here.
  • the indication information may also indicate the identification information of the AI model when indicating at least one of the above ten association relationships, and the identification information of the AI model may be used to determine multiple AI models with association relationships.
  • the indication information when indicating the identification information of the AI model, may include at least one of the following:
  • a first list wherein the first list includes identification information of multiple AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationship.
  • the specified order may be the order of precedence, top-bottom order, etc. of the identification information of the AI models in the first list.
  • the association order of the association relationship may be the order of precedence in which the AI models have the association relationship.
  • the first list contains identification information of two AI models, namely model A and model B.
  • Model A and model B have a sequence, that is, model A comes first and model B comes later.
  • the association relationship indicated by the indication information is the eighth association relationship mentioned above, in the eighth association relationship, the association order is the first model in front and the second model in the back, it can be determined that model A is a submodel of model B.
  • the indication information when used to indicate at least one of the above-mentioned eighth, ninth and tenth association relationships, the indication information may include a first list, so that the communication device that receives the indication information can easily determine which AI model is a sub-model of other AI models, or which AI model is a sub-structure of other AI models, or which AI model's output is the input of other AI models.
  • the remaining AI models may be models pre-agreed upon by the communicating parties.
  • the indication information includes identification information of model A
  • model B is a model pre-agreed upon by the communicating parties
  • the indication information is used for the above-mentioned first association relationship.
  • the indication information is specifically used to indicate that the model input of model A is the same as the model input of model B. According to the indication information, it can be determined that the model input of model A is the pre-agreed input of model B.
  • the remaining AI models may also be AI models that the communication device that receives the indication information needs to determine or construct.
  • the indication information includes identification information of model A, and the indication information is used to indicate the ninth association relationship.
  • the indication information is specifically used to indicate that there is a ninth association relationship between model B and model A.
  • the communication device can determine or construct model B based on the indication information, and model B is a substructure of model A.
  • the indication information can indicate the identification information of multiple AI models with associated relationships, and at the same time indicate what kind of association the multiple AI models have.
  • the indication information can include the identification information of model A and the identification information of model B.
  • the indication information also indicates that the model outputs of model A and model B are the same, and that some structures of model A and model B are the same.
  • the technical solution provided in the embodiment of the present application can avoid the problem of waste of transmission resources caused by repeated transmission of model features when the AI model is transmitted or the AI model information is indicated.
  • AI models such as model input, model output, structure and parameters, etc.
  • indication information it can potentially indicate the association relationship between the models, thereby providing a more flexible operation method for the joint processing of multiple AI-based applications.
  • the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model.
  • the traditional model indication since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
  • the model indication method provided in the embodiment of the present application can be executed by a model indication device.
  • the model indication device provided in the embodiment of the present application is described by taking the execution of the model indication method by the model indication device as an example.
  • Fig. 3 is a schematic diagram of the structure of an AI model indication device according to an embodiment of the present application, which may correspond to a communication device in other embodiments. As shown in Fig. 3, the device 300 includes the following modules.
  • the communication module 301 is used to send or receive indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  • the association relationship includes at least one of the following:
  • a first association relationship wherein the first association relationship represents that model inputs of the multiple AI models are the same;
  • a second association relationship wherein the second association relationship represents that the model outputs of the multiple AI models are the same;
  • a third association relationship wherein the third association relationship represents that the model inputs and model outputs of the multiple AI models are the same;
  • a fourth association relationship wherein the fourth association relationship represents that the model structures of the multiple AI models are the same;
  • a fifth association relationship wherein the fifth association relationship represents that some model structures of the multiple AI models are the same;
  • a sixth association relationship wherein the sixth association relationship represents that some model structures and corresponding parameters of the multiple AI models are the same;
  • a seventh association relationship wherein the seventh association relationship represents that the model structures and corresponding parameters of the multiple AI models are the same;
  • an eighth association relationship wherein the eighth association relationship represents that the first model among the multiple AI models is a submodel of the second model among the multiple AI models;
  • the ninth association relationship represents that a structure of a third model among the multiple AI models is a substructure of a fourth model among the multiple AI models;
  • the tenth association relationship represents that the output of the fifth model among the multiple AI models is the input of the sixth model among the multiple AI models.
  • model input is the same, including at least one of the following:
  • the model inputs are of the same type
  • the format of the model inputs is the same.
  • model outputs are the same, including at least one of the following:
  • the model outputs are of the same type
  • the model outputs are in the same format.
  • the indication information is also used to indicate the difference in model structures of the multiple AI models.
  • the indication information when the association relationship includes the eighth association relationship, also includes a first part in the second model, a structure of the first part is the same as a model structure of the first model, and parameters of the first part are the same as parameters of the first model.
  • the indication information when the association relationship includes the ninth association relationship, the indication information also includes the second part in the fourth model, the structure of the second part is the same as the model structure of the third model, and the parameters of the second part are different from the parameters of the third model.
  • the indication information includes at least one of the following:
  • a first list wherein the first list includes identification information of the plurality of AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationships;
  • Identification information of some AI models among the multiple AI models, and the remaining AI models have the association relationship with the some AI models
  • the identification information of the multiple AI models and the association relationship between the multiple AI models are described.
  • the process of the method 200 corresponding to the embodiment of the present application can be referred to, and the various units/modules in the device 300 and the above-mentioned other operations and/or functions are respectively for implementing the corresponding processes in the method 200, and can achieve the same or equivalent technical effects. For the sake of brevity, they will not be repeated here.
  • the model indicating device in the embodiment of the present application can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip.
  • the electronic device can be a terminal, or it can be other devices other than a terminal.
  • the terminal can include but is not limited to the types of terminals 11 listed above, and other devices can be servers, network attached storage (NAS), etc., which are not specifically limited in the embodiment of the present application.
  • the AI model indication device provided in the embodiment of the present application can implement each process implemented by the method embodiment of Figure 2 and achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the present embodiment also provides a communication device 400, including a processor 401 and a memory. 402, the memory 402 stores a program or instruction that can be run on the processor 401.
  • the program or instruction is executed by the processor 401 to implement the various steps of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect.
  • the communication device 400 is a network side device, the program or instruction is executed by the processor 401 to implement the various steps of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the embodiment of the present application also provides a communication device, including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a program or instruction to implement the steps in the method embodiment shown in Figure 2.
  • the communication device embodiment corresponds to the above-mentioned communication device side method embodiment, and each implementation process and implementation method of the above-mentioned method embodiment can be applied to the communication device embodiment and can achieve the same technical effect.
  • Figure 5 is a schematic diagram of the hardware structure of a communication device implementing an embodiment of the present application.
  • the communication device 500 includes but is not limited to: a radio frequency unit 501, a network module 502, an audio output unit 503, an input unit 504, a sensor 505, a display unit 506, a user input unit 507, an interface unit 508, a memory 509 and at least some of the components of a processor 510.
  • the communication device 500 may also include a power source (such as a battery) for supplying power to each component, and the power source may be logically connected to the processor 5 10 through a power management system, so that the power management system can manage charging, discharging, and power consumption.
  • a power source such as a battery
  • the communication device structure shown in FIG5 does not constitute a limitation on the terminal, and the terminal may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently, which will not be described in detail here.
  • the input unit 504 may include a graphics processing unit (GPU) 5041 and a microphone 5042, and the graphics processor 5041 processes the image data of the static picture or video obtained by the image capture device (such as a camera) in the video capture mode or the image capture mode.
  • the display unit 506 may include a display panel 5061, and the display panel 5061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc.
  • the user input unit 507 includes a touch panel 5071 and at least one of other input devices 5072.
  • the touch panel 5071 is also called a touch screen.
  • the touch panel 5071 may include two parts: a touch detection device and a touch controller.
  • Other input devices 5072 may include, but are not limited to, a physical keyboard, function keys (such as a volume control key, a switch key, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.
  • the radio frequency unit 501 after receiving downlink data from the network side device, can transmit the data to the processor 510 for processing; in addition, the radio frequency unit 501 can send uplink data to the network side device.
  • the radio frequency unit 501 includes but is not limited to an antenna, an amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc.
  • the memory 509 can be used to store software programs or instructions and various data.
  • the memory 509 can mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area can store an operating system, an application program or instruction required for at least one function (such as a sound playback function, an image playback function, etc.), etc.
  • the memory 509 can include a volatile memory or a non-volatile memory.
  • the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (ROM), or a programmable read-only memory (PROM).
  • the volatile memory may be a random access memory (RAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDRSDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synchronous link dynamic random access memory (SLDRAM) and a direct memory bus random access memory (DRRAM).
  • RAM random access memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • SDRAM synchronous dynamic random access memory
  • DDRSDRAM double data rate synchronous dynamic random access memory
  • ESDRAM enhanced synchronous dynamic random access memory
  • SLDRAM synchronous link dynamic random access memory
  • DRRAM direct memory bus random access memory
  • the processor 510 may include one or more processing units; optionally, the processor 510 integrates an application processor and a modem processor, wherein the application processor mainly processes operations related to an operating system, a user interface, and application programs, and the modem processor mainly processes wireless communication signals, such as a baseband processor. It is understandable that the modem processor may not be integrated into the processor 510.
  • the radio frequency unit 501 is used to send or receive indication information, and the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  • the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model.
  • the traditional model indication since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
  • the communication device 500 provided in the embodiment of the present application can also implement the various processes of the embodiment shown in Figure 2 above, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the embodiment of the present application also provides a communication device, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the steps of the method embodiment shown in Figure 2.
  • the communication device embodiment corresponds to the above communication device method embodiment, and each implementation process and implementation method of the above method embodiment can be applied to the communication device embodiment, and can achieve the same technical effect.
  • the communication device 600 includes: an antenna 61, a radio frequency device 62, a baseband device 63, a processor 64, and a memory 65.
  • the antenna 61 is connected to the radio frequency device 62.
  • the radio frequency device 62 receives information through the antenna 61 and sends the received information to the baseband device 63 for processing.
  • the baseband device 63 processes the information to be sent and sends it to the radio frequency device 62.
  • the radio frequency device 62 processes the received information and sends it out through the antenna 61.
  • the method executed by the communication device in the above embodiment may be implemented in the baseband device 63, which includes a baseband processor.
  • the baseband device 63 may include, for example, at least one baseband board on which a plurality of chips are arranged. As shown, one of the chips is, for example, a baseband processor, which is connected to the memory 65 through a bus interface to call the program in the memory 65 to execute the communication device operations shown in the above method embodiment.
  • the communication device may also include a network interface 66, which is, for example, a Common Public Radio Interface (CPRI).
  • a network interface 66 which is, for example, a Common Public Radio Interface (CPRI).
  • CPRI Common Public Radio Interface
  • the communication device 600 of the embodiment of the present application also includes: instructions or programs stored in the memory 65 and executable on the processor 64.
  • the processor 64 calls the instructions or programs in the memory 65 to execute the methods executed by the modules shown in Figure 3 and achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored.
  • a program or instruction is stored.
  • the various processes of the above-mentioned AI model indication method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
  • the processor is a processor in the communication device described in the above embodiment.
  • the readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.
  • the readable storage medium may be a non-transient readable storage medium.
  • An embodiment of the present application further provides a chip, which includes 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 the various processes of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
  • the embodiments of the present application further provide a computer program/program product, which is stored in a storage medium.
  • the computer program/program product is executed by at least one processor to implement the various processes of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the above-mentioned embodiment method can be implemented by means of a computer software product plus a necessary general hardware platform, and of course, it can also be implemented by hardware.
  • the computer software product is stored in a storage medium (such as ROM, RAM, disk, CD, etc.), including several instructions to enable the terminal or network side device to execute the method described in each embodiment of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The present application belongs to the technical field of communications. Disclosed are an AI model indication method and a communication device. The AI model indication method in the embodiments of the present application comprises: a communication device sending or receiving indication information, wherein the indication information is used for indicating an association relationship between a plurality of AI models.

Description

AI模型指示方法及通信设备AI model indication method and communication device
相关申请的交叉引用CROSS-REFERENCE TO RELATED APPLICATIONS
本申请主张在2023年02月13日在中国提交的申请号为202310114133.0的中国专利的优先权,其全部内容通过引用包含于此。This application claims priority to Chinese Patent Application No. 202310114133.0 filed in China on February 13, 2023, the entire contents of which are incorporated herein by reference.
技术领域Technical Field
本申请属于通信技术领域,具体涉及一种AI模型指示方法及通信设备。The present application belongs to the field of communication technology, and specifically relates to an AI model indication method and communication equipment.
背景技术Background Art
在通信网络中,通常会引入人工智能(Artificial Intelligence,AI)模型来执行一些任务,以提升网络的吞吐量、时延以及用户容量等。比如,可以使用定位模型预测终端的位置信息,使用信道测量模型进行信道估计等。一般地,某个通信设备在使用AI模型时,可以由其他通信设备将AI模型指示或传输给该通信设备,且在指示或传输AI模型时,会指示或传输AI模型的全部信息,以便通信设备可以获得该AI模型。然而,这种方法通常会消耗较多的传输资源,导致传输资源的浪费。In communication networks, artificial intelligence (AI) models are usually introduced to perform some tasks to improve network throughput, latency, and user capacity. For example, a positioning model can be used to predict the location information of a terminal, and a channel measurement model can be used to perform channel estimation. Generally, when a communication device uses an AI model, other communication devices can indicate or transmit the AI model to the communication device, and when indicating or transmitting the AI model, all the information of the AI model will be indicated or transmitted so that the communication device can obtain the AI model. However, this method usually consumes more transmission resources, resulting in a waste of transmission resources.
发明内容Summary of the invention
本申请实施例提供一种AI模型指示方法及通信设备,能够解决目前在传输或指示AI模型时,消耗的传输资源较多,导致传输资源浪费的问题。The embodiments of the present application provide an AI model indication method and a communication device, which can solve the problem that a large amount of transmission resources are consumed when transmitting or indicating an AI model, resulting in a waste of transmission resources.
第一方面,提供了一种AI模型指示方法,由通信设备执行,该方法包括:In a first aspect, an AI model indication method is provided, which is performed by a communication device, and the method includes:
通信设备发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。The communication device sends or receives indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
第二方面,提供了一种AI模型指示装置,该装置包括:In a second aspect, an AI model indication device is provided, the device comprising:
通信模块,用于发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。A communication module is used to send or receive indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
第三方面,提供了一种通信设备,该通信设备包括处理器和存储器,所述存储器存储可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如第一方面所述的方法的步骤。According to a third aspect, a communication device is provided, comprising a processor and a memory, wherein the memory stores a program or instruction that can be executed on the processor, and when the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.
第四方面,提供了一种通信设备,包括处理器及通信接口,其中,所述通信接口用于发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。In a fourth aspect, a communication device is provided, comprising a processor and a communication interface, wherein the communication interface is used to send or receive indication information, and the indication information is used to indicate an association relationship between multiple artificial intelligence AI models.
第五方面,提供了一种可读存储介质,所述可读存储介质上存储程序或指令,所述程序或指令被处理器执行时实现如第一方面所述的方法的步骤。In a fifth aspect, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.
第六方面,提供了一种芯片,所述芯片包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现如第一方面所述的方法。 In a sixth aspect, a chip is provided, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method described in the first aspect.
第七方面,提供了一种计算机程序/程序产品,所述计算机程序/程序产品被存储在存储介质中,所述程序/程序产品被至少一个处理器执行以实现如第一方面所述的AI模型指示方法的步骤。In the seventh aspect, a computer program/program product is provided, wherein the computer program/program product is stored in a storage medium, and the program/program product is executed by at least one processor to implement the steps of the AI model indication method as described in the first aspect.
在本申请实施例中,在对AI模型进行指示时,可以通过指示信息指示多个AI模型之间的关联关系,这样,可以通过AI模型间的关联关系确定需要指示的模型信息,从而实现对AI模型的指示。相较于传统的模型指示而言,由于可以无需将全部的模型信息进行指示,因此,可以减少传输资源,避免传输资源的浪费。In the embodiment of the present application, when indicating an AI model, the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model. Compared with the traditional model indication, since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
附图说明BRIEF DESCRIPTION OF THE DRAWINGS
图1是根据本申请实施例的无线通信系统的示意图;FIG1 is a schematic diagram of a wireless communication system according to an embodiment of the present application;
图2是根据本申请实施例的AI模型指示方法的示意性流程图;FIG2 is a schematic flow chart of an AI model indication method according to an embodiment of the present application;
图3是根据本申请实施例的AI模型指示装置的结构示意图;FIG3 is a schematic diagram of the structure of an AI model indication device according to an embodiment of the present application;
图4是根据本申请实施例的通信设备的结构示意图;FIG4 is a schematic diagram of the structure of a communication device according to an embodiment of the present application;
图5是根据本申请实施例的通信设备的结构示意图;FIG5 is a schematic diagram of the structure of a communication device according to an embodiment of the present application;
图6是根据本申请实施例的通信设备的结构示意图。FIG6 is a schematic diagram of the structure of a communication device according to an embodiment of the present application.
具体实施方式DETAILED DESCRIPTION
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员所获得的所有其他实施例,都属于本申请保护的范围。The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of this application.
本申请的术语“第一”、“第二”等是用于区别类似的对象,而不用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换,以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施,且“第一”、“第二”所区别的对象通常为一类,并不限定对象的个数,例如第一对象可以是一个,也可以是多个。此外,本申请中的“或”表示所连接对象的至少其中之一。例如“A或B”涵盖三种方案,即,方案一:包括A且不包括B;方案二:包括B且不包括A;方案三:既包括A又包括B。字符“/”一般表示前后关联对象是一种“或”的关系。The terms "first", "second", etc. of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of one type, and the number of objects is not limited, for example, the first object can be one or more. In addition, "or" in the present application represents at least one of the connected objects. For example, "A or B" covers three schemes, namely, Scheme 1: including A but not including B; Scheme 2: including B but not including A; Scheme 3: including both A and B. The character "/" generally indicates that the objects associated with each other are in an "or" relationship.
本申请的术语“指示”既可以是一个直接的指示(或者说显式的指示),也可以是一个间接的指示(或者说隐含的指示)。其中,直接的指示可以理解为,发送方在发送的指示中明确告知了接收方具体的信息、需要执行的操作或请求结果等内容;间接的指示可以理解为,接收方根据发送方发送的指示确定对应的信息,或者进行判断并根据判断结果确定需要执行的操作或请求结果等。The term "indication" in this application can be a direct indication (or explicit indication) or an indirect indication (or implicit indication). A direct indication can be understood as the sender explicitly informing the receiver of specific information, operations to be performed, or request results in the sent indication; an indirect indication can be understood as the receiver determining the corresponding information according to the indication sent by the sender, or making a judgment and determining the operation to be performed or the request result according to the judgment result.
值得指出的是,本申请实施例所描述的技术不限于长期演进型(Long Term Evolution,LTE)/LTE的演进(LTE-Advanced,LTE-A)系统,还可用于其他无线通信系统,诸如码分多址(Code Division Multiple Access,CDMA)、时分多址(Time Division Multiple Access,TDMA)、频分多址(Frequency Division Multiple Access,FDMA)、正交频分多址(Orthogonal Frequency Division Multiple Access,OFDMA)、 单载波频分多址(Single-carrier Frequency-Division Multiple Access,SC-FDMA)或其他系统。本申请实施例中的术语“系统”和“网络”常被可互换地使用,所描述的技术既可用于以上提及的系统和无线电技术,也可用于其他系统和无线电技术。以下描述出于示例目的描述了新空口(New Radio,NR)系统,并且在以下大部分描述中使用NR术语,但是这些技术也可应用于NR系统以外的系统,如第6代(6th Generation,6G)通信系统。It is worth noting that the technology described in the embodiments of the present application is not limited to the Long Term Evolution (LTE)/LTE-Advanced (LTE-A) system, 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 the embodiments of the present application are often used interchangeably, and the described techniques can be used for the systems and radio technologies mentioned above as well as for other systems and radio technologies. The following description describes a New Radio (NR) system for example purposes, and NR terminology is used in most of the following descriptions, but these techniques may also be applied to systems other than NR systems, such as 6th Generation (6G) communication systems.
图1示出本申请实施例可应用的一种无线通信系统的框图。无线通信系统包括终端11和网络侧设备12。其中,终端11可以是手机、平板电脑(Tablet Personal Computer)、膝上型电脑(Laptop Computer)、笔记本电脑、个人数字助理(Personal Digital Assistant,PDA)、掌上电脑、上网本、超级移动个人计算机(Ultra-mobile Personal Computer,UMPC)、移动上网装置(Mobile Internet Device,MID)、增强现实(Augmented Reality,AR)、虚拟现实(Virtual Reality,VR)设备、机器人、可穿戴式设备(Wearable Device)、飞行器(flight vehicle)、车载设备(Vehicle User Equipment,VUE)、船载设备、行人终端(Pedestrian User Equipment,PUE)、智能家居(具有无线通信功能的家居设备,如冰箱、电视、洗衣机或者家具等)、游戏机、个人计算机(Personal Computer,PC)、柜员机或者自助机等终端侧设备。可穿戴式设备包括:智能手表、智能手环、智能耳机、智能眼镜、智能首饰(智能手镯、智能手链、智能戒指、智能项链、智能脚镯、智能脚链等)、智能腕带、智能服装等。其中,车载设备也可以称为车载终端、车载控制器、车载模块、车载部件、车载芯片或车载单元等。需要说明的是,在本申请实施例并不限定终端11的具体类型。网络侧设备12可以包括接入网设备或核心网设备,其中,接入网设备也可以称为无线接入网(Radio Access Network,RAN)设备、无线接入网功能或无线接入网单元。接入网设备可以包括基站、无线局域网(Wireless Local Area Network,WLAN)接入点(Access Point,AS)或无线保真(Wireless Fidelity,WiFi)节点等。其中,基站可被称为节点B(Node B,NB)、演进节点B(Evolved Node B,eNB)、下一代节点B(the next generation Node B,gNB)、新空口节点B(New Radio Node B,NR Node B)、接入点、中继站(Relay Base Station,RBS)、服务基站(Serving Base Station,SBS)、基收发机站(Base Transceiver Station,BTS)、无线电基站、无线电收发机、基本服务集(Basic Service Set,BSS)、扩展服务集(Extended Service Set,ESS)、家用B节点(home Node B,HNB)、家用演进型B节点(home evolved Node B)、发送接收点(Transmission Reception Point,TRP)或所述领域中其他某个合适的术语,只要达到相同的技术效果,所述基站不限于特定技术词汇,需要说明的是,在本申请实施例中仅以NR系统中的基站为例进行介绍,并不限定基站的具体类型。FIG1 shows a block diagram of a wireless communication system applicable to an embodiment of the present application. The wireless communication system includes a terminal 11 and a network side device 12 . Among them, the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a handheld computer, a netbook, an ultra-mobile personal computer (Ultra-mobile Personal Computer, UMPC), a mobile Internet device (Mobile Internet Device, MID), an augmented reality (Augmented Reality, AR), a virtual reality (Virtual Reality, VR) device, a robot, a wearable device (Wearable Device), a flight vehicle (flight vehicle), a vehicle user equipment (VUE), a shipborne equipment, a pedestrian terminal (Pedestrian User Equipment, PUE), a smart home (home appliances with wireless communication functions, such as refrigerators, televisions, washing machines or furniture, etc.), a game console, a personal computer (Personal Computer, PC), a teller machine or a self-service machine and other terminal side devices. Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among them, the vehicle-mounted device can also be called a vehicle-mounted terminal, a vehicle-mounted controller, a vehicle-mounted module, a vehicle-mounted component, a vehicle-mounted chip or a vehicle-mounted unit, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application. The network side device 12 may include an access network device or a core network device, wherein the access network device may also be called a radio access network (Radio Access Network, RAN) device, a radio access network function or a radio access network unit. The access network device may include a base station, a wireless local area network (Wireless Local Area Network, WLAN) access point (Access Point, AS) or a wireless fidelity (Wireless Fidelity, WiFi) node, etc. Among them, the base station can be called Node B (Node B, NB), Evolved Node B (Evolved Node B, eNB), the next generation Node B (the next generation Node B, gNB), New Radio Node B (New Radio Node B, NR Node B), access point, Relay Base Station (Relay Base Station, RBS), Serving Base Station (Serving Base Station, SBS), Base Transceiver Station (Base Transceiver Station, BTS), radio base station, radio transceiver, base The base station is not limited to specific technical terms as long as the same technical effect is achieved. It should be noted that in the embodiments of the present application, only the base station in the NR system is taken as an example for introduction, and the specific type of the base station is not limited.
目前,在通信网络中,在对AI模型进行指示或传输时,通常会将模型的全部信息或特征指示给需要使用模型的通信设备。然而,模型的全部信息或特征一般会消耗较 多的传输资源,容易造成传输资源的浪费。此外,如果需要指示的模型之间有相同的特征,或本次指示的模型与之前指示的模型有相同的特征,那么,就会出现相同特征重复传输的情况,也容易造成传输资源的浪费。Currently, in communication networks, when an AI model is indicated or transmitted, all the information or features of the model are usually indicated to the communication device that needs to use the model. However, all the information or features of the model generally consumes a lot of space. In addition, if the models to be indicated have the same features, or the model indicated this time has the same features as the model indicated previously, then the same features will be repeatedly transmitted, which can easily lead to a waste of transmission resources.
本申请实施例提供一种AI模型指示方法及通信设备,在对AI模型进行指示时,可以通过指示信息指示多个AI模型之间的关联关系,这样,可以通过AI模型间的关联关系确定需要指示的模型信息,从而实现对AI模型的指示。相较于传统的模型指示而言,由于可以无需将全部的模型信息进行指示,因此,可以减少传输资源,避免传输资源的浪费。The embodiment of the present application provides an AI model indication method and communication device. When indicating an AI model, the association relationship between multiple AI models can be indicated through indication information. In this way, the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model. Compared with traditional model indication, since all model information does not need to be indicated, transmission resources can be reduced and waste of transmission resources can be avoided.
需要说明的是,本申请实施例中的人工智能(Artificial Intelligence,AI)模型包括但不限于神经网络、决策树、支持向量机、贝叶斯分类器等。It should be noted that the artificial intelligence (AI) models in the embodiments of the present application include but are not limited to neural networks, decision trees, support vector machines, Bayesian classifiers, etc.
下面结合附图,通过一些实施例及其应用场景对本申请实施例提供的AI模型指示方法及通信设备进行详细地说明。The following, in combination with the accompanying drawings, describes in detail the AI model indication method and communication device provided in the embodiments of the present application through some embodiments and their application scenarios.
如图2所示,本申请实施例提供一种AI模型指示方法200,该方法可以由通信设备执行,换言之,该AI模型指示方法可以由安装在通信设备中的软件或硬件来执行,该AI模型指示方法包括如下步骤。As shown in Figure 2, an embodiment of the present application provides an AI model indication method 200, which can be executed by a communication device. In other words, the AI model indication method can be executed by software or hardware installed in the communication device. The AI model indication method includes the following steps.
S202:通信设备发送或接收指示信息,指示信息用于指示多个人工智能AI模型之间的关联关系。S202: The communication device sends or receives indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
通信设备在需要向其他通信设备指示AI模型时,通信设备可以向其他通信设备发送指示信息,以向其他通信设备指示多个AI模型之间的关联关系,进而由其他通信设备根据该关联关系确定通信设备指示的模型信息。或者,通信设备在需要由其他通信设备指示AI模型时,可以接收其他通信设备发送的指示信息,以根据该指示信息确定多个AI模型之间的关联关系,进而根据该关联关系确定其他通信设备指示的模型信息。以下实施例中,“多个”指两个及两个以上。When a communication device needs to indicate an AI model to other communication devices, the communication device can send indication information to other communication devices to indicate the association relationship between multiple AI models to other communication devices, and then the other communication devices determine the model information indicated by the communication device based on the association relationship. Alternatively, when the communication device needs to indicate an AI model by other communication devices, it can receive indication information sent by other communication devices to determine the association relationship between multiple AI models based on the indication information, and then determine the model information indicated by other communication devices based on the association relationship. In the following embodiments, "multiple" refers to two or more.
通信设备可以是终端或网络侧设备。在通信设备为终端的情况下,通信设备发送指示信息,可以是向其他终端发送指示信息,通信设备接收指示信息,可以是通信设备从其他终端或网络侧设备接收指示信息。在通信设备为网络侧设备的情况下,通信设备发送指示信息,可以是向终端或其他网络侧设备发送指示信息,通信设备接收指示信息,可以是通信设备从其他网络侧设备接收指示信息。The communication device may be a terminal or a network side device. In the case where the communication device is a terminal, the communication device sending the indication information may be sending the indication information to other terminals, and the communication device receiving the indication information may be receiving the indication information from other terminals or network side devices. In the case where the communication device is a network side device, the communication device sending the indication information may be sending the indication information to a terminal or other network side device, and the communication device receiving the indication information may be receiving the indication information from other network side devices.
多个AI模型可以是待指示的模型,也可以是一部分已经指示(或已预先配置),另一部分未指示,还可以是均已进行指示(或已预先配置)。通过指示多个AI模型之间的关联关系,可以便于接收到指示信息的通信设备根据该关联关系确定(或构建)多个AI模型,或多个AI模型中的部分模型、或与多个AI模型相关的其他模型。The multiple AI models may be models to be indicated, or some may have been indicated (or pre-configured) and others may not have been indicated, or all may have been indicated (or pre-configured). By indicating the association relationship between the multiple AI models, it is convenient for the communication device that receives the indication information to determine (or construct) the multiple AI models, or some of the multiple AI models, or other models related to the multiple AI models according to the association relationship.
以通信设备发送指示信息为例,多个AI模型可以是通信设备需要指示给其他通信设备的模型,在这种情况下,通过发送指示信息可以在后续指示该多个AI模型时避免重复传输相同的模型信息,从而减少传输资源,其他通信设备在接收到指示信息后, 可以根据指示信息构建多个AI模型;或者,多个AI模型也可以是其中一部分AI模型已指示给了其他通信设备(或者是其他通信设备中已预先配置),另一部分AI模型还未指示,在这种情况下,通过发送指示信息,可以在后续传输另一部分AI模型时避免重复传输相同的模型信息,从而减少传输资源,其他通信设备在接收到指示信息后,可以根据指示信息以及已指示(或已配置)的模型构建另一部分AI模型;或者,多个AI模型也可以是均已指示给了其他通信设备(或者是其他通信设备中已预先配置),在这种情况下,其他通信设备可以根据指示信息以及已指示(或已配置)的多个AI模型构建新的模型,这样,通信设备可以无需指示该新的模型,从而减少传输资源。Taking the communication device sending indication information as an example, multiple AI models may be models that the communication device needs to indicate to other communication devices. In this case, by sending the indication information, the same model information can be avoided from being repeatedly transmitted when the multiple AI models are subsequently indicated, thereby reducing transmission resources. After receiving the indication information, other communication devices Multiple AI models can be constructed according to the indication information; or, multiple AI models may be that some of the AI models have been indicated to other communication devices (or have been pre-configured in other communication devices), and the other AI models have not yet been indicated. In this case, by sending the indication information, the same model information can be avoided from being repeatedly transmitted when the other part of the AI model is subsequently transmitted, thereby reducing transmission resources. After receiving the indication information, the other communication devices can construct another part of the AI model according to the indication information and the indicated (or configured) model; or, multiple AI models may be that all of them have been indicated to other communication devices (or have been pre-configured in other communication devices). In this case, the other communication devices can construct a new model according to the indication information and the multiple AI models that have been indicated (or configured). In this way, the communication device does not need to indicate the new model, thereby reducing transmission resources.
多个AI模型之间的关联关系可以是多个AI模型的模型特征之间的关联关系。通过指示AI模型之间的关联关系,可以知晓AI模型之间的哪些特征相关联。此外,通过指示AI模型之间的关联关系,也能够指示AI模型所关联的任务/功能之间的关联关系。The association relationship between multiple AI models can be the association relationship between the model features of multiple AI models. By indicating the association relationship between AI models, it can be known which features between AI models are associated. In addition, by indicating the association relationship between AI models, it is also possible to indicate the association relationship between the tasks/functions associated with the AI models.
可选地,多个AI模型之间的关联关系可以包括以下十种关联关系中的至少一项:Optionally, the association relationship between the multiple AI models may include at least one of the following ten association relationships:
(1)第一关联关系,第一关联关系表征多个AI模型的模型输入相同。(1) A first association relationship, where the first association relationship indicates that the model inputs of multiple AI models are the same.
模型输入相同可以包括模型输入的类型和/或模型输入的格式相同。模型输入的类型可以是模型输入具体是什么信息,比如,时域信道脉冲响应等。模型输入的格式可以是模型输入的排列方式等,比如,模型输入为时域信道脉冲响应的情况下,模型输入的格式可以是每个传输接收点(Transmission and Reception Point,TRP)关联的载波干扰比(Carrier to Interference Ratio,CIR)采样点数,CIR的排列方式、不同TRP ID的时域信道脉冲响应的排列方式、时域信道脉冲响应的采样点数等。The same model input may include the same type of model input and/or the same format of model input. The type of model input may be the specific information of the model input, such as a time domain channel impulse response. The format of the model input may be the arrangement of the model input, etc. For example, when the model input is a time domain channel impulse response, the format of the model input may be the number of carrier to interference ratio (CIR) sampling points associated with each transmission and reception point (TRP), the arrangement of CIR, the arrangement of time domain channel impulse responses of different TRP IDs, the number of sampling points of the time domain channel impulse response, etc.
以模型A和模型B为例,在基于AI的模型监督中,模型A为定位模型,输入为多TRP的信道脉冲响应,输出为UE位置,模型B用于监督模型A的有效性,输入也为多TRP的信道脉冲响应,输出为当前输入下模型A有效的置信度,那么,模型A和模型B的模型输入相同。Taking Model A and Model B as examples, in AI-based model supervision, Model A is a positioning model, the input is the channel impulse response of multiple TRPs, and the output is the UE position. Model B is used to supervise the effectiveness of Model A, and the input is also the channel impulse response of multiple TRPs, and the output is the confidence that Model A is valid under the current input. Therefore, the model inputs of Model A and Model B are the same.
通过指示多个AI模型的模型输入相同,可以根据其中一个AI模型的模型输入确定其他AI模型的模型输入,这样,可以无需对其他AI模型的模型输入进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model inputs of multiple AI models are the same, the model inputs of other AI models can be determined based on the model inputs of one of the AI models. In this way, there is no need to indicate the model inputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将具有相同模型输入的模型A和模型B指示给设备2,那么,设备1可以将模型A指示给设备2,同时指示模型A和模型B的模型输入相同,设备2可以根据模型B确定模型A的模型输入,这样,设备1就不需要指示模型B的模型输入了,从而可以避免重复传输相同的模型输入,减少传输资源,避免传输资源的浪费。For example, if device 1 wants to indicate model A and model B with the same model input to device 2, then device 1 can indicate model A to device 2 and indicate that the model inputs of model A and model B are the same. Device 2 can determine the model input of model A based on model B. In this way, device 1 does not need to indicate the model input of model B, thereby avoiding repeated transmission of the same model input, reducing transmission resources, and avoiding waste of transmission resources.
可选地,在指示信息指示第一关联关系的情况下,即指示多个AI模型的模型输入相同的情况下,指示信息还可以进一步指示多个AI模型的除模型输入以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模 型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates the first association relationship, that is, when indicating that the model inputs of the multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model inputs, so that the model information of one of the AI models and the model information of the multiple AI models can be used to distinguish the differences between the model information of the multiple AI models. The distinction between the model information determines the model information of the remaining AI models.
(2)第二关联关系,第二关联关系表征多个AI模型的模型输出相同。(2) A second association relationship, where the second association relationship indicates that the model outputs of multiple AI models are the same.
模型输出相同可以包括模型输出的类型和/或模型输出的格式相同。模型输出的类型可以是模型输出具体是什么信息,比如,位置坐标等。模型输出的格式可以是模型输出的维度等,比如,在模型输出为坐标位置的情况下,模型输出的格式可以是第一个坐标位置表示第一维度,第二个坐标位置表示第二维度。The same model output may include the same type of model output and/or the same format of model output. The type of model output may be what information the model output specifically is, such as position coordinates, etc. The format of model output may be the dimension of model output, etc. For example, when the model output is a coordinate position, the format of the model output may be that the first coordinate position represents the first dimension, and the second coordinate position represents the second dimension.
以模型A和模型B为例,模型A是场景A的定位模型,模型B是场景B的定位模型,模型A的输入是场景A下N个TRP的CIR信息,输出是位置,模型B是场景B下M个TRP的CIR信息,输出也是位置,则模型A和模型B的模型输出相同。Taking model A and model B as examples, model A is the positioning model of scene A, and model B is the positioning model of scene B. The input of model A is the CIR information of N TRPs in scene A, and the output is the position. The input of model B is the CIR information of M TRPs in scene B, and the output is also the position. Therefore, the model outputs of model A and model B are the same.
通过指示多个AI模型的模型输出相同,可以根据其中一个AI模型的模型输出确定其他AI模型的模型输出,这样,可以无需对其他AI模型的模型输出进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model outputs of multiple AI models are the same, the model outputs of other AI models can be determined based on the model output of one of the AI models. In this way, there is no need to indicate the model outputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备2中已预先配置了模型B,设备1想要将模型A指示给设备2,且模型A的模型输出和模型B的模型输出相同,那么,设备1在向设备2指示模型A时,可以指示模型A和模型B的模型输出相同,设备2可以根据模型B确定模型A的模型输出,这样,设备1就不需要指示模型A的模型输出了,从而可以减少传输资源,避免传输资源的浪费。For example, model B has been pre-configured in device 2, and device 1 wants to indicate model A to device 2, and the model output of model A is the same as the model output of model B. Then, when device 1 indicates model A to device 2, it can indicate that the model outputs of model A and model B are the same, and device 2 can determine the model output of model A based on model B. In this way, device 1 does not need to indicate the model output of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示第二关联关系的情况下,即指示多个AI模型的模型输出相同的情况下,指示信息还可以进一步指示多个AI模型的除模型输出以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates a second association relationship, that is, indicating that the model outputs of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model outputs, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
(3)第三关联关系,第三关联关系表征多个AI模型的模型输入和模型输出均相同。(3) The third association relationship: The third association relationship represents that the model input and model output of multiple AI models are the same.
模型输入相同包括模型输入的类型和/或模型输入的格式相同。模型输出相同可以包括模型输出的类型和/或模型输出的格式相同。The same model inputs may include the same model input types and/or the same model input formats. The same model outputs may include the same model output types and/or the same model output formats.
以模型A和模型B为例,模型A和B是两个不同复杂度的定位模型,两个模型的中间层的结构不同,但输入均为N个TRP的CIR信息,输出均为位置信息,则模型A和模型B的模型输入和模型输出均相同。Taking Model A and Model B as examples, Model A and Model B are two positioning models of different complexity. The structures of the middle layers of the two models are different, but the input is the CIR information of N TRPs, and the output is the location information. Therefore, the model input and model output of Model A and Model B are the same.
通过指示多个AI模型的模型输入和模型输出均相同,可以根据其中一个AI模型的模型输入和输出确定其他AI模型的模型输入和输出,这样,可以无需对其他AI模型的模型输入和输出进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model inputs and model outputs of multiple AI models are the same, the model inputs and outputs of other AI models can be determined based on the model inputs and outputs of one of the AI models. In this way, there is no need to indicate the model inputs and outputs of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将具有相同的模型输入和输出的模型A和模型B指示给设备2,那么,设备1可以将模型A指示给设备2,同时指示模型A和模型B的模型输入和输出均相同,设备2可以根据模型B确定模型A的模型输入和输出,这样,设备1在指 示模型B时,就不需要指示模型B的模型输入和输出了,从而可以避免重复传输相同的模型输入和输出,减少传输资源,避免传输资源的浪费。For example, if device 1 wants to indicate model A and model B with the same model input and output to device 2, then device 1 can indicate model A to device 2 and indicate that the model input and output of model A and model B are the same. Device 2 can determine the model input and output of model A based on model B. In this way, device 1 can indicate model A to device 2. When indicating model A, there is no need to indicate the model input and output of model B, thereby avoiding repeated transmission of the same model input and output, reducing transmission resources, and avoiding waste of transmission resources.
可选地,在指示信息指示第三关联关系的情况下,即指示多个AI模型的模型输入和输出均相同的情况下,指示信息还可以进一步指示多个AI模型的除模型输入和输出以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates a third association relationship, that is, indicating that the model inputs and outputs of multiple AI models are the same, the indication information may further indicate the differences between other model information of the multiple AI models except for the model inputs and outputs. In this way, the model information of the remaining AI models can be determined based on the model information of one of the AI models and the differences between the model information of multiple AI models.
(4)第四关联关系,第四关联关系表征多个AI模型的模型结构相同。(4) A fourth association relationship, wherein the fourth association relationship indicates that the model structures of multiple AI models are the same.
模型结构相同包括但不限于模型层的数量相同、和/或模型层中包括的元素数量相同、和/或模型层的类型相同、和/或模型层中激活函数(若有)的类型相同。比如,以多个AI模型为神经网络模型为例,多个AI模型的模型结构相同可以包括以下至少一项:The same model structure includes but is not limited to the same number of model layers, and/or the same number of elements included in the model layers, and/or the same type of model layers, and/or the same type of activation functions (if any) in the model layers. For example, taking multiple AI models as neural network models as an example, the same model structure of multiple AI models may include at least one of the following:
神经元的层数相同;The number of layers of neurons is the same;
每层神经元中包括的神经元的数量相同;The number of neurons included in each layer is the same;
每层神经元中包括的激活函数的类型相同;The type of activation function included in each layer of neurons is the same;
每层神经元的类型(比如,批归一化层、层归一化层、卷积层等)相同。The type of neurons in each layer (e.g., batch normalization layer, layer normalization layer, convolutional layer, etc.) is the same.
以模型A和模型B为例,对于AI定位,网络侧设备会为UE部署一个包含多个AI模型的模型池,模型A和模型B均为模型池中的模型,功能都是定位,包含的卷积层数量相同,但是模型A和模型B对应不同的使用场景,卷积层的参数不同,如不同的定位参考信号(Positioning reference signal,PRS)配置,则模型A和模型B具有相同的结构。Taking Model A and Model B as an example, for AI positioning, the network side device will deploy a model pool containing multiple AI models for the UE. Model A and Model B are both models in the model pool, and their functions are positioning. They contain the same number of convolutional layers, but Model A and Model B correspond to different usage scenarios and have different parameters of the convolutional layers, such as different positioning reference signal (PRS) configurations. Then Model A and Model B have the same structure.
通过指示多个AI模型的模型结构相同,可以根据其中一个AI模型的模型结构确定其他AI模型的结构,这样,可以无需对其他AI模型的模型结构进行指示,即在传输模型时,不需要传输其他AI模型的模型结构,只需要传输模型结构对应的参数即可,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model structures of multiple AI models are the same, the structures of other AI models can be determined based on the model structure of one of the AI models. In this way, there is no need to indicate the model structures of other AI models. That is, when transmitting the model, there is no need to transmit the model structures of other AI models, but only the parameters corresponding to the model structure. This can reduce the transmission resources consumed when indicating the model and avoid waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A和模型B的模型结构相同,那么,设备1在指示模型A时,可以指示模型A和模型B的模型结构相同,设备2可以根据模型B确定模型A的模型结构,这样,设备1就不需要传输模型A的模型结构信息了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. Model A and model B have the same model structure. Then, when device 1 indicates model A, it can indicate that the model structures of model A and model B are the same. Device 2 can determine the model structure of model A based on model B. In this way, device 1 does not need to transmit the model structure information of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
再比如,设备1将一个模型池发送给设备2,模型池包含多个模型,这些模型的模型结构相同,只有模型参数不同,如果设备2中配置的模型失效,则设备1可以通过仅配置参数来更新失效的模型。For another example, device 1 sends a model pool to device 2. The model pool contains multiple models. These models have the same model structure but different model parameters. If the model configured in device 2 fails, device 1 can update the failed model by only configuring the parameters.
可选地,在指示信息指示第四关联关系的情况下,即指示多个AI模型的模型结构相同的情况下,指示信息还可以进一步指示多个AI模型的除模型结构以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。 Optionally, when the indication information indicates a fourth association relationship, that is, indicating that the model structures of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model structure, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
(5)第五关联关系,第五关联关系表征多个AI模型的部分模型结构相同。(5) The fifth association relationship: The fifth association relationship represents that some model structures of multiple AI models are the same.
多个AI模型的部分模型结构相同,即多个AI模型的模型结构有相同的部分也有不同的部分。比如,以两个神经网络模型A和模型B为例,模型A和模型B的部分结构相同可以是:Multiple AI models have the same partial model structure, that is, the model structures of multiple AI models have both common and different parts. For example, taking two neural network models A and B as examples, the common partial structures of model A and model B can be:
模型A的前N1个模型层的结构和模型B的前N1个模型层的结构相同,N1小于模型A的总层数M1,也小于模型B的总层数M2;或者,The structure of the first N1 model layers of model A is the same as the structure of the first N1 model layers of model B, and N1 is less than the total number of layers M1 of model A and less than the total number of layers M2 of model B; or,
模型A的模型层数量与模型B的模型层的数量相同,但模型A的每个模型层中的神经元数量小于模型B的每个模型层中的神经元数量。The number of model layers of model A is the same as the number of model layers of model B, but the number of neurons in each model layer of model A is smaller than the number of neurons in each model layer of model B.
当然,模型A和模型B的部分结构相同也可以是除上述三种情况外的其他情况,这里不再一一举例说明。Of course, the parts of the structures of Model A and Model B may be the same in other cases besides the above three cases, which will not be explained one by one here.
对应到具体的应用场景,比如可以是,在基于AI的模型监督中,模型A为定位模型,输入为多TRP的信道脉冲响应,输出为UE位置,模型B用于监督模型A的有效性,输入也为多TRP的信道脉冲响应,输出为当前输入下模型A有效的置信度,模型B和模型A之间只有最后一层结构不同,如将模型B的最后一层神经元数量由模型A的2维改为1维度,激活函数由linear改为sigmoid,则,模型A和模型B的部分结构相同。其中,模型A和模型B也具有上述第一关联关系和第二关联关系。Corresponding to specific application scenarios, for example, in AI-based model supervision, model A is a positioning model, the input is the channel impulse response of multiple TRPs, and the output is the UE position. Model B is used to supervise the effectiveness of model A, and the input is also the channel impulse response of multiple TRPs, and the output is the confidence that model A is effective under the current input. Only the last layer structure is different between model B and model A. For example, the number of neurons in the last layer of model B is changed from 2 dimensions of model A to 1 dimension, and the activation function is changed from linear to sigmoid. Then, some structures of model A and model B are the same. Among them, model A and model B also have the above-mentioned first association relationship and second association relationship.
通过指示多个AI模型的模型部分结构相同,可以根据其中一个AI模型的模型结构确定其他AI模型相同的部分结构,这样,可以无需对其他AI模型的相同的模型结构部分进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model parts of multiple AI models are the same, the same partial structures of other AI models can be determined based on the model structure of one of the AI models. In this way, there is no need to indicate the same model structure parts of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A和模型B的部分模型结构相同,那么,设备1在指示模型A时,可以指示模型A和模型B的部分模型结构相同,设备2可以根据模型B确定模型A的部分模型结构,这样,设备1就不需要传输模型A中与模型B相同的模型结构信息了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. Model A and model B have some identical model structures. Then, when device 1 indicates model A, it can indicate that model A and model B have some identical model structures. Device 2 can determine the partial model structure of model A based on model B. In this way, device 1 does not need to transmit the model structure information in model A that is identical to that in model B, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示第五关联关系的情况下,即指示多个AI模型的部分模型结构相同的情况下,指示信息还可以进一步指示多个AI模型的除部分模型结构以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates the fifth association relationship, that is, indicating that some model structures of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except for the partial model structure, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
比如,在指示信息指示了第五关联关系的情况下,指示信息还可以包括多个AI模型的模型结构的区别。这样,可以根据其中一个AI模型的模型结构以及指示信息指示的多个模型的结构中相同的部分和不同的部分,确定其余模型的模型结构。For example, when the indication information indicates the fifth association relationship, the indication information may also include the differences in the model structures of the multiple AI models. In this way, the model structures of the remaining models may be determined based on the model structure of one of the AI models and the same and different parts of the structures of the multiple models indicated by the indication information.
仍以上述模型A和模型B为例,若指示信息指示了模型A的前N1个模型层的结构和模型B的前N1个模型层的结构相同,则指示信息还可以指示模型A的后(M1-N1)个模型层与模型B的后(M2-N1)个模型层的区别,比如,模型A的后(M1-N1)个模型层中均包含1个神经元,模型B的后(M2-N1)个模型层中均包含2个神经元。 这样,在已知模型A的结构的情况下,根据指示信息可以确定模型B的结构,或者,在已知模型B的结构的情况下,根据指示信息可以确定模型A的结构。Still taking the above-mentioned model A and model B as an example, if the indication information indicates that the structure of the first N1 model layers of model A is the same as the structure of the first N1 model layers of model B, the indication information can also indicate the difference between the last (M1-N1) model layers of model A and the last (M2-N1) model layers of model B. For example, the last (M1-N1) model layers of model A each contain 1 neuron, and the last (M2-N1) model layers of model B each contain 2 neurons. In this way, when the structure of model A is known, the structure of model B can be determined according to the indication information, or, when the structure of model B is known, the structure of model A can be determined according to the indication information.
(6)第六关联关系,第六关联关系表征多个AI模型的部分模型结构和对应参数均相同。(6) The sixth association relationship: The sixth association relationship represents that some model structures and corresponding parameters of multiple AI models are the same.
与上述第五关联关系不同的是,第六关联关系除了指示多个AI模型的部分结构相同外,还进一步指示了该相同的部分结构所对应的参数相同。这里的参数可以是模型结构对应的模型参数,比如,在模型为神经网络模型的情况下,模型结构对应的参数可以是模型结构的权重和偏置等。Different from the fifth association relationship, the sixth association relationship not only indicates that the partial structures of multiple AI models are the same, but also further indicates that the parameters corresponding to the same partial structures are the same. The parameters here can be model parameters corresponding to the model structure. For example, when the model is a neural network model, the parameters corresponding to the model structure can be the weights and biases of the model structure.
以迁移学习场景为例,在迁移学习中,在一个场景训练的AI模型可以迁移到另一个具有相似性的场景,在新的场景下,需要用新场景的数据重新训练或微调模型参数,得到新的AI模型,此时可以保留原模型的前N层的模型结构和参数不变,只改变原模型的后M层的结构及对应的参数,在这种情况下,原模型和重新训练的新模型之间的前N层结构和对应参数是相同的。Taking the transfer learning scenario as an example, in transfer learning, an AI model trained in one scenario can be migrated to another similar scenario. In the new scenario, it is necessary to retrain or fine-tune the model parameters using the data from the new scenario to obtain a new AI model. At this time, the model structure and parameters of the first N layers of the original model can be kept unchanged, and only the structure and corresponding parameters of the last M layers of the original model are changed. In this case, the first N layer structures and corresponding parameters between the original model and the retrained new model are the same.
通过指示多个AI模型的模型部分结构及对应参数相同,可以根据其中一个AI模型的模型结构和对应的参数,确定其他AI模型相同的部分结构和对应参数,这样,可以无需对其他AI模型的相同的模型结构部分及对应参数进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model partial structures and corresponding parameters of multiple AI models are the same, the same partial structures and corresponding parameters of other AI models can be determined based on the model structure and corresponding parameters of one of the AI models. In this way, there is no need to indicate the same model structure parts and corresponding parameters of other AI models, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A和模型B的部分模型结构和对应参数均相同,那么,设备1在指示模型A时,可以指示模型A和模型B部分模型结构和对应参数均相同,设备2可以根据模型B确定模型A的部分模型结构和对应参数,这样,设备1就不需要传输模型A中与模型B相同的模型结构和对应参数的信息了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. Some model structures and corresponding parameters of model A and model B are the same. Then, when device 1 indicates model A, it can indicate that some model structures and corresponding parameters of model A and model B are the same. Device 2 can determine some model structures and corresponding parameters of model A based on model B. In this way, device 1 does not need to transmit the information of model structure and corresponding parameters in model A that are the same as those in model B, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示第六关联关系的情况下,即指示多个AI模型的部分模型结构和对应参数相同的情况下,指示信息还可以进一步指示多个AI模型的除部分模型结构和对应参数以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates the sixth association relationship, that is, indicating that some model structures and corresponding parameters of multiple AI models are the same, the indication information may further indicate the difference between other model information of multiple AI models except for the partial model structure and corresponding parameters. In this way, the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
比如,在指示信息指示了第六关联关系的情况下,指示信息还可以包括多个AI模型的模型结构的区别。这样,可以根据其中一个AI模型的模型结构以及指示信息指示的多个模型的结构中相同的部分和不同的部分,确定其余模型的模型结构。For example, when the indication information indicates the sixth association relationship, the indication information may also include the differences in the model structures of the multiple AI models. In this way, the model structures of the remaining models may be determined based on the model structure of one of the AI models and the same and different parts of the structures of the multiple models indicated by the indication information.
(7)第七关联关系,第七关联关系表征多个AI模型的模型结构和对应参数均相同。(7) The seventh relationship: The seventh relationship represents that the model structures and corresponding parameters of multiple AI models are the same.
多个AI模型的模型结构和对应参数均相同,可以认为多个AI模型是相同的模型。比如,在同一场景的多个UE,他们对应的模型可能是完全相同的。If the model structures and corresponding parameters of multiple AI models are the same, then the multiple AI models can be considered to be the same model. For example, the models corresponding to multiple UEs in the same scene may be exactly the same.
通过指示多个AI模型的模型结构及对应参数均相同,可以根据其中一个AI模型的模型结构和对应的参数,确定其他AI模型的结构和对应参数,这样,可以无需对其 他AI模型的结构及对应参数进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the model structures and corresponding parameters of multiple AI models are the same, the structures and corresponding parameters of other AI models can be determined based on the model structure and corresponding parameters of one AI model, so that there is no need to The structure and corresponding parameters of other AI models can be indicated, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A和模型B的模型结构和对应参数均相同,那么,设备1在指示模型A时,可以指示模型A和模型B的模型结构和对应参数均相同,设备2可以根据模型B确定模型A的模型结构和对应参数,这样,设备1就不需要传输模型A的模型结构和对应参数的信息了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. The model structures and corresponding parameters of model A and model B are the same. Then, when device 1 indicates model A, it can indicate that the model structures and corresponding parameters of model A and model B are the same. Device 2 can determine the model structure and corresponding parameters of model A based on model B. In this way, device 1 does not need to transmit the information of the model structure and corresponding parameters of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示第七关联关系的情况下,即指示多个AI模型的模型结构和对应参数相同的情况下,指示信息还可以进一步指示多个AI模型的除模型结构和对应参数以外的其他模型信息之间的区别,这样,可以根据其中一个AI模型的模型信息以及多个AI模型的模型信息之间的区别确定其余AI模型的模型信息。Optionally, when the indication information indicates the seventh association relationship, that is, indicating that the model structures and corresponding parameters of multiple AI models are the same, the indication information may further indicate the difference between other model information of the multiple AI models except the model structure and corresponding parameters, so that the model information of the remaining AI models can be determined based on the model information of one of the AI models and the difference between the model information of multiple AI models.
(8)第八关联关系,第八关联关系表征多个AI模型中的第一模型是多个AI模型中的第二模型的子模型。(8) An eighth association relationship, the eighth association relationship represents that the first model among the multiple AI models is a sub-model of the second model among the multiple AI models.
第一模型是第二模型的子模型,可以认为是第一模型是第二模型中的一部分模型。比如,以模型A和模型B为例,模型A可以是模型B的前N3个模型层,该前N3个模型层的结构和对应参数均相同。其中,N3小于模型B的总模型层数。The first model is a sub-model of the second model, which can be considered as a part of the second model. For example, taking model A and model B as examples, model A can be the first N3 model layers of model B, and the structures and corresponding parameters of the first N3 model layers are the same. Among them, N3 is less than the total number of model layers of model B.
以信道状态信息(Channel State Information,CSI)压缩为例,模型B是auto-encoder模型,共有T层,其中包含encoder N层,decoder M层,T=M+N-1,模型B训练好之后,模型A是它的前N+1层encoder部分或后M+1层decoder部分,则模型A是模型B的子模型。Taking Channel State Information (CSI) compression as an example, model B is an auto-encoder model with a total of T layers, including N encoder layers and M decoder layers, T=M+N-1. After model B is trained, model A is its first N+1 layers of encoder part or the last M+1 layers of decoder part, then model A is a submodel of model B.
通过指示多个AI模型中的第一模型是第二模型的子模型,可以根据第二模型确定第一模型,这样可以无需对第一模型进行指示,或者,也可以根据第一模型确定第二模型中的部分结构和参数,这样可以无需指示第二模型中与第一模型相同的部分,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the first model among multiple AI models is a sub-model of the second model, the first model can be determined according to the second model, so there is no need to indicate the first model. Alternatively, some structures and parameters in the second model can be determined according to the first model, so there is no need to indicate the parts of the second model that are the same as the first model, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A是模型B的子模型,那么,设备1在指示模型A时,可以指示模型A是模型B的子模型,设备2可以根据模型B确定模型A,这样,设备1就不需要传输模型A了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2, and model A is a sub-model of model B. Then, when device 1 indicates model A, it can indicate that model A is a sub-model of model B. Device 2 can determine model A based on model B. In this way, device 1 does not need to transmit model A, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示了第八关联关系的情况下,指示信息还可以包括第二模型中的第一部分,第一部分的结构与第一模型的结构相同,第一部分的参数与第一模型的参数相同。也就是说,在指示信息指示了第八关联关系的情况下,还可以指示第一模型是第二模型的哪一部分,该部分即为第一部分。比如,第一模型是第二模型的前N层、encoder,或者是第二模型的后M层、decoder等。这样,可以根据第二模型确定第一模型的具体结构和具体参数,或者,根据第一模型确定第二模型中与第一模型相同的那部分模型的结构和对应参数。 Optionally, when the indication information indicates the eighth association relationship, the indication information may also include the first part in the second model, the structure of the first part is the same as the structure of the first model, and the parameters of the first part are the same as the parameters of the first model. That is to say, when the indication information indicates the eighth association relationship, it may also indicate which part of the second model the first model is, and the part is the first part. For example, the first model is the first N layers of the second model, the encoder, or the last M layers of the second model, the decoder, etc. In this way, the specific structure and specific parameters of the first model can be determined according to the second model, or the structure and corresponding parameters of the part of the second model that is the same as the first model can be determined according to the first model.
可选地,在指示信息指示了第八关联关系的情况下,指示信息还可以包括第二模型中与第一模型不同的模型部分,剩余部分即为与第一模型相同的部分。这样,也可以根据第二模型确定第一模型的具体结构和具体参数,或者,根据第一模型确定第二模型。Optionally, when the indication information indicates the eighth association relationship, the indication information may further include a model portion of the second model that is different from the first model, and the remaining portion is the same as the first model. In this way, the specific structure and specific parameters of the first model may also be determined based on the second model, or the second model may be determined based on the first model.
(9)第九关联关系,第九关联关系表征多个AI模型中的第三模型的结构是多个AI模型中的第四模型的子结构。(9) A ninth association relationship, wherein the ninth association relationship represents that the structure of the third model among the multiple AI models is a substructure of the fourth model among the multiple AI models.
第三模型是第四模型的子结构,可以认为是第三模型与第四模型中的一部分模型的结构相同,但对应的参数不同。比如,以模型A和模型B为例,模型A的结构与模型B的前N3个模型层的结构相同,但模型A的参数与模型B的前N3个模型层的参数不同。其中,N3小于模型B的总模型层数。一种可能的应用场景可以是,截选复杂模型B结构的一部分作为模型A的结构,以降低模型复杂度。The third model is a substructure of the fourth model. It can be considered that the third model has the same structure as a part of the fourth model, but the corresponding parameters are different. For example, taking model A and model B as examples, the structure of model A is the same as the structure of the first N3 model layers of model B, but the parameters of model A are different from the parameters of the first N3 model layers of model B. Among them, N3 is less than the total number of model layers of model B. A possible application scenario may be to select a part of the complex model B structure as the structure of model A to reduce the complexity of the model.
通过指示多个AI模型中的第三模型是第四模型的子结构,可以根据第四模型的结构确定第三模型的结构,这样可以无需对第三模型的结构进行指示,或者,也可以根据第三模型确定第四模型中的部分结构,这样可以无需指示第四模型中与第三模型的结构相同的部分,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the third model among multiple AI models is a substructure of the fourth model, the structure of the third model can be determined based on the structure of the fourth model, so that there is no need to indicate the structure of the third model. Alternatively, part of the structure in the fourth model can be determined based on the third model, so that there is no need to indicate the part of the fourth model that has the same structure as the third model, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型A指示给设备2,设备2中预先配置了模型B,模型A的结构是模型B的子结构,那么,设备1在指示模型A时,可以指示模型A是模型B的子结构,设备2根据模型B的结构确定模型A的结构,这样,设备1就不需要传输模型A的结构信息了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model A to device 2, and model B is pre-configured in device 2. The structure of model A is a substructure of model B. Then, when device 1 indicates model A, it can indicate that model A is a substructure of model B. Device 2 determines the structure of model A based on the structure of model B. In this way, device 1 does not need to transmit the structural information of model A, thereby reducing transmission resources and avoiding waste of transmission resources.
可选地,在指示信息指示了第九关联关系的情况下,指示信息还可以用于第四模型中的第二部分,第二部分的结构与第三模型的结构相同,第二部分的参数与第三模型的参数不同。也就是说,在指示信息指示了第九关联关系的情况下,还可以指示第三模型是第四模型的哪一部分,该部分即为第二部分。比如,第三模型是第四模型的前N层、encoder,或者是第四模型的后M层、decoder等,但对应的参数是不同的。这样,可以根据第四模型确定第三模型的具体结构,或者,根据第三模型的结构确定第四模型中与第三模型的结构相同的那部分模型的结构。Optionally, when the indication information indicates the ninth association relationship, the indication information can also be used for the second part in the fourth model, the structure of the second part is the same as the structure of the third model, and the parameters of the second part are different from the parameters of the third model. That is to say, when the indication information indicates the ninth association relationship, it can also indicate which part of the fourth model the third model is, and this part is the second part. For example, the third model is the first N layers, encoder of the fourth model, or the last M layers, decoder, etc. of the fourth model, but the corresponding parameters are different. In this way, the specific structure of the third model can be determined according to the fourth model, or the structure of the part of the model in the fourth model that has the same structure as the third model can be determined according to the structure of the third model.
可选地,在指示信息指示了第九关联关系的情况下,指示信息还可以包括第四模型中与第三模型不同的结构部分,剩余结构部分即为与第三模型相同的部分。这样,也可以根据第四模型确定第三模型的具体结构,或者,根据第三模型的结构确定第四模型的结构。Optionally, when the indication information indicates the ninth association relationship, the indication information may further include structural parts of the fourth model that are different from the third model, and the remaining structural parts are the same as the third model. In this way, the specific structure of the third model may be determined based on the fourth model, or the structure of the fourth model may be determined based on the structure of the third model.
(10)第十关联关系,第十关联关系表征多个模型中的第五模型的输出是多个模型中的第六模型的输入。(10) The tenth association relationship, the tenth association relationship represents that the output of the fifth model among the multiple models is the input of the sixth model among the multiple models.
第五模型是第六模型的输入,可以认为第五模型和第六模型是级联关系,第五模型的输出可以作为第六模型的输入。比如,对CSI进行预测和压缩的场景中,原始的CSI可以先经过模型A进行CSI预测后,再输入到模型B进行压缩,则,模型A和模 型B是级联关系,模型A的输出是模型B的输入。The fifth model is the input of the sixth model. It can be considered that the fifth model and the sixth model are in a cascade relationship, and the output of the fifth model can be used as the input of the sixth model. For example, in the scenario of predicting and compressing CSI, the original CSI can be first predicted by model A and then input to model B for compression. In this case, model A and model B are Model B is a cascade relationship, and the output of model A is the input of model B.
通过指示第五模型的输出是第六模型的输入,可以确定第五模型和第六模型的级联关系,进而基于该级联关系使用第五模型和第六模型,无需将第五模型和第六模型按照该级联关系组合后的模型进行指示,从而可以减少模型指示时消耗的传输资源,避免传输资源浪费。By indicating that the output of the fifth model is the input of the sixth model, the cascade relationship between the fifth model and the sixth model can be determined, and then the fifth model and the sixth model can be used based on the cascade relationship. There is no need to indicate the fifth model and the sixth model according to the model after combining the cascade relationship, thereby reducing the transmission resources consumed during model indication and avoiding waste of transmission resources.
比如,设备1想要将模型C指示给设备2,设备2中预先配置了模型A和模型B,模型C为模型A和模型B的组合,且模型A的模型输出是模型B的模型输入,那么,设备1在指示模型C时,可以指示模型A的模型输出是模型B的模型输入,设备2可以根据设备1的指示和模型A、模型B确定模型C,这样,设备1就不需要传输模型C了,从而可以减少传输资源,避免传输资源的浪费。For example, device 1 wants to indicate model C to device 2. Model A and model B are pre-configured in device 2. Model C is a combination of model A and model B, and the model output of model A is the model input of model B. Then, when device 1 indicates model C, it can indicate that the model output of model A is the model input of model B. Device 2 can determine model C based on the instruction of device 1 and models A and B. In this way, device 1 does not need to transmit model C, thereby reducing transmission resources and avoiding waste of transmission resources.
在上述十种关联关系中,指示信息可以指示其中任一种或多种的组合,具体可以根据实际的应用场景确定,这里不做具体限定。In the above ten association relationships, the indication information may indicate any one or more combinations thereof, which may be specifically determined according to the actual application scenario and is not specifically limited here.
本申请实施例中,指示信息在指示了上述十种关联关系中的至少一种的情况下,还可以指示AI模型的标识信息,该AI模型的标识信息可以用于确定具有关联关系的多个AI模型。可选地,指示信息在指示AI模型的标识信息时,可以包括以下至少一项:In the embodiment of the present application, the indication information may also indicate the identification information of the AI model when indicating at least one of the above ten association relationships, and the identification information of the AI model may be used to determine multiple AI models with association relationships. Optionally, when indicating the identification information of the AI model, the indication information may include at least one of the following:
(1)第一列表,第一列表中包括按照指定顺序排列的多个AI模型的标识信息,指定顺序与关联关系的关联顺序相对应。(1) A first list, wherein the first list includes identification information of multiple AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationship.
指定顺序可以是AI模型的标识信息在第一列表中的先后顺序、上下顺序等。关联关系的关联顺序可以是AI模型具有关联关系的先后顺序。The specified order may be the order of precedence, top-bottom order, etc. of the identification information of the AI models in the first list. The association order of the association relationship may be the order of precedence in which the AI models have the association relationship.
比如,第一列表中包含两个AI模型的标识信息,分别为模型A和模型B,模型A和模型B具有先后顺序,即模型A在前,模型B在后,那么,假设指示信息指示的关联关系为上述第八关联关系,在第八关联关系中,关联顺序为第一模型在前,第二模型在后,则可以确定模型A是模型B的子模型。For example, the first list contains identification information of two AI models, namely model A and model B. Model A and model B have a sequence, that is, model A comes first and model B comes later. Then, assuming that the association relationship indicated by the indication information is the eighth association relationship mentioned above, in the eighth association relationship, the association order is the first model in front and the second model in the back, it can be determined that model A is a submodel of model B.
可选地,在指示信息用于指示上述第八关联关系、第九关联关系和第十关联关系中的至少一项的情况下,指示信息中可以包含第一列表,这样,可以便于接收到指示信息的通信设备确定哪个AI模型是其他AI模型的子模型、或哪个AI模型是其他AI模型的子结构、或哪个AI模型的输出是其他AI模型的输入。Optionally, when the indication information is used to indicate at least one of the above-mentioned eighth, ninth and tenth association relationships, the indication information may include a first list, so that the communication device that receives the indication information can easily determine which AI model is a sub-model of other AI models, or which AI model is a sub-structure of other AI models, or which AI model's output is the input of other AI models.
(2)多个AI模型中部分AI模型的标识信息,其余AI模型与该部分AI模型具备关联关系。(2) Identification information of some AI models among multiple AI models, and the remaining AI models are associated with these AI models.
其余AI模型可以是通信双方预先约定的模型。比如,指示信息中包含模型A的标识信息,模型B为通信双方预先约定的模型,指示信息用于上述第一关联关系,则该指示信息具体用于指示模型A的模型输入和模型B的模型输入相同,根据该指示信息可以确定模型A的模型输入为预先约定的模型B的输入。The remaining AI models may be models pre-agreed upon by the communicating parties. For example, the indication information includes identification information of model A, model B is a model pre-agreed upon by the communicating parties, and the indication information is used for the above-mentioned first association relationship. Then, the indication information is specifically used to indicate that the model input of model A is the same as the model input of model B. According to the indication information, it can be determined that the model input of model A is the pre-agreed input of model B.
或者,其余AI模型也可以是接收到指示信息的通信设备需要确定或构建的AI模 型。比如,指示信息中包含模型A的标识信息,指示信息用于指示上述第九关联关系,则该指示信息具体用于指示模型B与模型A之间具备第九关联关系,通信设备基于该指示信息可以确定或构建模型B,模型B是模型A的子结构。Alternatively, the remaining AI models may also be AI models that the communication device that receives the indication information needs to determine or construct. For example, the indication information includes identification information of model A, and the indication information is used to indicate the ninth association relationship. The indication information is specifically used to indicate that there is a ninth association relationship between model B and model A. The communication device can determine or construct model B based on the indication information, and model B is a substructure of model A.
(3)多个AI模型的标识信息以及多个AI模型具备的关联关系。(3) Identification information of multiple AI models and the relationship between multiple AI models.
也就是说,指示信息可以将具备关联关系的多个AI模型的标识信息均进行指示,同时指示多个AI模型具体具备哪种关联关系。比如,指示信息中可以包含模型A的标识信息和模型B的标识信息,同时,指示信息还指示了模型A和模型B的模型输出相同,且模型A和模型B的部分结构相同。That is to say, the indication information can indicate the identification information of multiple AI models with associated relationships, and at the same time indicate what kind of association the multiple AI models have. For example, the indication information can include the identification information of model A and the identification information of model B. At the same time, the indication information also indicates that the model outputs of model A and model B are the same, and that some structures of model A and model B are the same.
本申请实施例提供的技术方案,在AI模型传输或AI模型信息指示的时候,可以避免模型特征的重复传输导致的传输资源浪费的问题,另外,通过指示信息指示AI模型间的相似特征(比如模型输入、模型输出、结构和参数等)能潜在地指示模型的关联关系,为多种基于AI的应用的联合处理提供了比较灵活的操作方法。The technical solution provided in the embodiment of the present application can avoid the problem of waste of transmission resources caused by repeated transmission of model features when the AI model is transmitted or the AI model information is indicated. In addition, by indicating similar features between AI models (such as model input, model output, structure and parameters, etc.) through indication information, it can potentially indicate the association relationship between the models, thereby providing a more flexible operation method for the joint processing of multiple AI-based applications.
在本申请实施例中,在对AI模型进行指示时,可以通过指示信息指示多个AI模型之间的关联关系,这样,可以通过AI模型间的关联关系确定需要指示的模型信息,从而实现对AI模型的指示。相较于传统的模型指示而言,由于可以无需将全部的模型信息进行指示,因此,可以减少传输资源,避免传输资源的浪费。In the embodiment of the present application, when indicating an AI model, the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model. Compared with the traditional model indication, since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
本申请实施例提供的模型指示方法,执行主体可以为模型指示装置。本申请实施例中以模型指示装置执行模型指示方法为例,说明本申请实施例提供的模型指示装置。The model indication method provided in the embodiment of the present application can be executed by a model indication device. In the embodiment of the present application, the model indication device provided in the embodiment of the present application is described by taking the execution of the model indication method by the model indication device as an example.
图3是根据本申请实施例的AI模型指示装置的结构示意图,该装置可以对应于其他实施例中的通信设备。如图3所示,装置300包括如下模块。Fig. 3 is a schematic diagram of the structure of an AI model indication device according to an embodiment of the present application, which may correspond to a communication device in other embodiments. As shown in Fig. 3, the device 300 includes the following modules.
通信模块301,用于发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。The communication module 301 is used to send or receive indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
可选地,作为一个实施例,所述关联关系包括以下至少一项:Optionally, as an embodiment, the association relationship includes at least one of the following:
第一关联关系,所述第一关联关系表征所述多个AI模型的模型输入相同;A first association relationship, wherein the first association relationship represents that model inputs of the multiple AI models are the same;
第二关联关系,所述第二关联关系表征所述多个AI模型的模型输出相同;A second association relationship, wherein the second association relationship represents that the model outputs of the multiple AI models are the same;
第三关联关系,所述第三关联关系表征所述多个AI模型的模型输入和模型输出均相同;A third association relationship, wherein the third association relationship represents that the model inputs and model outputs of the multiple AI models are the same;
第四关联关系,所述第四关联关系表征所述多个AI模型的模型结构相同;A fourth association relationship, wherein the fourth association relationship represents that the model structures of the multiple AI models are the same;
第五关联关系,所述第五关联关系表征所述多个AI模型的部分模型结构相同;A fifth association relationship, wherein the fifth association relationship represents that some model structures of the multiple AI models are the same;
第六关联关系,所述第六关联关系表征所述多个AI模型的部分模型结构和对应参数均相同;A sixth association relationship, wherein the sixth association relationship represents that some model structures and corresponding parameters of the multiple AI models are the same;
第七关联关系,所述第七关联关系表征所述多个AI模型的模型结构和对应参数均相同;A seventh association relationship, wherein the seventh association relationship represents that the model structures and corresponding parameters of the multiple AI models are the same;
第八关联关系,所述第八关联关系表征所述多个AI模型中的第一模型是所述多个AI模型中的第二模型的子模型; an eighth association relationship, wherein the eighth association relationship represents that the first model among the multiple AI models is a submodel of the second model among the multiple AI models;
第九关联关系,所述第九关联关系表征所述多个AI模型中的第三模型的结构是所述多个AI模型中的第四模型的子结构;a ninth association relationship, wherein the ninth association relationship represents that a structure of a third model among the multiple AI models is a substructure of a fourth model among the multiple AI models;
第十关联关系,所述第十关联关系表征所述多个AI模型中的第五模型的输出是所述多个AI模型中的第六模型的输入。The tenth association relationship represents that the output of the fifth model among the multiple AI models is the input of the sixth model among the multiple AI models.
可选地,作为一个实施例,所述模型输入相同,包括以下至少一项:Optionally, as an embodiment, the model input is the same, including at least one of the following:
所述模型输入的类型相同;The model inputs are of the same type;
所述模型输入的格式相同。The format of the model inputs is the same.
可选地,作为一个实施例,所述模型输出相同,包括以下至少一项:Optionally, as an embodiment, the model outputs are the same, including at least one of the following:
所述模型输出的类型相同;The model outputs are of the same type;
所述模型输出的格式相同。The model outputs are in the same format.
可选地,作为一个实施例,在所述关联关系包括所述第五关联关系或所述第六关联关系的情况下,所述指示信息还用于指示所述多个AI模型的模型结构的区别。Optionally, as an embodiment, when the association relationship includes the fifth association relationship or the sixth association relationship, the indication information is also used to indicate the difference in model structures of the multiple AI models.
可选地,作为一个实施例,在所述关联关系包括所述第八关联关系的情况下,所述指示信息还包括所述第二模型中的第一部分,所述第一部分的结构与所述第一模型的模型结构相同,所述第一部分的参数与所述第一模型的参数相同。Optionally, as an embodiment, when the association relationship includes the eighth association relationship, the indication information also includes a first part in the second model, a structure of the first part is the same as a model structure of the first model, and parameters of the first part are the same as parameters of the first model.
可选地,作为一个实施例,在所述关联关系包括所述第九关联关系的情况下,所述指示信息还包括所述第四模型中的第二部分,所述第二部分的结构与所述第三模型的模型结构相同,所述第二部分的参数与所述第三模型的参数不同。Optionally, as an embodiment, when the association relationship includes the ninth association relationship, the indication information also includes the second part in the fourth model, the structure of the second part is the same as the model structure of the third model, and the parameters of the second part are different from the parameters of the third model.
可选地,作为一个实施例,所述指示信息包括以下至少一项:Optionally, as an embodiment, the indication information includes at least one of the following:
第一列表,所述第一列表中包括按照指定顺序排列的所述多个AI模型的标识信息,所述指定顺序与所述关联关系的关联顺序相对应;A first list, wherein the first list includes identification information of the plurality of AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationships;
所述多个AI模型中部分AI模型的标识信息,其余AI模型与所述部分AI模型具备所述关联关系;Identification information of some AI models among the multiple AI models, and the remaining AI models have the association relationship with the some AI models;
所述多个AI模型的标识信息以及所述多个AI模型具备的关联关系。The identification information of the multiple AI models and the association relationship between the multiple AI models.
根据本申请实施例的装置300可以参照对应本申请实施例的方法200的流程,并且,该装置300中的各个单元/模块和上述其他操作和/或功能分别为了实现方法200中的相应流程,并且能够达到相同或等同的技术效果,为了简洁,在此不再赘述。According to the device 300 of the embodiment of the present application, the process of the method 200 corresponding to the embodiment of the present application can be referred to, and the various units/modules in the device 300 and the above-mentioned other operations and/or functions are respectively for implementing the corresponding processes in the method 200, and can achieve the same or equivalent technical effects. For the sake of brevity, they will not be repeated here.
本申请实施例中的模型指示装置可以是电子设备,例如具有操作系统的电子设备,也可以是电子设备中的部件,例如集成电路或芯片。该电子设备可以是终端,也可以为除终端之外的其他设备。示例性的,终端可以包括但不限于上述所列举的终端11的类型,其他设备可以为服务器、网络附属存储器(Network Attached Storage,NAS)等,本申请实施例不作具体限定。The model indicating device in the embodiment of the present application can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or it can be other devices other than a terminal. Exemplarily, the terminal can include but is not limited to the types of terminals 11 listed above, and other devices can be servers, network attached storage (NAS), etc., which are not specifically limited in the embodiment of the present application.
本申请实施例提供的AI模型指示装置能够实现图2的方法实施例实现的各个过程,并达到相同的技术效果,为避免重复,这里不再赘述。The AI model indication device provided in the embodiment of the present application can implement each process implemented by the method embodiment of Figure 2 and achieve the same technical effect. To avoid repetition, it will not be repeated here.
如图4所示,本申请实施例还提供一种通信设备400,包括处理器401和存储器 402,存储器402上存储有可在所述处理器401上运行的程序或指令,例如,该通信设备400为终端时,该程序或指令被处理器401执行时实现上述AI模型指示方法实施例的各个步骤,且能达到相同的技术效果。该通信设备400为网络侧设备时,该程序或指令被处理器401执行时实现上述AI模型指示方法实施例的各个步骤,且能达到相同的技术效果,为避免重复,这里不再赘述。As shown in FIG4 , the present embodiment also provides a communication device 400, including a processor 401 and a memory. 402, the memory 402 stores a program or instruction that can be run on the processor 401. For example, when the communication device 400 is a terminal, the program or instruction is executed by the processor 401 to implement the various steps of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. When the communication device 400 is a network side device, the program or instruction is executed by the processor 401 to implement the various steps of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
本申请实施例还提供一种通信设备,包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现如图2所示方法实施例中的步骤。该通信设备实施例与上述通信设备侧方法实施例对应,上述方法实施例的各个实施过程和实现方式均可适用于该通信设备实施例中,且能达到相同的技术效果。具体地,图5为实现本申请实施例的一种通信设备的硬件结构示意图。The embodiment of the present application also provides a communication device, including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a program or instruction to implement the steps in the method embodiment shown in Figure 2. The communication device embodiment corresponds to the above-mentioned communication device side method embodiment, and each implementation process and implementation method of the above-mentioned method embodiment can be applied to the communication device embodiment and can achieve the same technical effect. Specifically, Figure 5 is a schematic diagram of the hardware structure of a communication device implementing an embodiment of the present application.
该通信设备500包括但不限于:射频单元501、网络模块502、音频输出单元503、输入单元504、传感器505、显示单元506、用户输入单元507、接口单元508、存储器509以及处理器510等中的至少部分部件。The communication device 500 includes but is not limited to: a radio frequency unit 501, a network module 502, an audio output unit 503, an input unit 504, a sensor 505, a display unit 506, a user input unit 507, an interface unit 508, a memory 509 and at least some of the components of a processor 510.
本领域技术人员可以理解,通信设备500还可以包括给各个部件供电的电源(比如电池),电源可以通过电源管理系统与处理器5 10逻辑相连,从而通过电源管理系统实现管理充电、放电以及功耗管理等功能。图5中示出的通信设备结构并不构成对终端的限定,终端可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置,在此不再赘述。Those skilled in the art will appreciate that the communication device 500 may also include a power source (such as a battery) for supplying power to each component, and the power source may be logically connected to the processor 5 10 through a power management system, so that the power management system can manage charging, discharging, and power consumption. The communication device structure shown in FIG5 does not constitute a limitation on the terminal, and the terminal may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently, which will not be described in detail here.
应理解的是,本申请实施例中,输入单元504可以包括图形处理单元(Graphics Processing Unit,GPU)5041和麦克风5042,图形处理器5041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元506可包括显示面板5061,可以采用液晶显示器、有机发光二极管等形式来配置显示面板5061。用户输入单元507包括触控面板5071以及其他输入设备5072中的至少一种。触控面板5071,也称为触摸屏。触控面板5071可包括触摸检测装置和触摸控制器两个部分。其他输入设备5072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆,在此不再赘述。It should be understood that in the embodiment of the present application, the input unit 504 may include a graphics processing unit (GPU) 5041 and a microphone 5042, and the graphics processor 5041 processes the image data of the static picture or video obtained by the image capture device (such as a camera) in the video capture mode or the image capture mode. The display unit 506 may include a display panel 5061, and the display panel 5061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 507 includes a touch panel 5071 and at least one of other input devices 5072. The touch panel 5071 is also called a touch screen. The touch panel 5071 may include two parts: a touch detection device and a touch controller. Other input devices 5072 may include, but are not limited to, a physical keyboard, function keys (such as a volume control key, a switch key, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.
本申请实施例中,射频单元501接收来自网络侧设备的下行数据后,可以传输给处理器510进行处理;另外,射频单元501可以向网络侧设备发送上行数据。通常,射频单元501包括但不限于天线、放大器、收发信机、耦合器、低噪声放大器、双工器等。In the embodiment of the present application, after receiving downlink data from the network side device, the radio frequency unit 501 can transmit the data to the processor 510 for processing; in addition, the radio frequency unit 501 can send uplink data to the network side device. Generally, the radio frequency unit 501 includes but is not limited to an antenna, an amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc.
存储器509可用于存储软件程序或指令以及各种数据。存储器509可主要包括存储程序或指令的第一存储区和存储数据的第二存储区,其中,第一存储区可存储操作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外,存储器509可以包括易失性存储器或非易失性存储器。其中,非易失性存储器可以是只读存储器(Read-Only Memory,ROM)、可编程只读存储器(Programmable  ROM,PROM)、可擦除可编程只读存储器(Erasable PROM,EPROM)、电可擦除可编程只读存储器(Electrically EPROM,EEPROM)或闪存。易失性存储器可以是随机存取存储器(Random Access Memory,RAM),静态随机存取存储器(Static RAM,SRAM)、动态随机存取存储器(Dynamic RAM,DRAM)、同步动态随机存取存储器(Synchronous DRAM,SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM,DDRSDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM,ESDRAM)、同步连接动态随机存取存储器(Synch link DRAM,SLDRAM)和直接内存总线随机存取存储器(Direct Rambus RAM,DRRAM)。本申请实施例中的存储器509包括但不限于这些和任意其它适合类型的存储器。The memory 509 can be used to store software programs or instructions and various data. The memory 509 can mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area can store an operating system, an application program or instruction required for at least one function (such as a sound playback function, an image playback function, etc.), etc. In addition, the memory 509 can include a volatile memory or a non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (ROM), or a programmable read-only memory (PROM). The volatile memory may be a random access memory (RAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDRSDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synchronous link dynamic random access memory (SLDRAM) and a direct memory bus random access memory (DRRAM). The memory 509 in the embodiment of the present application includes but is not limited to these and any other suitable types of memories.
处理器510可包括一个或多个处理单元;可选的,处理器510集成应用处理器和调制解调处理器,其中,应用处理器主要处理涉及操作系统、用户界面和应用程序等的操作,调制解调处理器主要处理无线通信信号,如基带处理器。可以理解的是,上述调制解调处理器也可以不集成到处理器510中。The processor 510 may include one or more processing units; optionally, the processor 510 integrates an application processor and a modem processor, wherein the application processor mainly processes operations related to an operating system, a user interface, and application programs, and the modem processor mainly processes wireless communication signals, such as a baseband processor. It is understandable that the modem processor may not be integrated into the processor 510.
其中,射频单元501,用于发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。Among them, the radio frequency unit 501 is used to send or receive indication information, and the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
在本申请实施例中,在对AI模型进行指示时,可以通过指示信息指示多个AI模型之间的关联关系,这样,可以通过AI模型间的关联关系确定需要指示的模型信息,从而实现对AI模型的指示。相较于传统的模型指示而言,由于可以无需将全部的模型信息进行指示,因此,可以减少传输资源,避免传输资源的浪费。In the embodiment of the present application, when indicating an AI model, the association relationship between multiple AI models can be indicated through the indication information, so that the model information that needs to be indicated can be determined through the association relationship between the AI models, thereby realizing the indication of the AI model. Compared with the traditional model indication, since it is not necessary to indicate all the model information, it is possible to reduce transmission resources and avoid the waste of transmission resources.
本申请实施例提供的通信设备500还可以实现上述图2所示实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。The communication device 500 provided in the embodiment of the present application can also implement the various processes of the embodiment shown in Figure 2 above, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
可以理解,本实施例中提及的各实现方式的实现过程可以参照方法实施例200的相关描述,并达到相同或相应的技术效果,为避免重复,在此不再赘述。It can be understood that the implementation process of each implementation method mentioned in this embodiment can refer to the relevant description of method embodiment 200, and achieve the same or corresponding technical effect. To avoid repetition, it will not be repeated here.
本申请实施例还提供一种通信设备,包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现如图2所示的方法实施例的步骤。该通信设备实施例与上述通信设备方法实施例对应,上述方法实施例的各个实施过程和实现方式均可适用于该通信设备实施例中,且能达到相同的技术效果。The embodiment of the present application also provides a communication device, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the steps of the method embodiment shown in Figure 2. The communication device embodiment corresponds to the above communication device method embodiment, and each implementation process and implementation method of the above method embodiment can be applied to the communication device embodiment, and can achieve the same technical effect.
具体地,本申请实施例还提供了一种通信设备。如图6所示,该通信设备600包括:天线61、射频装置62、基带装置63、处理器64和存储器65。天线61与射频装置62连接。在上行方向上,射频装置62通过天线61接收信息,将接收的信息发送给基带装置63进行处理。在下行方向上,基带装置63对要发送的信息进行处理,并发送给射频装置62,射频装置62对收到的信息进行处理后经过天线61发送出去。Specifically, an embodiment of the present application further provides a communication device. As shown in FIG6 , the communication device 600 includes: an antenna 61, a radio frequency device 62, a baseband device 63, a processor 64, and a memory 65. The antenna 61 is connected to the radio frequency device 62. In the uplink direction, the radio frequency device 62 receives information through the antenna 61 and sends the received information to the baseband device 63 for processing. In the downlink direction, the baseband device 63 processes the information to be sent and sends it to the radio frequency device 62. The radio frequency device 62 processes the received information and sends it out through the antenna 61.
以上实施例中通信设备执行的方法可以在基带装置63中实现,该基带装置63包括基带处理器。The method executed by the communication device in the above embodiment may be implemented in the baseband device 63, which includes a baseband processor.
基带装置63例如可以包括至少一个基带板,该基带板上设置有多个芯片,如图6 所示,其中一个芯片例如为基带处理器,通过总线接口与存储器65连接,以调用存储器65中的程序,执行以上方法实施例中所示的通信设备操作。The baseband device 63 may include, for example, at least one baseband board on which a plurality of chips are arranged. As shown, one of the chips is, for example, a baseband processor, which is connected to the memory 65 through a bus interface to call the program in the memory 65 to execute the communication device operations shown in the above method embodiment.
该通信设备还可以包括网络接口66,该接口例如为通用公共无线接口(Common Public Radio Interface,CPRI)。The communication device may also include a network interface 66, which is, for example, a Common Public Radio Interface (CPRI).
具体地,本申请实施例的通信设备600还包括:存储在存储器65上并可在处理器64上运行的指令或程序,处理器64调用存储器65中的指令或程序执行图3所示各模块执行的方法,并达到相同的技术效果,为避免重复,故不在此赘述。Specifically, the communication device 600 of the embodiment of the present application also includes: instructions or programs stored in the memory 65 and executable on the processor 64. The processor 64 calls the instructions or programs in the memory 65 to execute the methods executed by the modules shown in Figure 3 and achieve the same technical effect. To avoid repetition, it will not be repeated here.
本申请实施例还提供一种可读存储介质,所述可读存储介质上存储有程序或指令,该程序或指令被处理器执行时实现上述AI模型指示方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned AI model indication method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
其中,所述处理器为上述实施例中所述的通信设备中的处理器。所述可读存储介质,包括计算机可读存储介质,如计算机只读存储器ROM、随机存取存储器RAM、磁碟或者光盘等。在一些示例中,可读存储介质可以是非瞬态的可读存储介质。The processor is a processor in the communication device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk. In some examples, the readable storage medium may be a non-transient readable storage medium.
本申请实施例另提供了一种芯片,所述芯片包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现上述AI模型指示方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。An embodiment of the present application further provides a chip, which includes 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 the various processes of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
应理解,本申请实施例提到的芯片还可以称为系统级芯片,系统芯片,芯片系统或片上系统芯片等。It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
本申请实施例另提供了一种计算机程序/程序产品,所述计算机程序/程序产品被存储在存储介质中,所述计算机程序/程序产品被至少一个处理器执行以实现上述AI模型指示方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。The embodiments of the present application further provide a computer program/program product, which is stored in a storage medium. The computer program/program product is executed by at least one processor to implement the various processes of the above-mentioned AI model indication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
需要说明的是,在本文中,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外,需要指出的是,本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能,还可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能,例如,可以按不同于所描述的次序来执行所描述的方法,并且还可以添加、省去或组合各种步骤。另外,参照某些示例所描述的特征可在其他示例中被组合。It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises one..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted or combined. In addition, the features described with reference to certain examples may be combined in other examples.
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助计算机软件产品加必需的通用硬件平台的方式来实现,当然也可以通过硬件。该计算机软件产品存储在存储介质(如ROM、RAM、磁碟、光盘等)中,包括若干指令,用以使得终端或者网络侧设备执行本申请各个实施例所述的方法。 Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of a computer software product plus a necessary general hardware platform, and of course, it can also be implemented by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, disk, CD, etc.), including several instructions to enable the terminal or network side device to execute the method described in each embodiment of the present application.
上面结合附图对本申请的实施例进行了描述,但是本申请并不局限于上述的具体实施方式,上述的具体实施方式仅仅是示意性的,而不是限制性的,本领域的普通技术人员在本申请的启示下,在不脱离本申请宗旨和权利要求所保护的范围情况下,还可做出很多形式的实施方式,这些实施方式均属于本申请的保护之内。 The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms of implementation methods without departing from the purpose of the present application and the scope of protection of the claims, and these implementation methods are all within the protection of the present application.

Claims (20)

  1. 一种AI模型指示方法,包括:An AI model indication method, comprising:
    通信设备发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。The communication device sends or receives indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  2. 根据权利要求1所示的方法,其中,所述关联关系包括以下至少一项:The method according to claim 1, wherein the association relationship includes at least one of the following:
    第一关联关系,所述第一关联关系表征所述多个AI模型的模型输入相同;A first association relationship, wherein the first association relationship represents that model inputs of the multiple AI models are the same;
    第二关联关系,所述第二关联关系表征所述多个AI模型的模型输出相同;A second association relationship, wherein the second association relationship represents that the model outputs of the multiple AI models are the same;
    第三关联关系,所述第三关联关系表征所述多个AI模型的模型输入和模型输出均相同;A third association relationship, wherein the third association relationship represents that the model inputs and model outputs of the multiple AI models are the same;
    第四关联关系,所述第四关联关系表征所述多个AI模型的模型结构相同;A fourth association relationship, wherein the fourth association relationship represents that the model structures of the multiple AI models are the same;
    第五关联关系,所述第五关联关系表征所述多个AI模型的部分模型结构相同;A fifth association relationship, wherein the fifth association relationship represents that some model structures of the multiple AI models are the same;
    第六关联关系,所述第六关联关系表征所述多个AI模型的部分模型结构和对应参数均相同;A sixth association relationship, wherein the sixth association relationship represents that some model structures and corresponding parameters of the multiple AI models are the same;
    第七关联关系,所述第七关联关系表征所述多个AI模型的模型结构和对应参数均相同;A seventh association relationship, wherein the seventh association relationship represents that the model structures and corresponding parameters of the multiple AI models are the same;
    第八关联关系,所述第八关联关系表征所述多个AI模型中的第一模型是所述多个模型中的第二模型的子模型;An eighth association relationship, wherein the eighth association relationship represents that a first model among the multiple AI models is a submodel of a second model among the multiple models;
    第九关联关系,所述第九关联关系表征所述多个AI模型中的第三模型的结构是所述多个AI模型中的第四模型的子结构;a ninth association relationship, wherein the ninth association relationship represents that a structure of a third model among the multiple AI models is a substructure of a fourth model among the multiple AI models;
    第十关联关系,所述第十关联关系表征所述多个AI模型中的第五模型的输出是所述多个AI模型中的第六模型的输入。The tenth association relationship represents that the output of the fifth model among the multiple AI models is the input of the sixth model among the multiple AI models.
  3. 根据权利要求2所示的方法,其中,所述模型输入相同,包括以下至少一项:The method according to claim 2, wherein the model inputs are the same and include at least one of the following:
    所述模型输入的类型相同;The model inputs are of the same type;
    所述模型输入的格式相同。The format of the model inputs is the same.
  4. 根据权利要求2所示的方法,其中,所述模型输出相同,包括以下至少一项:The method according to claim 2, wherein the model outputs are the same, including at least one of the following:
    所述模型输出的类型相同;The model outputs are of the same type;
    所述模型输出的格式相同。The model outputs are in the same format.
  5. 根据权利要求2所示的方法,其中,The method according to claim 2, wherein:
    在所述关联关系包括所述第五关联关系或所述第六关联关系的情况下,所述指示信息还包括所述多个AI模型的模型结构的区别。In a case where the association relationship includes the fifth association relationship or the sixth association relationship, the indication information further includes the difference in model structures of the multiple AI models.
  6. 根据权利要求2所示的方法,其中,The method according to claim 2, wherein:
    在所述关联关系包括所述第八关联关系的情况下,所述指示信息还还包括所述第二模型中的第一部分,所述第一部分的结构与所述第一模型的模型结构相同,所述第一部分的参数与所述第一模型的参数相同。In the case where the association relationship includes the eighth association relationship, the indication information further includes a first part in the second model, a structure of the first part is the same as a model structure of the first model, and parameters of the first part are the same as parameters of the first model.
  7. 根据权利要求2所示的方法,其中, The method according to claim 2, wherein:
    在所述关联关系包括所述第九关联关系的情况下,所述指示信息还包括所述第四模型中的第二部分,所述第二部分的结构与所述第三模型的模型结构相同,所述第二部分的参数与所述第三模型的参数不同。In the case where the association relationship includes the ninth association relationship, the indication information also includes a second part in the fourth model, a structure of the second part is the same as a model structure of the third model, and parameters of the second part are different from parameters of the third model.
  8. 根据权利要求1所示的方法,其中,所述指示信息包括以下至少一项:The method according to claim 1, wherein the indication information includes at least one of the following:
    第一列表,所述第一列表中包括按照指定顺序排列的所述多个AI模型的标识信息,所述指定顺序与所述关联关系的关联顺序相对应;A first list, wherein the first list includes identification information of the plurality of AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationships;
    所述多个AI模型中部分AI模型的标识信息,其余AI模型与所述部分AI模型具备所述关联关系;Identification information of some AI models among the multiple AI models, and the remaining AI models have the association relationship with the some AI models;
    所述多个AI模型的标识信息以及所述多个AI模型具备的关联关系。The identification information of the multiple AI models and the association relationship between the multiple AI models.
  9. 一种AI模型指示装置,包括:An AI model indicating device, comprising:
    通信模块,用于发送或接收指示信息,所述指示信息用于指示多个人工智能AI模型之间的关联关系。A communication module is used to send or receive indication information, where the indication information is used to indicate the association relationship between multiple artificial intelligence AI models.
  10. 根据权利要求9所示的装置,其中,所述关联关系包括以下至少一项:The apparatus according to claim 9, wherein the association relationship includes at least one of the following:
    第一关联关系,所述第一关联关系表征所述多个AI模型的模型输入相同;A first association relationship, wherein the first association relationship represents that model inputs of the multiple AI models are the same;
    第二关联关系,所述第二关联关系表征所述多个AI模型的模型输出相同;A second association relationship, wherein the second association relationship represents that the model outputs of the multiple AI models are the same;
    第三关联关系,所述第三关联关系表征所述多个AI模型的模型输入和模型输出均相同;A third association relationship, wherein the third association relationship represents that the model inputs and model outputs of the multiple AI models are the same;
    第四关联关系,所述第四关联关系表征所述多个AI模型的模型结构相同;A fourth association relationship, wherein the fourth association relationship represents that the model structures of the multiple AI models are the same;
    第五关联关系,所述第五关联关系表征所述多个AI模型的部分模型结构相同;A fifth association relationship, wherein the fifth association relationship represents that some model structures of the multiple AI models are the same;
    第六关联关系,所述第六关联关系表征所述多个AI模型的部分模型结构和对应参数均相同;A sixth association relationship, wherein the sixth association relationship represents that some model structures and corresponding parameters of the multiple AI models are the same;
    第七关联关系,所述第七关联关系表征所述多个AI模型的模型结构和对应参数均相同;A seventh association relationship, wherein the seventh association relationship represents that the model structures and corresponding parameters of the multiple AI models are the same;
    第八关联关系,所述第八关联关系表征所述多个AI模型中的第一模型是所述多个模型中的第二模型的子模型;An eighth association relationship, wherein the eighth association relationship represents that a first model among the multiple AI models is a submodel of a second model among the multiple models;
    第九关联关系,所述第九关联关系表征所述多个AI模型中的第三模型的结构是所述多个AI模型中的第四模型的子结构;a ninth association relationship, wherein the ninth association relationship represents that a structure of a third model among the multiple AI models is a substructure of a fourth model among the multiple AI models;
    第十关联关系,所述第十关联关系表征所述多个AI模型中的第五模型的输出是所述多个AI模型中的第六模型的输入。The tenth association relationship represents that the output of the fifth model among the multiple AI models is the input of the sixth model among the multiple AI models.
  11. 根据权利要求10所示的装置,其中,所述模型输入相同,包括以下至少一项:The apparatus according to claim 10, wherein the model inputs are the same and include at least one of the following:
    所述模型输入的类型相同;The model inputs are of the same type;
    所述模型输入的格式相同。The format of the model inputs is the same.
  12. 根据权利要求10所示的装置,其中,所述模型输出相同,包括以下至少一项:The apparatus according to claim 10, wherein the model outputs are the same, including at least one of the following:
    所述模型输出的类型相同;The model outputs are of the same type;
    所述模型输出的格式相同。 The model outputs are in the same format.
  13. 根据权利要求10所示的装置,其中,The device according to claim 10, wherein
    在所述关联关系包括所述第五关联关系或所述第六关联关系的情况下,所述指示信息还包括所述多个AI模型的模型结构的区别。In a case where the association relationship includes the fifth association relationship or the sixth association relationship, the indication information further includes the difference in model structures of the multiple AI models.
  14. 根据权利要求10所示的装置,其中,The device according to claim 10, wherein
    在所述关联关系包括所述第八关联关系的情况下,所述指示信息还包括所述第二模型中的第一部分,所述第一部分的结构与所述第一模型的模型结构相同,所述第一部分的参数与所述第一模型的参数相同。In the case where the association relationship includes the eighth association relationship, the indication information also includes a first part in the second model, a structure of the first part is the same as a model structure of the first model, and parameters of the first part are the same as parameters of the first model.
  15. 根据权利要求10所示的装置,其中,The device according to claim 10, wherein
    在所述关联关系包括所述第九关联关系的情况下,所述指示信息还包括所述第四模型中的第二部分,所述第二部分的结构与所述第三模型的模型结构相同,所述第二部分的参数与所述第三模型的参数不同。In the case where the association relationship includes the ninth association relationship, the indication information also includes a second part in the fourth model, a structure of the second part is the same as a model structure of the third model, and parameters of the second part are different from parameters of the third model.
  16. 根据权利要求9所示的装置,其中,所述指示信息包括以下至少一项:The apparatus according to claim 9, wherein the indication information comprises at least one of the following:
    第一列表,所述第一列表中包括按照指定顺序排列的所述多个AI模型的标识信息,所述指定顺序与所述关联关系的关联顺序相对应;A first list, wherein the first list includes identification information of the plurality of AI models arranged in a specified order, and the specified order corresponds to an association order of the association relationships;
    所述多个AI模型中部分AI模型的标识信息,其余AI模型与所述部分AI模型具备所述关联关系;Identification information of some AI models among the multiple AI models, and the remaining AI models have the association relationship with the some AI models;
    所述多个AI模型的标识信息以及所述多个AI模型具备的关联关系。The identification information of the multiple AI models and the association relationship between the multiple AI models.
  17. 一种通信设备,包括处理器和存储器,所述存储器存储可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如权利要求1至8任一项所述的AI模型指示方法的步骤。A communication device comprises a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, the steps of the AI model indication method as described in any one of claims 1 to 8 are implemented.
  18. 一种可读存储介质,所述可读存储介质上存储程序或指令,所述程序或指令被处理器执行时实现如权利要求1至8任一项所述的AI模型指示方法的步骤。A readable storage medium storing a program or instruction, wherein the program or instruction, when executed by a processor, implements the steps of the AI model indication method according to any one of claims 1 to 8.
  19. 一种计算机程序产品,所述计算机程序产品被至少一个处理器执行以实现如权利要求1至8任一项所述的AI模型指示方法。A computer program product, wherein the computer program product is executed by at least one processor to implement the AI model indication method according to any one of claims 1 to 8.
  20. 一种电子设备,包括所述电子设备被配置成用于执行如权利要求1至8任一项所述的AI模型指示方法。 An electronic device, comprising: the electronic device is configured to execute the AI model indication method as described in any one of claims 1 to 8.
PCT/CN2024/076413 2023-02-13 2024-02-06 Ai model indication method and communication device WO2024169797A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202310114133.0 2023-02-13
CN202310114133.0A CN118487952A (en) 2023-02-13 2023-02-13 AI model indication method and communication device

Publications (1)

Publication Number Publication Date
WO2024169797A1 true WO2024169797A1 (en) 2024-08-22

Family

ID=92195475

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2024/076413 WO2024169797A1 (en) 2023-02-13 2024-02-06 Ai model indication method and communication device

Country Status (2)

Country Link
CN (1) CN118487952A (en)
WO (1) WO2024169797A1 (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111930346A (en) * 2020-07-15 2020-11-13 北京百度网讯科技有限公司 Artificial intelligence information processing method and device, electronic equipment and storage medium
US20210248447A1 (en) * 2020-02-12 2021-08-12 Palantir Technologies Inc. System and method for chaining discrete models
CN113377503A (en) * 2020-03-09 2021-09-10 阿尔法云计算(深圳)有限公司 Task scheduling method, device and system for collaborative AI (artificial intelligence)
US11119630B1 (en) * 2018-06-19 2021-09-14 Palantir Technologies Inc. Artificial intelligence assisted evaluations and user interface for same
US20220318640A1 (en) * 2021-03-31 2022-10-06 Fujitsu Limited Automated empathetic reconciliation of decisions of artificial intelligence (ai) models
CN115460100A (en) * 2021-06-09 2022-12-09 华为技术有限公司 A communication method and communication device applying artificial intelligence
CN115511085A (en) * 2022-10-31 2022-12-23 上海浦东发展银行股份有限公司 Model data processing method, device, equipment and storage medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11119630B1 (en) * 2018-06-19 2021-09-14 Palantir Technologies Inc. Artificial intelligence assisted evaluations and user interface for same
US20210248447A1 (en) * 2020-02-12 2021-08-12 Palantir Technologies Inc. System and method for chaining discrete models
CN113377503A (en) * 2020-03-09 2021-09-10 阿尔法云计算(深圳)有限公司 Task scheduling method, device and system for collaborative AI (artificial intelligence)
CN111930346A (en) * 2020-07-15 2020-11-13 北京百度网讯科技有限公司 Artificial intelligence information processing method and device, electronic equipment and storage medium
US20220318640A1 (en) * 2021-03-31 2022-10-06 Fujitsu Limited Automated empathetic reconciliation of decisions of artificial intelligence (ai) models
CN115460100A (en) * 2021-06-09 2022-12-09 华为技术有限公司 A communication method and communication device applying artificial intelligence
CN115511085A (en) * 2022-10-31 2022-12-23 上海浦东发展银行股份有限公司 Model data processing method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN118487952A (en) 2024-08-13

Similar Documents

Publication Publication Date Title
WO2023160547A1 (en) Information response method, information sending method, terminal, and network side device
US20240330776A1 (en) Client screening method and apparatus, client, and central device
WO2023143581A1 (en) Information transmission method and apparatus, and terminal and network-side device
WO2024012239A1 (en) Condition configuration processing method, condition configuration processing apparatus, and terminal
US20240405955A1 (en) Information activation method, terminal, and network side device
WO2024169797A1 (en) Ai model indication method and communication device
WO2023174253A1 (en) Ai model processing method and device
WO2023109759A1 (en) Prach transmission method and apparatus, and terminal
WO2024169794A1 (en) Model supervision method and communication device
CN112886991B (en) Interference elimination method and device and electronic equipment
EP4451740A1 (en) Path preference determination method, terminal and network side device
WO2024120444A1 (en) Model supervision method and apparatus, terminal, network side device and readable storage medium
WO2024067878A1 (en) Channel prediction method and apparatus, and communication device
WO2024169911A1 (en) Multi-antenna transmit diversity indication method, and terminal and network-side device
US20250036978A1 (en) Model fine-tuning method and apparatus, and device
US20240314045A1 (en) Information interaction method and apparatus, and communication device
WO2024235043A1 (en) Information acquisition method and apparatus, and communication device
CN120238945A (en) Reporting method, transmission method, device, terminal and network equipment
WO2025108288A1 (en) Model monitoring method and apparatus, and communication device
CN120282160A (en) AI unit activation method, terminal and network side equipment
WO2024061287A1 (en) Artificial intelligence (ai) model transmission method and apparatus, and terminal and medium
WO2025140604A1 (en) Information reporting methods and apparatuses, and first devices and second devices
CN120390247A (en) RRM measurement processing method, device, terminal and network side equipment
WO2025140432A1 (en) Information reporting method and apparatus
WO2025157103A1 (en) Rrm measurement processing methods and apparatuses, and terminal and network-side device

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 24756170

Country of ref document: EP

Kind code of ref document: A1