Synchronization of twin models

Abstract

Example embodiments of the present disclosure are directed to synchronization of models. A method comprises obtaining one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and transmitting, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

Description

SYNCHRONIZATION OF TWIN MODELSCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of Fl application No. 20245886, filed July 12, 2024. The content of which are hereby incorporated by reference in their entirety.FIELD

[0002] Various example embodiments of the present disclosure generally relate to the field of telecommunication and in particular, to methods, devices, apparatuses and computer readable storage medium for synchronization of twin models.BACKGROUND

[0003] Artificial Intelligence (Al) and Machine Learning (ML) techniques are being increasingly employed in 5G system (5GS) and are considered as a key enabler of 5G-Advanced and 6G mobile network generation. Life cycle management (LCM) of AI / ML models is a crucial aspect for current and future deployments. Thus, it is worth studying LCM of AI / ML models.SUMMARY

[0004] In a first aspect of the present disclosure, there is provided a first apparatus. The first apparatus comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus to: obtain one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and transmit, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

[0005] In a second aspect of the present disclosure, there is provided a second apparatus. The second apparatus comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the second apparatus to: receive, from a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model.

[0006] In a third aspect of the present disclosure, there is provided a first apparatus. The first apparatus comprises at least one processor; and at least one memory storing instructions that, iwhen executed by the at least one processor, cause the first apparatus to: obtain a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus , wherein the parameter set is used in a training process of the model; and transmit, to the second apparatus, an indication related to the parameter set for initializing the parameter set at the second apparatus.

[0007] In a fourth aspect of the present disclosure, there is provided a second apparatus. The second apparatus comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the second apparatus to: receive, from a first apparatus, an indication related to a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus , wherein the parameter set is used in a training process of the model.

[0008] In a fifth aspect of the present disclosure, there is provided a method. The method comprises: obtaining, at a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and transmitting, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

[0009] In a sixth aspect of the present disclosure, there is provided a method. The method comprises: receiving, at a second apparatus and from a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model.

[0010] In a seventh aspect of the present disclosure, there is provided a method. The method comprises: obtaining, at a first apparatus, a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model; and transmitting, to the second apparatus, an indication related to the parameter set for initializing the parameter set at the second apparatus.

[0011] In an eighth aspect of the present disclosure, there is provided a method. The method comprises: receiving, at a second apparatus and from a first apparatus, an indication related to a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model.

[0012] In a ninth aspect of the present disclosure, there is provided a first apparatus. The first apparatus comprises means for obtaining one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and means for transmitting, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

[0013] In a tenth aspect of the present disclosure, there is provided a second apparatus. The second apparatus comprises means for receiving, from a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model.

[0014] In an eleventh aspect of the present disclosure, there is provided a first apparatus. The first apparatus comprises means for obtaining a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model; and means for transmitting, to the second apparatus, an indication related to the parameter set for initializing the parameter set at the second apparatus.

[0015] In a twelfth aspect of the present disclosure, there is provided a second apparatus. The second apparatus comprises means for receiving, from a first apparatus, an indication related to a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model.

[0016] In a thirteenth aspect of the present disclosure, there is provided a computer readable medium. The computer readable medium comprises instructions stored thereon for causing an apparatus to perform at least the method according to the fifth aspect.

[0017] In a fourteenth aspect of the present disclosure, there is provided a computer readable medium. The computer readable medium comprises instructions stored thereon for causing an apparatus to perform at least the method according to the sixth aspect.

[0018] In a fifteenth aspect of the present disclosure, there is provided a computer readable medium. The computer readable medium comprises instructions stored thereon for causing an apparatus to perform at least the method according to the seventh aspect.

[0019] In a sixteenth aspect of the present disclosure, there is provided a computer readable medium. The computer readable medium comprises instructions stored thereon for causing anapparatus to perform at least the method according to the eighth aspect.

[0020] According to some aspects, there is provided the subject matter of the independent claims. Some further aspects are defined in the dependent claims.

[0021] It is to be understood that the Summary section is not intended to identify key or essential features of embodiments of the present disclosure, nor is it intended to be used to limit the scope of the present disclosure. Other features of the present disclosure will become easily comprehensible through the following description.BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Some example embodiments will now be described with reference to the accompanying drawings, where:

[0023] FIG. 1 illustrates an example communication environment in which example embodiments of the present disclosure can be implemented;

[0024] FIG. 2 illustrates a schematic diagram of twin model architecture;

[0025] FIG. 3 illustrates an example signaling diagram of twin model synchronization according to some example embodiments of the present disclosure;

[0026] FIG. 4 illustrates another example signaling diagram of twin model synchronization according to some other example embodiments of the present disclosure;

[0027] FIG. 5 illustrates a flowchart of a method for twin model synchronization according to some example embodiments of the present disclosure;

[0028] FIG. 6 illustrates an example signaling diagram of twin model synchronization initiated by a terminal device according to some example embodiments of the present disclosure;

[0029] FIG. 7 illustrates a flowchart of a method implemented at a first apparatus in accordance with some example embodiments of the present disclosure;

[0030] FIG. 8 illustrates a flowchart of a method implemented at a second apparatus in accordance with some example embodiments of the present disclosure;

[0031] FIG. 9 illustrates a flowchart of a method implemented at a first apparatus in accordance with some example embodiments of the present disclosure;

[0032] FIG. 10 illustrates a flowchart of a method implemented at a second apparatus inaccordance with some example embodiments of the present disclosure;

[0033] FIG. 11 illustrates a simplified block diagram of a device that is suitable for implementing example embodiments of the present disclosure; and

[0034] FIG. 12 illustrates a block diagram of an example computer readable medium in accordance with some example embodiments of the present disclosure.

[0035] Throughout the drawings, the same or similar reference numerals represent the same or similar element.DETAILED DESCRIPTION

[0036] Principle of the present disclosure will now be described with reference to some example embodiments. It is to be understood that these embodiments are described only for the purpose of illustration and help those skilled in the art to understand and implement the present disclosure, without suggesting any limitation as to the scope of the disclosure. Embodiments described herein can be implemented in various manners other than the ones described below.

[0037] In the following description and claims, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skills in the art to which this disclosure belongs.

[0038] References in the present disclosure to “one embodiment,” “an embodiment,” “an example embodiment,” and the like indicate that the embodiment described may include a particular feature, structure, or characteristic, but it is not necessary that every embodiment includes the particular feature, structure, or characteristic. Moreover, such phrases are not necessarily referring to the same embodiment. Further, when a particular feature, structure, or characteristic is described in connection with an embodiment, it is submitted that it is within the knowledge of one skilled in the art to affect such feature, structure, or characteristic in connection with other embodiments whether or not explicitly described.

[0039] It shall be understood that although the terms “first,” “second,”..., etc. in front of noun(s) and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another and they do not limit the order of the noun(s). For example, a first element could be termed a second element, and similarly, a second element could be termed a first element, without departing from the scope of example embodiments. As used herein, the term “and / or” includes any and all combinations of one or more of the listed terms.

[0040] As used herein, “at least one of the following: ” and “at least one of ” and similar wording, where the list of two or more elements are joined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements.

[0041] As used herein, unless stated explicitly, performing a step “in response to A” does not indicate that the step is performed immediately after “A” occurs and one or more intervening steps may be included.

[0042] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises”, “comprising”, “has”, “having”, “includes” and / or “including”, when used herein, specify the presence of stated features, elements, and / or components etc., but do not preclude the presence or addition of one or more other features, elements, components and / or combinations thereof.

[0043] As used in this application, the term “circuitry” may refer to one or more or all of the following:(a) hardware-only circuit implementations (such as implementations in only analog and / or digital circuitry) and(b) combinations of hardware circuits and software, such as (as applicable):(i) a combination of analog and / or digital hardware circuit(s) with software / firmware and(ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and(c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.

[0044] This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers animplementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.

[0045] As used herein, the term “communication network” refers to a network following any suitable communication standards, such as New Radio (NR), Long Term Evolution (LTE), LTE- Advanced (LTE-A), Wideband Code Division Multiple Access (WCDMA), High-Speed Packet Access (HSPA), Narrow Band Internet of Things (NB-loT) and so on. Furthermore, the communications between a terminal device and a network device in the communication network may be performed according to any suitable generation communication protocols, including, but not limited to, the first generation (1 G), the second generation (2G), 2.5G, 2.75G, the third generation (3G), the fourth generation (4G), 4.5G, the fifth generation (5G), 5.5G, the sixth generation (6G) communication protocols, and / or any other protocols either currently known or to be developed in the future. Embodiments of the present disclosure may be applied in various communication systems. Given the rapid development in communications, there will of course also be future type communication technologies and systems with which the present disclosure may be embodied. It should not be seen as limiting the scope of the present disclosure to only the aforementioned system.

[0046] As used herein, the term “network device” refers to a node in a communication network via which a terminal device accesses the network and receives services therefrom. The network device may refer to a base station (BS) or an access point (AP), for example, a node B (NodeB or NB), an evolved NodeB (eNodeB or eNB), an NR NB (also referred to as a gNB), a Remote Radio Unit (RRU), a radio header (RH), a remote radio head (RRH), a relay, an Integrated Access and Backhaul (IAB) node, a low power node such as a femto, a pico, a non-terrestrial network (NTN) or non-ground network device such as a satellite network device, a low earth orbit (LEO) satellite and a geosynchronous earth orbit (GEO) satellite, an aircraft network device, and so forth, depending on the applied terminology and technology. In some example embodiments, radio access network (RAN) split architecture comprises a Centralized Unit (CU) and a Distributed Unit (DU) at an IAB donor node. An IAB node comprises a Mobile Terminal (IAB-MT) part that behaves like a UE toward the parent node, and a DU part of an IAB node behaves like a base station toward the next-hop IAB node.

[0047] The term “terminal device” refers to any end device that may be capable of wireless communication. By way of example rather than limitation, a terminal device may also be referred to as a communication device, user equipment (UE), a Subscriber Station (SS), a Portable Subscriber Station, a Mobile Station (MS), or an Access Terminal (AT). The terminal device may include, but not limited to, a mobile phone, a cellular phone, a smart phone, voice over IP (VoIP) phones, wireless local loop phones, a tablet, a wearable terminal device, a personal digital assistant (PDA), portable computers, desktop computer, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback appliances, vehiclemounted wireless terminal devices, wireless endpoints, mobile stations, laptop-embedded equipment (LEE), laptop-mounted equipment (LME), USB dongles, smart devices, wireless customer-premises equipment (CPE), an I nternet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medical device and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and / or other wireless devices operating in an industrial and / or an automated processing chain contexts), a consumer electronics device, a device operating on commercial and / or industrial wireless networks, and the like. The terminal device may also correspond to a Mobile Termination (MT) part of an IAB node (e.g., a relay node). In the following description, the terms “terminal device”, “communication device”, “terminal”, “user equipment” and “UE” may be used interchangeably.

[0048] As used herein, the term “resource,” “transmission resource,” “resource block,” “physical resource block” (PRB), “uplink resource,” or “downlink resource” may refer to any resource for performing a communication, for example, a communication between a terminal device and a network device, such as a resource in time domain, a resource in frequency domain, a resource in space domain, a resource in code domain, or any other combination of the time, frequency, space and / or code domain resource enabling a communication, and the like. In the following, unless explicitly stated, a resource in both frequency domain and time domain will be used as an example of a transmission resource for describing some example embodiments of the present disclosure. It is noted that example embodiments of the present disclosure are equally applicable to other resources in other domains.

[0049] The term “data processing model” used herein may refer to an algorithm that is used to process data. The term “AI / ML model” used herein may refer to a data driven algorithm that applies AI / ML techniques to generate a set of outputs based on a set of inputs. The term “AI / ML model” may be interchangeably with the term “data processing model” or “model.” The term “AI / ML entity” used herein may refer to a stand-alone ML model characterized either by a model architecture andaccompanying hyper-parameters, an executable software component (.exe format), or an imagebased software version to be used by a compatible application. An AI / ML entity may also be an application, network managed function, or network management function utilizing a ML model to perform a certain functionality. The term “twin model” used herein may refer to two or more models that use the same dataset, same neural network architecture (model, number of layers, weights, activation functions etc.). The outputs of twin models are expected to be the same. The term “epoch” or “training epoch” used herein may refer to one complete pass through the entire training dataset.

[0050] FIG. 1 illustrates an example communication environment 100 in which example embodiments of the present disclosure can be implemented. In the communication environment 100, a plurality of communication devices, including a first apparatus 110 and a second apparatus 120, can communicate with each other. In some example embodiments, the first apparatus 110 may be a UE and the second apparatus 120 may be a base station serving the UE. Alternatively, the second apparatus 120 may be a UE and the first apparatus 110 may be a base station serving the UE.

[0051] It is to be understood that the number of devices and their connections shown in FIG. 1 are only for the purpose of illustration without suggesting any limitation. The communication environment 100 may include any suitable number of devices configured to implementing example embodiments of the present disclosure. Although not shown, it would be appreciated that one or more additional devices may be located in the cell, and one or more additional cells may be deployed in the communication environment 100. It is noted that although illustrated as a network device, the second apparatus 120 may be another device than a network device. Although illustrated as a terminal device, the first apparatus 110 may be another device than a terminal device.

[0052] In some example embodiments, a link from the second apparatus 120 to the first apparatus 110 is referred to as a downlink (DL), while a link from the first apparatus 110 to the second apparatus 120 is referred to as an uplink (UL). In DL, the second apparatus 120 is a transmitting (TX) device (or a transmitter) and the first apparatus 110 is a receiving (RX) device (or a receiver). In UL, the first apparatus 110 is a TX device (or a transmitter) and the second apparatus 120 is a RX device (or a receiver).

[0053] Communications in the communication environment 100 may be implemented according to any proper communication protocol(s), comprising, but not limited to, cellular communicationprotocols of the first generation (1 G), the second generation (2G), the third generation (3G), the fourth generation (4G), the fifth generation (5G), 5.5G, the sixth generation (6G), and the like, wireless local network communication protocols such as Institute for Electrical and Electronics Engineers (IEEE) 802.11 and the like, and / or any other protocols currently known or to be developed in the future. Moreover, the communication may utilize any proper wireless communication technology, comprising but not limited to: Code Division Multiple Access (CDMA), Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Frequency Division Duplex (FDD), Time Division Duplex (TDD), Multiple-Input Multiple-Output (MIMO), Orthogonal Frequency Division Multiple (OFDM), Discrete Fourier Transform spread OFDM (DFT- s-OFDM) and / or any other technologies currently known or to be developed in the future.

[0054] As mentioned above, LCM of AI / ML models is a crucial aspect for current and future deployments. For example, twin-model architectures have been proposed for upcoming network deployment to improve several performance metrics.

[0055] Twin-model architecture, as shown in FIG. 2, includes models at two different entities that generate identical outputs. The concept may be extended to multiple entities that maintain the same output using the same model available locally to each entity. One use case is channel state information (CSI) feedback compression where the UE estimates the channel using the CSI-RS and feeds back a compressed update of the channel by exploiting twin channel predictors to reduce feedback overhead. As an example, if the prediction at the UE is perfect then feedback overhead can be avoided, else the feedback overhead bit can be reduced to feedback update. The gNB can acquire the CSI as:where denotes the predicted channel, HUEdenotes estimated channel at the UE, and Q(- ) represents standard element-wise quantization function.

[0056] Another example is twin data augmentation algorithms, e.g., generative adversarial networks (GAN). For instance, twin GANs are used to generate identical data at gNB and UE.

[0057] Twin models need to produce identical outputs at different network entities. However, synchronization between the different models need to be ensured prior to inference without model transfer. Also, UE needs to report its capabilities to support architecture of an AI / ML model. Possibly, gNB may also informs its capabilities to support architecture of an AI / ML model.

[0058] In accordance with some example embodiments of the present disclosure, there isprovided a solution for twin-model synchronization. In particular, a first apparatus obtains one or more model parameter for a model and transmits the one or more model parameters to a second apparatus. The first apparatus and the second apparatus obtain a parameter set based on the one or more model parameters. In this way, the synchronization of models at the first apparatus and the second apparatus can be ensured. Further, since the one or more model parameters are exchanged instead of a whole set of parameters, signaling overheads between the first apparatus and the second apparatus can be reduced.

[0059] Example embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0060] Reference is made to FIG. 3, which illustrates a signaling flow 300 of twin model synchronization in accordance with some embodiments of the present disclosure. For the purpose of discussion, the signaling flow 300 will be discussed with reference to FIG. 1 , for example, by using the first apparatus 110 and the second apparatus 120. It is noted that the order of acts / steps shown in FIG. 3 is only an example not limitation. In some example embodiments, the first apparatus 110 is a terminal device and the second apparatus 120 is a network device. Alternatively, the first apparatus 110 is a network device and the second apparatus 120 is a terminal device. In some other example embodiments, both the first apparatus 110 and the second apparatus 120 are terminal devices. In some further example embodiments, both the first apparatus 110 and the second apparatus 120 are network devices. It is noted that FIG. 3 is taken twin models at two entities as an example. Embodiments with reference to FIG. 3 may be applied to other scenarios where more than two models need to be synchronized.

[0061] The first apparatus 110 obtains (3010) one or more model parameters for a model. The model is applied in a communication between the first apparatus 110 and the second apparatus 120. For example, the model may be used in CSI feedback scenario. Alternatively, the model may be used in positioning scenario or twin data augmentation algorithms. It is noted that the model can be applied in any proper communication scenarios.

[0062] The one or more model parameters indicates / derives a parameter set for training process of the model. For example, the parameter set may be derived or generated based on the one or more model parameters. In some example embodiments, the one or more model parameters may include a random number generator seed which may be represented as “Oseed - Alternatively, or in addition, the one or more model parameters may include an indication of a generation function (represented as “RNGinit”). In some further example embodiments, the one or more modelparameters may include an indication of codebook and / or an index of the indication of codebook.

[0063] In some example embodiments, if the first apparatus 110 is the terminal device and the second apparatus is the network device, the first apparatus 110 may obtain (3010’) the one or more model parameters based on a training capability of the first apparatus 110. For example, the training capability may include how many weights can the first apparatus 110 train. Alternatively, or in addition, the training capability may include how many layers can the first apparatus 110 support. It is noted that the training capability can be any suitable capabilities related to model training. For example, the first apparatus 110 may obtain / select (3010’) the random number generator seed and / or the generation function based on the training capability of the first apparatus 110. As another example, the first apparatus 110 may obtain / select (3010’) the indication of codebook (represented as “Cinit”) and / or the index of the indication of codebook based on the training capability of the first apparatus 110. In some embodiments, the first apparatus 110 may select weights of the model form the codebook based on its training capability. The codebook may be agreed in advance at the first apparatus 110 and the second apparatus 120.

[0064] Table 1 below shows example of the codebook and the training capability. For example, if the training capability of the first apparatus 110 is training capability 1 , the first apparatus 110 may select the codebook 1 which indicates the first parameter set. It is noted that Table 1 is only an example not limitation.Table 1

[0065] In some other example embodiments, one or more UE capabilities may be reported to the network device prior training. For example, if the first apparatus 110 is the network device and the second apparatus 120 is the terminal device, the second apparatus 120 may transmit (3005) a training capability supported by the second apparatus 120. In this case, after receiving (3005) the training capability from the second apparatus 120, the first apparatus 110 may obtain (3010”)the one or more model parameters based on the training capacity supported by the second apparatus 120. For example, the first apparatus 110 may obtain / select (3010”) the random number generator seed and / or the generation function based on the training capability of the second apparatus 120. As another example, the first apparatus 110 may obtain / select (3010”) the indication of codebook and / or the index of the indication of codebook based on the training capability of the second apparatus 120. For example, as shown in Table 1 , if the training capability of the second apparatus 120 is training capability 2, the first apparatus 110 may select the codebook 2 which indicates the second parameter set.

[0066] In some further example embodiments, a plurality of models may be predetermined and a plurality of parameter sets may be predetermined. In some example embodiments, one model may correspond to one or more parameter sets. In this case, the first apparatus 110 may select (3010) which parameter set to be used based on its training capability or the training capability from the second apparatus 120.

[0067] The first apparatus 110 transmits (3020) the one or more model parameters to the second apparatus 120. In other words, the second apparatus 120 receives (3020) the one or more model parameters from the first apparatus 110. In this way, signaling overheads between the first apparatus and the second apparatus can be reduced.

[0068] The second apparatus 120 may transmit (3025) an indication (also referred to as “third indication”) regarding whether the second apparatus supports the one or more model parameters to the first apparatus 110. That is, the first apparatus 110 may receive (3025) the third indication from the second apparatus 120. In some example embodiments, if the third indication indicates that the second apparatus 120 does not support the one or more model parameters, the first apparatus 110 may determine one or more updated model parameters. The first apparatus 110 may then transmit (3030) the one or more updated model parameters to the second apparatus 120. In other words, if the second apparatus 120 does not support the one or more model parameters, the second apparatus 120 may receive the one or more updated model parameters from the first apparatus 110.

[0069] The first apparatus 110 may initialize (3035) the parameter set based on the one or more model parameters. The second apparatus 120 may also initialize (3035’) the parameter set based on the one or more model parameters. After the initialization, the second apparatus 120 may transmit (3040) an indication (referred to as “first indication”) of the initialization of the parameter set at the second apparatus 120. That is, the first apparatus 110 may receive (3040) the firstindication from the second apparatus 120.

[0070] In some example embodiments, the parameter set may include one or more layer weights. The term “layer weight” or “weight of the model” may refer to parameters that are learned during the training process. These weights determine the strength and direction of the relationships between the input features and the output predictions. The weights of the AI / ML model may be initialized randomly before training begins. This random initialization helps in breaking symmetry and allows the model to learn different representations. During the training process, the model may adjust the weights iteratively to minimize the difference between the predicted outputs and the actual outputs. Alternatively, or in addition, the parameter set may include the number of layers. In some further example embodiments, the parameter set may also include the number of training epochs. In some other example embodiments, the parameter set may include an optimizer and / or a bias term. The bias term may represent an intercept or the baseline prediction when all input features are zero. It allows the model to make predictions even when the input features do not provide any information. In the present disclosure, when referring weights, bias terms are covered as well.

[0071] In some example embodiments, the first apparatus 110 may generate the parameter set using the random number generator seed (for example, 0seed) and the generation function (such as, RNGinit). As mentioned above, the generation function may be informed to the second apparatus 120 by the first apparatus 110. Alternatively, the generation function may be predefined by the first apparatus 110 and the second apparatus 120. Alternatively, the first apparatus 110 may conduct the codebook based on the one or more model parameters. For example, if the one or more model parameters indicates the index of the codebook 1 (shown in Table 1), the first apparatus 110 may apply the parameter set 1 which is corresponding to codebook 1. The initialization (3035’) of the parameter set at the second apparatus 120 is similar to the initialization (3035) of the parameter set at the first apparatus 110. For example, the second apparatus may generate the parameter set using the random number generator seed and the generation function or may conduct the codebook based on the one or more model parameters. Details of the initialization (3035’) are omitted here to avoid redundancy.

[0072] The first apparatus 110 may perform (3050) the training process of the model based on the parameter set initialized at the first apparatus 110. For example, the first apparatus 110 may train the model based on one or more of: the layer weights, the number of layers, the training epochs, the optimizer or the bias term included in the parameter set. The second apparatus 120may perform (3050’) the training process of the model based on the parameter set initialized at the second apparatus 120. The training (3050’) of the model at the second apparatus 120 is similar to the training (3050) of the model at the first apparatus 110. In this way, the twin models at the first apparatus 110 and the second apparatus 120 can be synchronized without introducing huge signaling overheads. The training of the models at the first apparatus 110 and the second apparatus 120 may begin by using pre-defined training data.

[0073] The second apparatus 120 may transmit (3060) a first output of the model at the second apparatus 120 to the first apparatus 110. For example, the second apparatus 120 may transmit (3060) a value of the loss function at a predefined epoch (such as, EPcheck) to the first apparatus 110. In some example embodiments, the first output may be a vector.

[0074] The first apparatus 110 may compare (3070) the first output with a second output of the model at the first apparatus 110. For example, the first apparatus 110 may compare the value at the predefined epoch from the second apparatus and another value at the predefined epoch at the first apparatus. The first apparatus 110 may transmit (3080) an indication (also referred to as “second indication”) regarding the comparison between the first output and the second output to the second apparatus 120. In other words, the second apparatus 120 may receive the second indication from the first apparatus 110. In this way, it can ensure the synchronization between the twin models.

[0075] In some example embodiments, if a difference between the first output and the second output is below or equal to a first threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. For example, if a mean squared error (such as loss) between the first and second outputs is below or equal to the first threshold (such as, Delta), the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. It is noted that the first threshold may be any suitable value. Alternatively, if a similarity between the first output and the second output is equal to or above a second threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. For example, if a cosine similarity (CS) between the first and second outputs is equal to or above the second threshold, the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. In some example embodiments, a maximum value of the second threshold may be 1 . For example, the second threshold may be 0.9. When the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized, the second indication transmitted by thefirst apparatus 110 may indicate the synchronization between the model at the first apparatus 110 and the model at the second apparatus 120.

[0076] In some other example embodiments, if a difference between the first output and the second output is above the first threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. For example, if a mean squared error between the first and second outputs is above the first threshold, the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. Alternatively, if the similarity between the first output and the second output is below a second threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. For example, if the CS between the first and second outputs is equal to or above the second threshold (such as, 1 or 0.9), the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. When the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized, the second indication transmitted by the first apparatus 110 may indicate the out-of-synchronization between the model at the first apparatus 110 and the model at the second apparatus 120. In some example embodiments, if the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized, the first apparatus 110 may transmit (3090) the parameter set to the second apparatus 120.

[0077] According to example embodiments described with reference to FIG. 3, to ensure the same training / inference outcome on each model, one device initializes weight parameters and conveys these parameters to the other device using a short initialization update. In this way, it can acquire identical weights at different network entities without weight transfer, thereby eliminating overheads for transferring millions of weight parameters.

[0078] Reference is made to FIG. 4, which illustrates a signaling flow 400 of twin model synchronization in accordance with some embodiments of the present disclosure. For the purpose of discussion, the signaling flow 400 will be discussed with reference to FIG. 1 , for example, by using the first apparatus 110 and the second apparatus 120. It is noted that the order of acts / steps shown in FIG. 4 is only an example not limitation. In some example embodiments, the first apparatus 110 is a terminal device and the second apparatus 120 is a network device. Alternatively, the first apparatus 110 is a network device and the second apparatus 120 is a terminal device. In some other example embodiments, both the first apparatus 110 and the second apparatus 120 are terminal devices. In some further example embodiments, both the first apparatus 110 and thesecond apparatus 120 are network devices. It is noted that FIG. 4 is taken twin models at two entities as an example. Embodiments with reference to FIG. 4 may be applied to other scenarios where more than two models need to be synchronized.

[0079] The first apparatus 110 obtains (4010) a parameter set for a model. The model is applied in a communication between the first apparatus 110 and the second apparatus 120. For example, the model may be used in CSI feedback scenario. Alternatively, the model may be used in positioning scenario. It is noted that the model can be applied in any proper communication scenarios.

[0080] In some example embodiments, if the first apparatus 110 is the terminal device and the second apparatus is the network device, the first apparatus 110 may obtain (3010’) the parameter set based on a training capability of the first apparatus 110. For example, the training capability may include how many weights can the first apparatus 110 train. Alternatively, or in addition, the training capability may include how many layers can the first apparatus 110 support. It is noted that the training capability can be any suitable capabilities related to model training. For example, the first apparatus 110 may obtain / select (4010’) weights of the model form the codebook based on its training capability. The codebook may be agreed in advance at the first apparatus 110 and the second apparatus 120.

[0081] In some other example embodiments, one or more UE capabilities may be reported to the network device prior training. For example, if the first apparatus 110 is the network device and the second apparatus 120 is the terminal device, the second apparatus 120 may transmit a training capability supported by the second apparatus 120. In this case, after receiving the training capability from the second apparatus 120, the first apparatus 110 may obtain (4010) the parameter set based on the training capacity supported by the second apparatus 120. For example, the first apparatus 110 may obtain / select the weights based on the training capability of the second apparatus 120.

[0082] The first apparatus 110 transmits (4020) an indication related to the parameter set to the second apparatus 120. In other words, the second apparatus 120 receives (3020) the indication related to the parameter set from the first apparatus 110. In some example embodiments, the indication may be an indication of a codebook. Alternatively, the indication may be an index related to the codebook. In some example embodiments, the first apparatus may obtain the indication related to the parameter set from the codebook. In this way, signaling overheads between the first apparatus and the second apparatus can be reduced.

[0083] The second apparatus 120 may transmit an indication regarding whether the second apparatus supports the parameter set parameters to the first apparatus 110. In some example embodiments, if the indication indicates that the second apparatus 120 does not support the parameter set, the first apparatus 110 may determine an updated parameter set. The first apparatus 110 may then transmit the updated parameter set to the second apparatus 120.

[0084] In some example embodiments, the first apparatus may conduct the codebook based on the parameter set. The first apparatus 110 may initialize (4035) the parameter set. The second apparatus 120 may also initialize (4035’) the parameter set. After the initialization, the second apparatus 120 may transmit (4040) an indication (referred to as “first indication”) of the initialization of the parameter set at the second apparatus 120. That is, the first apparatus 110 may receive (4040) the first indication from the second apparatus 120.

[0085] In some example embodiments, the parameter set may include one or more layer weights. The term “layer weight” or “weight of the model” may refer to parameters that are learned during the training process. These weights determine the strength and direction of the relationships between the input features and the output predictions. The weights of the AI / ML model may be initialized randomly before training begins. This random initialization helps in breaking symmetry and allows the model to learn different representations. During the training process, the model may adjust the weights iteratively to minimize the difference between the predicted outputs and the actual outputs. Alternatively, or in addition, the parameter set may include the number of layers. In some further example embodiments, the parameter set may also include the number of training epochs. In some other example embodiments, the parameter set may include an optimizer and / or a bias term. The bias term may represent an intercept or the baseline prediction when all input features are zero. It allows the model to make predictions even when the input features do not provide any information. In the present disclosure, when referring weights, bias terms are covered as well.

[0086] The first apparatus 110 may perform (4050) the training process of the model based on the parameter set initialized at the first apparatus 110. For example, the first apparatus 110 may train the model based on one or more of: the layer weights, the number of layers, the training epochs, the optimizer or the bias term included in the parameter set. The second apparatus 120 may perform (4050’) the training process of the model based on the parameter set initialized at the second apparatus 120. The training (4050’) of the model at the second apparatus 120 is similar to the training (4050) of the model at the first apparatus 110. In this way, the twin modelsat the first apparatus 110 and the second apparatus 120 can be synchronized without introducing huge signaling overheads. The training of the models at the first apparatus 110 and the second apparatus 120 may begin by using pre-defined training data.

[0087] The second apparatus 120 may transmit (4060) a first output of the model at the second apparatus 120 to the first apparatus 110. For example, the second apparatus 120 may transmit (4060) a value of the loss function at a predefined epoch (such as, EPcheck) to the first apparatus 110. In some example embodiments, the first output may be a vector.

[0088] The first apparatus 110 may compare (4070) the first output with a second output of the model at the first apparatus 110. For example, the first apparatus 110 may compare the value at the predefined epoch from the second apparatus and another value at the predefined epoch at the first apparatus. The first apparatus 110 may transmit (4080) an indication (also referred to as “second indication”) regarding the comparison between the first output and the second output to the second apparatus 120. In other words, the second apparatus 120 may receive the second indication from the first apparatus 110. In this way, it can ensure the synchronization between the twin models.

[0089] In some example embodiments, if a difference between the first output and the second output is below or equal to a first threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. For example, if a mean squared error (such as loss) between the first and second outputs is below or equal to the first threshold (such as, Delta), the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. It is noted that the first threshold may be any suitable value. Alternatively, if a similarity between the first output and the second output is equal to or above a second threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. For example, if a cosine similarity (CS) between the first and second outputs is equal to or above the second threshold, the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized. In some example embodiments, a maximum value of the second threshold may be 1 . For example, the second threshold may be 0.9. When the model at the first apparatus 110 and the model at the second apparatus 120 are synchronized, the second indication transmitted by the first apparatus 110 may indicate the synchronization between the model at the first apparatus 110 and the model at the second apparatus 120.

[0090] In some other example embodiments, if a difference between the first output and thesecond output is above the first threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. For example, if a mean squared error between the first and second outputs is above the first threshold, the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. Alternatively, if the similarity between the first output and the second output is below a second threshold, the first apparatus 110 may determine that the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. For example, if the CS between the first and second outputs is equal to or above the second threshold (such as, 1 or 0.9), the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized. When the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized, the second indication transmitted by the first apparatus 110 may indicate the out-of-synchronization between the model at the first apparatus 110 and the model at the second apparatus 120. In some example embodiments, if the model at the first apparatus 110 and the model at the second apparatus 120 are not synchronized, the first apparatus 110 may transmit (4090) the parameter set to the second apparatus 120.

[0091] FIG. 5 shows an example of a flowchart of embodiments. In the first step, the first network entity, e.g., a UE, may initialize the weight parameters using the process of the random number generator seed or the codebook (for example, as discussed in FIG. 3 and FIG. 4). The UE may convey the used parameters to the gNB for generating the identical weights. Once the weights are initialized using the previous step, the training process can begin at both the network entities. Optionally, during the training phase, one of the network entities can check the loss at a predefined epoch and share the value with the other network entity. If the loss value is same, it means that both the network entities are learning the same target. Also, at the end of training, if both the models have generated identical output then it means that the synchronization is perfect. Otherwise, the entity that has achieved the objective of, e.g., accurately predicting the channel, can share the entire model with the other network entity.

[0092] FIG. 6 shows an example of required signaling for some embodiments of the present disclosure, which involve UE 610 and gNB 620. In summary, in the beginning, both the network entities (i.e., the UE 610 and gNB 620) agree 6005 to which of the entities can initiate the process. For example, the UE 610 may decide to initiate then it reports the information to gNB 620. After initializing the model using either of the embodiment, the UE 610 may report (6010; 6010’) one or more parameters to generate initialization weights. The UE 610 may transmit (6020) its capability (such as, NN weights can be trained) to the gNB 620. After initialization of weights at the gNB 620,the gNB 620 may inform (6025) the UE that weight parameters are initialized. Both the UE 610 and the gNB 620 may agree (6030) on the training parameters, e.g., epochs, length of dataset. Next, the training begins. Optionally, during the training phase, the gNB 620 can share (6035), e.g., loss value of the model at a pre-defined epoch. The UE 610 can compare the loss value and accordingly report (6040; 6040’) to the gNB 620. For example, if the two models are synchronized, the UE 610 may inform (6040) an acknowledgement (ACK) to the gNB 620. Alternatively, if the two models are not synchronized, the UE 610 may inform (6040’) a negative-acknowledgement (NACK) to the gNB 620. Further, if the two models are not synchronized, the UE 610 may share (6050) its model parameters with the gNB 620.

[0093] FIG. 7 shows a flowchart of an example method 700 implemented at a first apparatus in accordance with some example embodiments of the present disclosure. For the purpose of discussion, the method 700 will be described from the perspective of the first apparatus 110 in FIG. 1.

[0094] At block 710, the first apparatus obtains one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus. The one or more model parameters indicates a parameter set for a training process of the model.

[0095] At block 720, the first apparatus transmits, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

[0096] In some example embodiments, the one or more model parameters comprises at least one of: a random number generator seed, an indication of a generation function, an indication of codebook, or an indices of the indication of codebook.

[0097] In some example embodiments, the method 700 further comprises: initializing the parameter set based on the one or more model parameters.

[0098] In some example embodiments, the method 700 further comprises: generating the parameter set using the random number generator seed and the generation function; or conducting the codebook based on the one or more model parameters.

[0099] In some example embodiments, the method 700 further comprises: receiving, from the second apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0100] In some example embodiments, the method 700 further comprises: performing the training process of the model based on the initialized parameter set.

[0101] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0102] In some example embodiments, the method 700 further comprises: receiving, from the second apparatus, a first output of the model at the second apparatus; comparing the first output with a second output of the model at the first apparatus; and transmitting, to the second apparatus, a second indication regarding the comparison between the first output and the second output.

[0103] In some example embodiments, the method 700 further comprises: based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, determining that the model at the first apparatus and the model at the second apparatus are synchronized; and transmitting, to the second apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0104] In some example embodiments, the method 700 further comprises: based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, determining that the model at the first apparatus and the model at the second apparatus are not synchronized; and transmitting, to the second apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0105] In some example embodiments, the method 700 further comprises: transmitting the parameter set to the second apparatus.

[0106] In some example embodiments, the method 700 further comprises: obtaining one or more model parameters based on training capability supported by the first apparatus.

[0107] In some example embodiments, the method 700 further comprises: selecting the random number generator seed based on the training capability supported by the first apparatus; or selecting the codebook from a predefined codebook based on the training capability supported by the first apparatus.

[0108] In some example embodiments, the method 700 further comprises: receiving, from the second apparatus, a training capability supported by the second apparatus; and obtaining the one or more model parameters based on the training capability supported by the second apparatus.

[0109] In some example embodiments, the method 700 further comprises: selecting the random number generator seed based on the training capability supported by the second apparatus; or selecting the codebook from a predefined codebook based on the training capability supported by the second apparatus.

[0110] In some example embodiments, the method 700 further comprises: receiving, from the second apparatus, a third indication regarding whether the second apparatus supports the one or more model parameters; and based on a determination that the second apparatus does not supports the one or more model parameters, transmitting one or more updated model parameters to the second apparatus.

[0111] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0112] FIG. 8 shows a flowchart of an example method 800 implemented at a second apparatus in accordance with some example embodiments of the present disclosure. For the purpose of discussion, the method 800 will be described from the perspective of the second apparatus 120 in FIG. 1.

[0113] At block 810, the second apparatus receives, from a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus. The one or more model parameters indicates a parameter set for a training process of the model.

[0114] In some example embodiments, the one or more model parameters comprises at least one of: a random number generator seed, an indication of a generation function, an indication of codebook, or an indices of the indication of codebook.

[0115] In some example embodiments, the method 800 further comprises: initializing the parameter set based on the one or more model parameters.

[0116] In some example embodiments, the method 800 further comprises: generating the parameter set using the random number generator seed and the generation function; or conducting the codebook based on the one or more model parameters.

[0117] In some example embodiments, the method 800 further comprises: transmitting, to the first apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0118] In some example embodiments, the method 800 further comprises: performing the training process of the model based on the initialized parameter set.

[0119] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0120] In some example embodiments, the method 800 further comprises: transmitting, to the first apparatus, a first output of the model at the second apparatus; and receiving, from the first apparatus, a second indication regarding the comparison between the first output and a second output of the model at the first apparatus.

[0121] In some example embodiments, the method 800 further comprises: based on a determination that a difference between the first output and the second output is less than or equal to a threshold, receiving from the first apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0122] In some example embodiments, the method 800 further comprises: based on a determination that a difference between the first output and the second output is larger than or equal to a threshold, receiving from the first apparatus, the second indication indicating out-of- synchronization between the model at the first apparatus and the model at the second apparatus.

[0123] In some example embodiments, the method 800 further comprises: receiving the parameter set from the first apparatus.

[0124] In some example embodiments, the method 800 further comprises: transmitting, to the first apparatus, training capability supported by the second apparatus.

[0125] In some example embodiments, the method 800 further comprises: transmitting, to the first apparatus, a third indication regarding whether the second apparatus supports the one or more model parameters; and based on a determination that the second apparatus does not supports the one or more model parameters, receiving one or more updated model parameters from the first apparatus.

[0126] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0127] FIG. 9 shows a flowchart of an example method 900 implemented at a first apparatus in accordance with some example embodiments of the present disclosure. For the purpose ofdiscussion, the method 900 will be described from the perspective of the first apparatus 110 in FIG. 1.

[0128] At block 910, the first apparatus obtains a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus. The parameter set is used in a training process of the model.

[0129] At block 920, the first apparatus transmits, to the second apparatus, an indication related to the parameter set for initializing the parameter set at the second apparatus.

[0130] In some example embodiments, the indication comprises an indication of a codebook or an index related to the codebook.

[0131] In some example embodiments, the method 900 comprises conducting the codebook based on the parameter set.

[0132] In some example embodiments, the method 900 comprises obtaining the indication related to the parameter set from a codebook.

[0133] In some example embodiments, the method 900 comprises receiving, from the second apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0134] In some example embodiments, the method 900 comprises performing the training process of the model based on the parameter set.

[0135] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0136] In some example embodiments, the method 900 comprises receiving, from the second apparatus, a first output of the model at the second apparatus; comparing the first output with a second output of the model at the first apparatus; and transmitting, to the second apparatus, a second indication regarding the comparison between the first output and the second output.

[0137] In some example embodiments, the method 900 comprises based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, determining that the model at the first apparatus and the model at the second apparatus are synchronized; and transmitting, to the second apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0138] In some example embodiments, the method 900 comprises based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, determining that the model at the first apparatus and the model at the second apparatus are not synchronized; and transmitting, to the second apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0139] In some example embodiments, the method 900 comprises transmitting the parameter set to the second apparatus.

[0140] In some example embodiments, the apparatus is a terminal device and the second apparatus is a network device, and wherein the method 900 comprises determining the parameter set based on a training capability of the first apparatus.

[0141] In some example embodiments, the method 900 comprises selecting the codebook from a predefined codebook based on the training capability supported by the first apparatus.

[0142] In some example embodiments, the first apparatus is a network device and the second apparatus is a terminal device, and the method 900 comprises receiving, from the second apparatus, a training capability supported by the second apparatus; and obtaining the parameter set based on the training capability supported by the second apparatus.

[0143] In some example embodiments, the method 900 comprises selecting the codebook from a predefined codebook based on the training capability supported by the second apparatus.

[0144] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0145] FIG. 10 shows a flowchart of an example method 1000 implemented at a second apparatus in accordance with some example embodiments of the present disclosure. For the purpose of discussion, the method 1000 will be described from the perspective of the second apparatus 120 in FIG. 1.

[0146] At block 1010, the second apparatus receives, from a first apparatus, an indication related to a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model.

[0147] In some example embodiments, the indication comprises an indication of a codebook or an index related to the codebook.

[0148] In some example embodiments, the method 1000 comprises conducting the codebook based on the parameter set.

[0149] In some example embodiments, the method 1000 comprises determining the parameter set based on the indication from a codebook.

[0150] In some example embodiments, the method 1000 comprises transmitting, to the first apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0151] In some example embodiments, the method 1000 comprises performing the training process of the model based on the parameter set.

[0152] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0153] In some example embodiments, the method 1000 comprises transmitting, to the first apparatus, a first output of the model at the second apparatus; and receiving, from the first apparatus, a second indication regarding the comparison between the first output and a second output of the model at the first apparatus.

[0154] In some example embodiments, the method 1000 comprises based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, receiving from the first apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0155] In some example embodiments, the method 1000 comprises based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, receiving from the first apparatus, the second indication indicating out- of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0156] In some example embodiments, the method 1000 comprises receiving the parameter set from the first apparatus.

[0157] In some example embodiments, the first apparatus is a network device and the second apparatus is a terminal device, and the method 1000 comprises transmitting, to the first apparatus, a training capability supported by the second apparatus.

[0158] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0159] I n some example embodiments, a first apparatus capable of performing any of the method 700 (for example, the first apparatus 110 in FIG. 1) may comprise means for performing the respective operations of the method 700. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module. The first apparatus may be implemented as or included in the first apparatus 110 in FIG. 1 .

[0160] In some example embodiments, the first apparatus comprises means for obtaining one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and means for transmitting, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

[0161] In some example embodiments, the one or more model parameters comprises at least one of: a random number generator seed, an indication of a generation function, an indication of codebook, or an indices of the indication of codebook.

[0162] In some example embodiments, the first apparatus further comprises: means for initializing the parameter set based on the one or more model parameters.

[0163] In some example embodiments, the first apparatus further comprises: means for generating the parameter set using the random number generator seed and the generation function; or means for conducting the codebook based on the one or more model parameters.

[0164] In some example embodiments, the first apparatus further comprises: means for receiving, from the second apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0165] In some example embodiments, the first apparatus further comprises: means for performing the training process of the model based on the initialized parameter set.

[0166] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0167] In some example embodiments, the first apparatus further comprises: means for receiving, from the second apparatus, a first output of the model at the second apparatus; means for comparing the first output with a second output of the model at the first apparatus; and means for transmitting, to the second apparatus, a second indication regarding the comparison between the first output and the second output.

[0168] In some example embodiments, the first apparatus further comprises: means for based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, determining that the model at the first apparatus and the model at the second apparatus are synchronized; and means for transmitting, to the second apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0169] In some example embodiments, the first apparatus further comprises: means for based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, determining that the model at the first apparatus and the model at the second apparatus are not synchronized; and means for transmitting, to the second apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0170] In some example embodiments, the first apparatus further comprises: means for transmitting the parameter set to the second apparatus.

[0171] In some example embodiments, the first apparatus further comprises: means for obtaining one or more model parameters based on training capability supported by the first apparatus.

[0172] In some example embodiments, the first apparatus further comprises: means for selecting the random number generator seed based on the training capability supported by the first apparatus; or means for selecting the codebook from a predefined codebook based on the training capability supported by the first apparatus.

[0173] In some example embodiments, the first apparatus further comprises: means for meansfor receiving, from the second apparatus, a training capability supported by the second apparatus; and means for obtaining the one or more model parameters based on the training capability supported by the second apparatus.

[0174] In some example embodiments, the first apparatus further comprises: selecting the random number generator seed based on the training capability supported by the second apparatus; or means for selecting the codebook from a predefined codebook based on the training capability supported by the second apparatus.

[0175] In some example embodiments, the first apparatus further comprises: means for receiving, from the second apparatus, a third indication regarding whether the second apparatus supports the one or more model parameters; and means for based on a determination that the second apparatus does not supports the one or more model parameters, transmitting one or more updated model parameters to the second apparatus.

[0176] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0177] In some example embodiments, a second apparatus capable of performing any of the method 800 (for example, the second apparatus 120 in FIG. 1 ) may comprise means for performing the respective operations of the method 800. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module. The second apparatus may be implemented as or included in the second apparatus 120 in FIG. 1.

[0178] In some example embodiments, the second apparatus comprises means for receiving, from a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model.

[0179] In some example embodiments, the one or more model parameters comprises at least one of: a random number generator seed, an indication of a generation function, an indication of codebook, or an indices of the indication of codebook.

[0180] In some example embodiments, the second apparatus further comprises: means for initializing the parameter set based on the one or more model parameters.

[0181] In some example embodiments, the second apparatus further comprises: means for generating the parameter set using the random number generator seed and the generation function;or means for conducting the codebook based on the one or more model parameters.

[0182] In some example embodiments, the second apparatus further comprises: means for transmitting, to the first apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0183] In some example embodiments, the second apparatus further comprises: means for performing the training process of the model based on the initialized parameter set.

[0184] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0185] In some example embodiments, the second apparatus further comprises: means for transmitting, to the first apparatus, a first output of the model at the second apparatus; and means for receiving, from the first apparatus, a second indication regarding the comparison between the first output and a second output of the model at the first apparatus.

[0186] In some example embodiments, the second apparatus further comprises: means for based on a determination that a difference between the first output and the second output is less than or equal to a threshold, receiving from the first apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0187] In some example embodiments, the second apparatus further comprises: means for based on a determination that a difference between the first output and the second output is larger than or equal to a threshold, receiving from the first apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0188] In some example embodiments, the second apparatus further comprises: means for receiving the parameter set from the first apparatus.

[0189] In some example embodiments, the second apparatus further comprises: means for transmitting, to the first apparatus, training capability supported by the second apparatus.

[0190] In some example embodiments, the second apparatus further comprises: means for transmitting, to the first apparatus, a third indication regarding whether the second apparatus supports the one or more model parameters; and means for based on a determination that the second apparatus does not supports the one or more model parameters, receiving one or moreupdated model parameters from the first apparatus.

[0191] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0192] In some example embodiments, a first apparatus capable of performing any of the method 900 (for example, the first apparatus 110 in FIG. 1) may comprise means for performing the respective operations of the method 900. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module. The first apparatus may be implemented as or included in the first apparatus 110 in FIG. 1 .

[0193] In some example embodiments, the first apparatus comprises means for obtaining a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model; and means for transmitting, to the second apparatus, an indication related to the parameter set for initializing the parameter set at the second apparatus.

[0194] In some example embodiments, the indication comprises an indication of a codebook or an index related to the codebook.

[0195] In some example embodiments, the first apparatus comprises means for conducting the codebook based on the parameter set.

[0196] In some example embodiments, the first apparatus comprises means for obtaining the indication related to the parameter set from a codebook.

[0197] In some example embodiments, the first apparatus comprises means for receiving, from the second apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0198] In some example embodiments, the first apparatus comprises means for performing the training process of the model based on the parameter set.

[0199] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0200] In some example embodiments, the first apparatus comprises means for receiving, from the second apparatus, a first output of the model at the second apparatus; means for comparing the first output with a second output of the model at the first apparatus; and means for transmitting,to the second apparatus, a second indication regarding the comparison between the first output and the second output.

[0201] In some example embodiments, the first apparatus comprises means for based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, determining that the model at the first apparatus and the model at the second apparatus are synchronized; and means for transmitting, to the second apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0202] In some example embodiments, the first apparatus comprises means for based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, determining that the model at the first apparatus and the model at the second apparatus are not synchronized; and means for transmitting, to the second apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0203] In some example embodiments, the first apparatus comprises means for transmitting the parameter set to the second apparatus.

[0204] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, and wherein the first apparatus comprises means for determining the parameter set based on a training capability of the first apparatus.

[0205] In some example embodiments, the first apparatus comprises means for selecting the codebook from a predefined codebook based on the training capability supported by the first apparatus.

[0206] In some example embodiments, the first apparatus is a network device and the second apparatus is a terminal device, and wherein the first apparatus comprises means for receiving, from the second apparatus, a training capability supported by the second apparatus; and means for obtaining the parameter set based on the training capability supported by the second apparatus.

[0207] In some example embodiments, the first apparatus comprises means for selecting the codebook from a predefined codebook based on the training capability supported by the second apparatus.

[0208] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0209] In some example embodiments, a second apparatus capable of performing any of the method 1000 (for example, the second apparatus 120 in FIG. 1 ) may comprise means for performing the respective operations of the method 1000. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module. The second apparatus may be implemented as or included in the second apparatus 120 in FIG. 1.

[0210] In some example embodiments, the second apparatus comprises means for receiving, from a first apparatus, an indication related to a parameter set for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the parameter set is used in a training process of the model.

[0211] In some example embodiments, the indication comprises an indication of a codebook or an index related to the codebook.

[0212] In some example embodiments, the second apparatus comprises means for conducting the codebook based on the parameter set.

[0213] In some example embodiments, the second apparatus comprises means for determining the parameter set based on the indication from a codebook.

[0214] In some example embodiments, the second apparatus comprises means for transmitting, to the first apparatus, a first indication of the initialization of the parameter set at the second apparatus.

[0215] In some example embodiments, the second apparatus comprises means for performing the training process of the model based on the parameter set.

[0216] In some example embodiments, the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

[0217] In some example embodiments, the second apparatus comprises means for transmitting, to the first apparatus, a first output of the model at the second apparatus; and means for receiving, from the first apparatus, a second indication regarding the comparison between the first outputand a second output of the model at the first apparatus.

[0218] In some example embodiments, the second apparatus comprises means for based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, receiving from the first apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

[0219] In some example embodiments, the second apparatus comprises means for based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, receiving from the first apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

[0220] In some example embodiments, the second apparatus comprises means for receiving the parameter set from the first apparatus.

[0221] In some example embodiments, the first apparatus is a network device and the second apparatus is a terminal device, and the second apparatus comprises means for transmitting, to the first apparatus, a training capability supported by the second apparatus.

[0222] In some example embodiments, the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

[0223] FIG. 11 is a simplified block diagram of a device 1100 that is suitable for implementing example embodiments of the present disclosure. The device 1100 may be provided to implement a communication device, for example, the first apparatus 110 or the second apparatus 120 as shown in FIG. 1 . As shown, the device 1100 includes one or more processors 1110, one or more memories 1120 coupled to the processor 1110, and one or more communication modules 1140 coupled to the processor 1110.

[0224] The communication module 1140 is for bidirectional communications. The communication module 1140 has one or more communication interfaces to facilitate communication with one or more other modules or devices. The communication interfaces may represent any interface that is necessary for communication with other network elements. In some example embodiments, thecommunication module 1140 may include at least one antenna.

[0225] The processor 1110 may be of any type suitable to the local technical network and may include one or more of the following: general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs) and processors based on multicore processor architecture, as non-limiting examples. The device 1100 may have multiple processors, such as an application specific integrated circuit chip that is slaved in time to a clock which synchronizes the main processor.

[0226] The memory 1120 may include one or more non-volatile memories and one or more volatile memories. Examples of the non-volatile memories include, but are not limited to, a Read Only Memory (ROM) 1124, an electrically programmable read only memory (EPROM), a flash memory, a hard disk, a compact disc (CD), a digital video disk (DVD), an optical disk, a laser disk, and other magnetic storage and / or optical storage. Examples of the volatile memories include, but are not limited to, a random-access memory (RAM) 1122 and other volatile memories that will not last in the power-down duration.

[0227] A computer program 1130 includes computer executable instructions that are executed by the associated processor 1110. The instructions of the program 1130 may include instructions for performing operations / acts of some example embodiments of the present disclosure. The program 1130 may be stored in the memory, e.g., the ROM 1124. The processor 1110 may perform any suitable actions and processing by loading the program 1130 into the RAM 1122.

[0228] The example embodiments of the present disclosure may be implemented by means of the program 1130 so that the device 1100 may perform any process of the disclosure as discussed with reference to FIG. 2 to FIG. 10. The example embodiments of the present disclosure may also be implemented by hardware or by a combination of software and hardware.

[0229] In some example embodiments, the program 1130 may be tangibly contained in a computer readable medium which may be included in the device 1100 (such as in the memory 1120) or other storage devices that are accessible by the device 1100. The device 1100 may load the program 1130 from the computer readable medium to the RAM 1122 for execution. In some example embodiments, the computer readable medium may include any types of non-transitory storage medium, such as ROM, EPROM, a flash memory, a hard disk, CD, DVD, and the like. The term “non-transitory,” as used herein, is a limitation of the medium itself (i.e. , tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM).

[0230] FIG. 12 shows an example of the computer readable medium 1200 which may be in form of CD, DVD or other optical storage disk. The computer readable medium 1200 has the program 1130 stored thereon.

[0231] Generally, various embodiments of the present disclosure may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects may be implemented in hardware, and other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device. Although various aspects of embodiments of the present disclosure are illustrated and described as block diagrams, flowcharts, or using some other pictorial representations, it is to be understood that the block, apparatus, system, technique or method described herein may be implemented in, as nonlimiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.

[0232] Some example embodiments of the present disclosure also provide at least one computer program product tangibly stored on a computer readable medium, such as a non-transitory computer readable medium. The computer program product includes computer-executable instructions, such as those included in program modules, being executed in a device on a target physical or virtual processor, to carry out any of the methods as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, or the like that perform particular tasks or implement particular abstract data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. Machine-executable instructions for program modules may be executed within a local or distributed device. In a distributed device, program modules may be located in both local and remote storage media.

[0233] Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. The program code may be provided to a processor or controller of a general-purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

[0234] In the context of the present disclosure, the computer program code or related data maybe carried by any suitable carrier to enable the device, apparatus or processor to perform various processes and operations as described above. Examples of the carrier include a signal, computer readable medium, and the like.

[0235] The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable medium may include but not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium would include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random-access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0236] Further, although operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, although several specific implementation details are contained in the above discussions, these should not be construed as limitations on the scope of the present disclosure, but rather as descriptions of features that may be specific to particular embodiments. Unless explicitly stated, certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, unless explicitly stated, various features that are described in the context of a single embodiment may also be implemented in a plurality of embodiments separately or in any suitable sub-combination.

[0237] Although the present disclosure has been described in languages specific to structural features and / or methodological acts, it is to be understood that the present disclosure defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

CLAIMS1 . A first apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus to: obtain one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model; and transmit, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

2. The first apparatus of claim 1 , wherein the one or more model parameters comprises at least one of: a random number generator seed, an indication of a generation function, an indication of codebook, or an indices of the indication of codebook, and / or wherein the parameter set comprises at least one of: one or more layer weights, number of layers, number of training epochs, an optimizer, a bias term.

3. The first apparatus of claim 1 or 2, wherein the first apparatus is caused to: generate the parameter set using the random number generator seed and the generation function; or conduct the codebook based on the one or more model parameters.

4. The first apparatus of any of claims 1 -3, wherein the first apparatus is further caused to: receive, from the second apparatus, a first indication of the initialization of the parameter set at the second apparatus.

5. The first apparatus of any of claims 1 -4, wherein the first apparatus is further caused to: perform the training process of the model based on the initialized parameter set.

6. The first apparatus of any of claims 1 -5, wherein the first apparatus is further caused to: receive, from the second apparatus, a first output of the model at the second apparatus; compare the first output with a second output of the model at the first apparatus; andtransmit, to the second apparatus, a second indication regarding the comparison between the first output and the second output.

7. The first apparatus of claim 6, wherein the first apparatus is further caused to: based on a determination that a difference between the first output and the second output is below or equal to a first threshold or a determination that a similarity between the first output and the second output is equal to or above a second threshold, determine that the model at the first apparatus and the model at the second apparatus are synchronized; and transmit, to the second apparatus, the second indication indicating the synchronization between the model at the first apparatus and the model at the second apparatus.

8. The first apparatus of claim 6, wherein the first apparatus is further caused to: based on a determination that a difference between the first output and the second output is above or equal to a first threshold or a determination that a similarity between the first output and the second output is below a second threshold, determine that the model at the first apparatus and the model at the second apparatus are not synchronized; and transmit, to the second apparatus, the second indication indicating out-of-synchronization between the model at the first apparatus and the model at the second apparatus.

9. The first apparatus of claim 8, wherein the first apparatus is further caused to: transmit the parameter set to the second apparatus.

10. The first apparatus of any of claims 1-9, wherein the apparatus is a terminal device and the second apparatus is a network device, and wherein the first apparatus is caused to: obtain one or more model parameters based on training capability supported by the first apparatus.11 . The first apparatus of claim 10, wherein the first apparatus is at least caused to: select the random number generator seed based on the training capability supported by the first apparatus; or select the codebook from a predefined codebook based on the training capability supported by the first apparatus.

12. The first apparatus of any of claims 1 -9, wherein the first apparatus is a network device and the second apparatus is a terminal device, and wherein the first apparatus is caused to: receive, from the second apparatus, a training capability supported by the second apparatus; and obtain the one or more model parameters based on the training capability supported by the second apparatus.

13. The first apparatus of claim 12, wherein the first apparatus is at least caused to: select the random number generator seed based on the training capability supported by the second apparatus; or select the codebook from a predefined codebook based on the training capability supported by the second apparatus.

14. The first apparatus of claim 12, wherein the first apparatus is further caused to: receive, from the second apparatus, a third indication regarding whether the second apparatus supports the one or more model parameters; and based on a determination that the second apparatus does not supports the one or more model parameters, transmit one or more updated model parameters to the second apparatus.

15. The first apparatus of any of claims 1 -9, wherein the first apparatus is a terminal device and the second apparatus is a network device, or wherein the first apparatus is a network device and the second apparatus is a terminal device.

16. A method comprising: obtaining, at a first apparatus, one or more model parameters for a model which is applied in a communication between the first apparatus and a second apparatus, wherein the one or more model parameters indicates a parameter set for a training process of the model. transmitting, to a second apparatus, the one or more model parameters for initializing the parameter set at the second apparatus.

Images