Control of machine learning training and inference

Abstract

Necessity for a machine learning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect is determined, a machine learning training request is transmitted, upon reception of which, training takes place. Necessity, at a machine learning training timing, for the machine learning entity is determined, and a corresponding machine learning training configuration is transmitted and received. A machine learning inference capability report indicative of support of machine learning inferences, is transmitted, received, and considered for a machine learning inference request decision. The network aspects may be energy saving, load balancing, or mobility optimisation. The sub-aspects may be beam on / off prediction, cell on / off prediction, handover candidate beam determination, handover candidate cell determination, cell energy cost prediction, base station energy cost prediction, cell load prediction, energy efficiency prediction, target cell prediction, source cell resource status prediction, neighbouring cell resource status prediction, prediction of terminals to be handed over to a target cell, terminal trajectory prediction, terminal traffic prediction, prediction of handover target cell, or prediction of handover candidate cells.

Description

Field Various example embodiments relate to control of machine learning training and inference. More specifically, various example embodiments exemplarily relate to measures (including methods, apparatuses and computer program products) for realizing control of machine learning training and inference. Background The present specification generally relates to management of artificial intelligence (AI) and machine learning (ML), particularly in the context of (mobile) networks. While an AI / ML training function may be modeled in a generic way, for AI / ML inference, specific AI / ML inference functions, e.g. related to energy saving, mobility load balancing, mobility robust optimization, analytics logical function (AnLF), management data analytics (MDA), etc., may be modelled separately. The main reason being that the characteristics / capabilities of AI / ML inference functions may be specific to each use case (e.g. energy saving, (mobility) load balancing, mobility (robust) optimization) and cannot always be generalized. For the example use case (AI / ML-based) network energy saving, the following output data of AI / ML inference may for example be applicable: Energy saving strategy, such as recommended cell activation / deactivation, Handover strategy, including recommended candidate cells for taking over the traffic, Predicted energy efficiency, Predicted energy state (e.g., active, high, low, inactive). For the example use case (AI / ML-based) load balancing, the following output data of AI / ML inference may for example be applicable: Selection of target cell for load balancing, Predicted own resource status information, Predicted resource status information of neighboring NG-RAN node(s), Predicted UE(s) selected to be handed over to target NG-RAN node (used by radio access network (RAN) node internally). Considering, for example, the use case AI / ML-based RAN energy saving, an AI / ML entity (composed of one or more ML models) for RAN energy saving may be trained using available training data to provide either a single AI / ML inference output or multiple AI / ML inference outputs. Examples of such AI / ML inference outputs for RAN energy saving may include: (i) Beam on-off prediction, (ii) Cell on-off prediction, (iii) Candidate beam to handover the traffic before switching off the serving beam, (iv) Candidate cell to handover the traffic before switching off the serving cell, (v) Cell energy cost prediction, (vi) gNodeB energy cost prediction, (vii) Cell load prediction, (viii) Energy efficiency prediction, etc. An AI / ML training consumer can request the AI / ML training producer to train an AI / ML entity, e.g., for RAN energy saving (or other use cases such as mobility load balancing, mobility robust optimization). Once the AI / ML entity is trained and loaded into an AI / ML inference producer, the AI / ML inference consumer can request the AI / ML inference producer to generate and provide AI / ML inference reports e.g., for RAN energy saving (or other use cases such as mobility load balancing, mobility robust optimization). This currently known approach leads to inflexible AI / ML training and inference and in particular AI / ML training and inference hardly adaptable to AI / ML training and inference consumer needs. Hence, the problem arises that no flexible AI / ML training and inference and in particular no AI / ML training and inference easily adaptable to AI / ML training and inference consumer needs can be provided. Hence, there is a need to provide for control of machine learning training and inference. Summary Various example embodiments aim at addressing at least part of the above issues and / or problems and drawbacks. Various aspects of example embodiments are set out in the appended claims. According to an exemplary aspect, there is provided a method, comprising receiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one subaspect of a network control aspect, and training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided a method, comprising receiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided a method, comprising determining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided a method, comprising determining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. According to an exemplary aspect, there is provided a method, comprising generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and transmitting said machine learning inference capability report. According to an exemplary aspect, there is provided a method, comprising receiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and considering said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. According to an exemplary aspect, there is provided an apparatus, comprising receiving circuitry configured to receive a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and training circuitry configured to train said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising receiving circuitry configured to receive a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training circuitry configured to train, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising determining circuitry configured to determine necessity for a machine leaning entity supporting machine learning inference related to at least one subaspect of a network control aspect, and transmitting circuitry configured to transmit a machine learning training request for training said machine leaning 6 entity to support machine learning inference related to said at least one subaspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising determining circuitry configured to determine necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting circuitry configured to transmit a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. According to an exemplary aspect, there is provided an apparatus, comprising generating circuitry configured to generate a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and transmitting circuitry configured to transmit said machine learning inference capability report. According to an exemplary aspect, there is provided an apparatus, comprising receiving circuitry configured to receive a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and considering circuitry configured to consider said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform receiving a machine learning training request for training a machine 7 leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform receiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform determining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform determining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of 8 said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and transmitting said machine learning inference capability report. According to an exemplary aspect, there is provided an apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform receiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and considering said support of machine learning inference related to said at least one supported subaspect of said supported network control aspect for machine learning inference request decision. According to an exemplary aspect, there is provided a computer program product comprising computer-executable computer program code which, when the program is run on a computer (e.g. a computer of an apparatus according to any one of the aforementioned apparatus-related exemplary aspects of the present disclosure), is configured to cause the computer to carry out the method according to any one of the aforementioned method-related exemplary aspects of the present disclosure. Such computer program product may comprise (or be embodied) a (tangible) computer-readable (storage) medium or the like on which the computerexecutable computer program code is stored, and / or the program may be directly loadable into an internal memory of the computer or a processor thereof. Any one of the above aspects enables an efficient detailed and customized control of training of AI / ML entities and of inference output of such trained AI / ML entities to thereby solve at least part of the problems and drawbacks identified in relation to the prior art. By way of example embodiments, there is provided control of machine learning training and inference. More specifically, by way of example embodiments, there are provided measures and mechanisms for realizing control of machine learning training and inference. Thus, improvement is achieved by methods, apparatuses and computer program products enabling / realizing control of machine learning training and inference. Brief description of the drawings In the following, the present disclosure will be described in greater detail by way of non-limiting examples with reference to the accompanying drawings, in which FIG. 1 is a block diagram illustrating an apparatus according to example embodiments, FIG. 2 is a block diagram illustrating an apparatus according to example embodiments, FIG. 3 is a block diagram illustrating an apparatus according to example embodiments, FIG. 4 is a block diagram illustrating an apparatus according to example embodiments, FIG. 5 is a block diagram illustrating an apparatus according to example embodiments, FIG. 6 is a block diagram illustrating an apparatus according to example embodiments, FIG. 7 is a block diagram illustrating an apparatus according to example embodiments, FIG. 8 is a block diagram illustrating an apparatus according to example embodiments, FIG. 9 is a schematic diagram of procedure according to example embodiments FIG. 10 is a schematic diagram of procedure according to example embodiments FIG. 11 is a schematic diagram of procedure according to example embodiments FIG. 12 is a schematic diagram of procedure according to example embodiments FIG. 13 is a schematic diagram of procedure according to example embodiments FIG. 14 is a schematic diagram of a procedure according to example embodiments, FIG. 15 shows a schematic diagram of signaling sequences according to example embodiments, and FIG. 16 is a block diagram alternatively illustrating apparatuses according to example embodiments. Detailed description The present disclosure is described herein with reference to particular nonlimiting examples and to what are presently considered to be conceivable embodiments. A person skilled in the art will appreciate that the disclosure is by no means limited to these examples, and may be more broadly applied. It is to be noted that the following description of the present disclosure and its embodiments mainly refers to specifications being used as non-limiting examples for certain exemplary network configurations and deployments. Namely, the present disclosure and its embodiments are mainly described in relation to 3GPP specifications being used as non-limiting examples for certain exemplary network configurations and deployments. As such, the description of example embodiments given herein specifically refers to terminology which is directly related thereto. Such terminology is only used in the context of the presented non-limiting examples, and does naturally not limit the disclosure in any way. Rather, any other communication or communication related system deployment, etc. may also be utilized as long as compliant with the features described herein. Hereinafter, various embodiments and implementations of the present disclosure and Its aspects or embodiments are described using several variants and / or alternatives. It is generally noted that, according to certain needs and constraints, all of the described variants and / or alternatives may be provided alone or in any conceivable combination (also including combinations of individual features of the various variants and / or alternatives). 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. According to example embodiments, in general terms, there are provided measures and mechanisms for (enabling / realizing) control of machine learning training and inference. As mentioned above, a currently known approach in relation to AI / ML training and inference control leads to inflexible AI / ML training and inference and in particular AI / ML training and inference hardly adaptable to AI / ML training and inference consumer needs. Namely, as mentioned, an AI / ML training consumer can request the AI / ML training producer to train an AI / ML entity, e.g., for RAN energy saving (or other use cases such as mobility load balancing, mobility robust optimization). However, there is no possibility for the AI / ML training consumer e.g. to request the AI / ML training producer to train an AI / ML entity that can support specific (one or multiple) AI / ML inference outputs (e.g. beam on-off prediction and / or cell on-off prediction) for a specific use case (e.g. RAN energy saving). Further, as mentioned, once the AI / ML entity is trained and loaded into an AI / ML inference producer, the AI / ML inference consumer can request the AI / ML inference producer to generate and provide AI / ML inference reports e.g., for RAN energy saving (or other use cases such as mobility load balancing, mobility robust optimization). However, there is no possibility for the AI / ML inference consumer e.g. to request the AI / ML inference producer to generate and provide AI / ML inference reports with specific (one or multiple) AI / ML inference outputs (e.g. beam on-off prediction and / or cell on-off prediction) for a specific use case (e.g. RAN energy saving). In view thereof, in brief, according to example embodiments, an AI / ML training consumer is enabled to request the AI / ML training producer to train an AI / ML entity that can support specific (one or multiple) AI / ML inference outputs (as an example of sub-aspects of a network control aspect) for a specific AI / ML inference type (as an example of a network control aspect), e.g. RAN energy saving. According to example embodiments, this is achieved by an attribute, e.g., AIMLInferenceOutputType (as an example of an identifier indicative of at least one sub-aspect of a network control aspect) in the AI / ML training request to indicate specific (one or multiple) AI / ML inference outputs that the trained ML entity needs to support. In the example use case RAN energy saving, according to example embodiments, this attribute may include: - Beam on-off prediction, - Cell on-off prediction, - Candidate beam to handover the traffic before switching off the serving beam, - Candidate cell to handover the traffic before switching off the serving cell, - Cell energy cost prediction, - gNodeB energy cost prediction, - Cell load prediction, - Energy efficiency prediction, - etc. In the example use case load balancing, according to example embodiments, this attribute may include: - Target cell prediction, - Resource status prediction of source cell, - Resource status prediction of neighboring cell, - Prediction of user equipments (UE) (as an example of terminals) to be handed over to a target cell, - etc. In the example use case mobility optimization, according to example embodiments, this attribute may include: - UE trajectory prediction, - UE traffic prediction, - Predicted target cell for handover, - Predicted candidate cells for handover, - etc. It is to be noted that this attribute is different from existing attribute inferenceType (as an example of an identifier indicative of a network control aspect) which indicates the type of the use case, e.g., RAN energy saving, mobility load balancing, mobility robust optimization, etc. Depending on whether the AI / ML training producer manages to successfully train the AI / ML entity to support the requested AI / ML inference outputs or not, according to example embodiments, the AI / ML training producer indicates to the AI / ML training consumer either success or failure and provides the AI / ML training report for each AI / ML inference output. Additionally, according to example embodiments, once the AI / ML entity is trained, the AI / ML entity can either be stored in an AI / ML entity repository or loaded into an AI / ML inference producer by including the supported AI / ML inference outputs information in the metadata. Still further, in view of the above, in brief, according to example embodiments, an AI / ML inference consumer is enabled to request the AI / ML 15 inference producer to generate and provide AI / ML inference reports with specific (one or multiple) AI / ML inference outputs for RAN energy saving. According to example embodiments, this is achieved by a datatype, e.g., SupportedAIMLInferenceCapabilities, which includes information / attributes related to the supported AI / ML inference type(s) (example of indication of (a) network control aspect(s)) and supported AI / ML inference output(s) (example of indication of (a) sub-aspect(s) of a network control aspect) for each AI / ML inference type. In the case of RAN energy saving, supportedlnferenceType is e.g. 'RAN Energy Saving', and supportedlnferenceOutputType may include: - Beam on-off prediction, - Cell on-off prediction, - Candidate beam to handover the traffic before switching off the serving beam, - Candidate cell to handover the traffic before switching off the serving cell, - Cell energy cost prediction, - gNodeB energy cost prediction, - Cell load prediction, - Energy efficiency prediction, - etc. In the case of load balancing, supportedlnferenceType is e.g. 'Load Balancing', and supportedlnferenceOutputType may include: - Target cell prediction, - Resource status prediction of source cell, - Resource status prediction of neighboring cell, - Prediction of user equipments (UE) (as an example of terminals) to be handed over to a target cell, - etc. In the case of mobility optimization, supportedlnferenceType is e.g. 'Mobility Optimization', and supportedlnferenceOutputType may include: - UE trajectory prediction, - UE traffic prediction, - Predicted target cell for handover, - Predicted candidate cells for handover, - etc. According to example embodiments, an AI / ML inference consumer may request the AI / ML inference producer to report on the supported AI / ML inference type(s) and AI / ML inference output(s) for each AI / ML inference type. The datatype may thus be provided to potential AI / ML inference consumers. According to example embodiments, thus knowing the capabilities, an AI / ML inference consumer may, based on the capabilities, request the AI / ML inference producer to generate and provide AI / ML inference report(s) for specific (one or multiple) AI / ML inference type(s) and (one or multiple) AI / ML inference output(s) for each AI / ML inference type that may be reported by the AI / ML inference producer. Example embodiments are specified below in more detail. FIG. 1 is a block diagram illustrating an apparatus according to example embodiments. The apparatus may be a network node or entity 10 such as a machine learning training producer entity (or a network node or entity providing such functionality) comprising a receiving circuitry 11 and a training circuitry 12. The receiving circuitry 11 receives a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and the training circuitry 12 trains said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Alternatively, the receiving circuitry 11 receives a machine learning training configuration fortraining a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and the training circuitry 12 trains, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. FIG. 9 is a schematic diagram of a procedure according to example embodiments. FIG. 10 is a schematic diagram of a procedure according to example embodiments. The apparatus according to FIG. 1 may perform the method of FIG. 9 or FIG. 10 but is not limited to this method. The method of FIG. 9 or FIG. 10 may be performed by the apparatus of FIG. 1 but is not limited to being performed by this apparatus. As shown in FIG. 9, a procedure according to example embodiments comprises an operation of receiving (S91) a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and an operation of training (S92) said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. As shown in FIG. 10, a procedure according to example embodiments comprises an operation of receiving (S101) a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and an operation of training (S102), meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Meeting said machine learning training timing here means that (timing) requirements indicated by the machine learning training timing are fulfilled. FIG. 2 is a block diagram illustrating an apparatus according to example embodiments. In particular, FIG. 2 illustrates a variation of the apparatus shown in FIG. 1. The apparatus according to FIG. 2 may thus further comprise a generating circuitry 21, an evaluating circuitry 22, and / or a transmitting circuitry 23. In an embodiment at least some of the functionalities of the apparatus shown in FIG. 1 (or 2) may be shared between two physically separate devices forming one operational entity. Therefore, the apparatus may be seen to depict the operational entity comprising one or more physically separate devices for executing at least some of the described processes. According to a variation of the procedure shown in FIG. 9 or FIG. 10, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of generating a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect, and an operation of transmitting said machine learning training report. According to a variation of the procedure shown in FIG. 9 or FIG. 10, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of evaluating said inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. According to a variation of the procedure shown in FIG. 9 or FIG. 10, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of 19 transmitting information on said machine leaning entity to a machine learning entity repository entity or to a machine learning inference producer entity. According to a variation of the procedure shown in FIG. 9 or FIG. 10, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine leaning entity transmission request for transmitting said machine leaning entity to said machine learning inference producer entity. According to further example embodiments, said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning training request includes a second identifier indicative of said network control aspect. According to further example embodiments, said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. FIG. 3 is a block diagram illustrating an apparatus according to example embodiments. The apparatus may be a network node or entity 30 such as a machine learning training consumer entity (or a network node or entity providing such functionality) comprising a determining circuitry 31 and a transmitting circuitry 32. The determining circuitry 31 determines necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and the transmitting circuitry 32 transmits a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. Alternatively, the determining circuitry 31 determines necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and the transmitting circuitry 32 transmits a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. FIG. 11 is a schematic diagram of a procedure according to example embodiments. FIG. 12 is a schematic diagram of a procedure according to example embodiments. The apparatus according to FIG. 3 may perform the method of FIG. 11 or FIG. 12 but is not limited to this method. The method of FIG. 11 or FIG. 12 may be performed by the apparatus of FIG. 3 but is not limited to being performed by this apparatus. As shown in FIG. 11, a procedure according to example embodiments comprises an operation of determining (Sill) necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and an operation of transmitting (S112) a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one subaspect of said network control aspect. As shown in FIG. 12, a procedure according to example embodiments comprises an operation of determining (S121) necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and an operation of transmitting (S122) a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. Necessity, at a machine learning training timing, here means that the necessity is assumed / predicted as being present at the machine learning training timing. Determination of the necessity, thus, may be performed in advance before the machine learning training timing. FIG. 4 is a block diagram illustrating an apparatus according to example embodiments. In particular, FIG. 4 illustrates a variation of the apparatus shown in FIG. 3. The apparatus according to FIG. 4 may thus further comprise a receiving circuitry 41. In an embodiment at least some of the functionalities of the apparatus shown in FIG. 3 (or 4) may be shared between two physically separate devices forming one operational entity. Therefore, the apparatus may be seen to depict the operational entity comprising one or more physically separate devices for executing at least some of the described processes. According to a variation of the procedure shown in FIG. 11 or FIG. 12, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning training request includes a second identifier indicative of said network control aspect. According to further example embodiments, said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. FIG. 5 is a block diagram illustrating an apparatus according to example embodiments. The apparatus may be a network node or entity 50 such as a machine learning inference producer entity (or a network node or entity providing such functionality) comprising a generating circuitry 51 and a transmitting circuitry 52. The generating circuitry 51 generates a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect. The transmitting circuitry 52 transmits said machine learning inference capability report. FIG. 13 is a schematic diagram of a procedure according to example embodiments. The apparatus according to FIG. 5 may perform the method of FIG. 13 but is not limited to this method. The method of FIG. 13 may be performed by the apparatus of FIG. 5 but is not limited to being performed by this apparatus. As shown in FIG. 13, a procedure according to example embodiments comprises an operation of generating (S131) a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and an operation of transmitting (S132) said machine learning inference capability report. FIG. 6 is a block diagram illustrating an apparatus according to example embodiments. In particular, FIG. 6 illustrates a variation of the apparatus shown in FIG. 5. The apparatus according to FIG. 6 may thus further comprise a receiving circuitry 61. In an embodiment at least some of the functionalities of the apparatus shown in FIG. 5 (or 6) may be shared between two physically separate devices forming one operational entity. Therefore, the apparatus may be seen to depict the operational entity comprising one or more physically separate devices for executing at least some of the described processes. According to further example embodiments, said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect. According to further example embodiments, said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. According to a variation of the procedure shown in FIG. 13, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine learning inference capability report request. According to a variation of the procedure shown in FIG. 13, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine learning inference request for machine learning inference related to at least one sub-aspect of a network control aspect, and an operation of generating a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. According to a variation of the procedure shown in FIG. 13, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine learning inference configuration for machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning inference configuration is indicative of a machine learning inference timing, and an operation of generating, meeting said machine learning inference timing, a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Meeting said machine learning inference timing here means that (timing) requirements indicated by the machine learning inference timing are fulfilled. According to a variation of the procedure shown in FIG. 13, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of transmitting said machine learning inference report. According to further example embodiments, said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning inference request includes a second identifier indicative of said network control aspect. According to further example embodiments, said machine learning inference request is indicative of an inference time window. According to further example embodiments, said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. According to further example embodiments, said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. According to further example embodiments, said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. According to further example embodiments, said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one supported subaspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. FIG. 7 is a block diagram illustrating an apparatus according to example embodiments. The apparatus may be a network node or entity 70 such as a machine learning inference consumer entity (or a network node or entity providing such functionality) comprising a receiving circuitry 71 and a considering circuitry 72. The receiving circuitry 71 receives a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect. The considering circuitry 72 considers said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. FIG. 14 is a schematic diagram of a procedure according to example embodiments. The apparatus according to FIG. 7 may perform the method of FIG. 14 but is not limited to this method. The method of FIG. 14 may be performed by the apparatus of FIG. 7 but is not limited to being performed by this apparatus. As shown in FIG. 14, a procedure according to example embodiments comprises an operation of receiving (S141) a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and an operation of considering (S142) said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. FIG. 8 is a block diagram illustrating an apparatus according to example embodiments. In particular, FIG. 8 illustrates a variation of the apparatus shown in FIG. 7. The apparatus according to FIG. 8 may thus further comprise a transmitting circuitry 81 and / or a determining circuitry 82. In an embodiment at least some of the functionalities of the apparatus shown in FIG. 7 (or 8) may be shared between two physically separate devices forming one operational entity. Therefore, the apparatus may be seen to depict the operational entity comprising one or more physically separate devices for executing at least some of the described processes. According to further example embodiments, said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect. According to further example embodiments, said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. According to a variation of the procedure shown in FIG. 14, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of transmitting a machine learning inference capability report request. According to a variation of the procedure shown in FIG. 14, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of determining necessity for a machine leaning inference related to at least one sub-aspect of a network control aspect, and an operation of transmitting a machine learning inference request for machine learning inference related to said at least one sub-aspect of said network control aspect. According to a variation of the procedure shown in FIG. 14, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of determining necessity, at a machine learning inference timing, for a machine leaning inference related to at least one sub-aspect of a network control aspect, and an operation of transmitting a machine learning inference configuration for machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning inference configuration is indicative of said machine learning inference timing. Necessity, at a machine learning inference timing, here means that the necessity is assumed / predicted as being present at the machine learning inference timing. Determination of the necessity, thus, may be performed in advance before the machine learning inference timing. According to a variation of the procedure shown in FIG. 14, exemplary additional operations are given, which are inherently independent from each other as such. According to such variation, an exemplary method according to example embodiments may comprise an operation of receiving a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. According to further example embodiments, said machine learning inference request includes a second identifier indicative of said network control aspect. According to further example embodiments, said machine learning inference request is indicative of an inference time window. According to further example embodiments, said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. According to further example embodiments, said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. According to further example embodiments, said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. According to further example embodiments, said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization. According to further example embodiments, said at least one supported subaspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell. According to further example embodiments, said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Example embodiments outlined and specified above are explained below in more specific terms. FIG. 15 shows a schematic diagram of signaling sequences according to example embodiments, and in particular illustrates a detailed step-by-step workflow according to example embodiments. In a step 1 of FIG. 15, according to example embodiments, an AI / ML training consumer may request an AI / ML training producer to train an AI / ML entity for inferenceType, e.g., RAN energy saving, and inferenceOutputType, e.g., Beam on-off prediction and / or Cell on-off prediction and / or Candidate beam to handover the traffic before switching off the serving beam and / or Candidate cell to handover the traffic before switching off the serving cell and / or Cell energy cost prediction and / or gNodeB energy cost prediction and / or Cell load prediction and / or Energy efficiency prediction. The AI / ML training request information object class (IOC) may be specified as in the following table: Attribute name Support Qualifier is- Reada-ble is- Writa-ble isln-vari-ant is-Noti-fyable inferenceType CM T F F T inferenceOutputType CM T F F T candidateTrainingDataSource 0 T T F T trainingDataQualityScore 0 T T F T training Requestsource M T T F T requeststatus M T F F T expected Runtimecontext 0 T T F T performanceRequirements M T T F T cancel Request 0 T T F T suspendRequest 0 T T F T Attribute related to role mLEntityToTrainRef CM T F F T mLEntityCoordinationGroupT oTrainRef CM T F F T The AI / ML training request IOC (attribute constraints) may be specified as in the following table: Name Definition inferenceType Support Qualifier Condition: MLTrainingRequest MOI represents the request for initial ML training. inferenceOutputType Support Qualifier Condition: MLTrainingRequest MOI represents the request for initial ML training. mLEntityToTrainRef Support Qualifier Condition: MLTrainingRequest MOI represents the request for ML retraining. mLEntityCoordinationGroupToTrainRef Support Qualifier Condition: MLTrainingRequest MOI represents the request for joint training of a group of ML entities. Attribute definitions may be specified as in the following table: Attribute Name Documentation and Allowed Values Properties inferenceOutputType It indicates the type of inference output. allowedValues: BEAMONOFF, CELLONOFF, CAN DI DATEBEAMTOHAN DOVER, CAN DI DATECELLTOHAN DOVER, CELLENERGYCOSTPREDICTION, GNODEBENERGYCOSTPREDICTION, CELLLOADPREDICTION, ENERGYEFFICIENCYPREDICTION, type: Enum multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: True 5 In a step 2 of FIG. 15, according to example embodiments, the AI / ML training producer may train the AI / ML entity for a particular inference type, e.g., RAN energy saving, and inference output type (e.g., BEAMONOFF, CELLONOFF) using the available training data to satisfy the training performance requirements. The AI / ML training producer may use one or multiple AI / ML models to generate multiple inference outputs which is implementationspecific. 5 The AI / ML entity IOC may be specified as in the following table: Attribute name Support Qualifier is- Read-able is- Writ-able isln-vari-ant is-Noti-fyable mLEntityld M T F F T inferenceType M T F F T inferenceOutputType M T F F T mLEntityVersion M T F F T expected RunTimeContext 0 T T F T trainingcontext CM T F F T runTimeContext 0 T F F T supported Performanceindicators 0 T F F T Attribute related to role retrainingEventsMonitorRef 0 T T F T sourceTrainedMLEntityRef CM T F F T In a step 3 of FIG. 15, according to example embodiments, the AI / ML training 10 producer evaluates the performance of the AI / ML entity for each inference type, e.g., RAN energy saving, and for each inference output type (e.g., BEAMONOFF, CELLONOFF) and provides one AI / ML training report per inference output type. Since the AI / ML entity can provide multiple outputs using one or more AI / ML models, the performance of each inference output 15 type may vary depending on the features used to train the respective AI / ML models for each inference output type. The AI / ML training report IOC may be specified as in the following table: Attribute name Support Qualifier is- Read-able Is- Writ-able isln-vari-ant is-Noti-fyable mLEntityld M T F F T inferenceType M T F F T inferenceOutputType M T F F T areConsumerTrainingDataUsed M T F F T usedConsumerTrainingData CM T F F T confidenceindication 0 T F F T model PerformanceTraining M T F F T model Performancevalidation 0 T F F T dataRatioTrainingAndValidation 0 T F F T areNewTrainingDataUsed M T F F T Attribute related to role trainingRequestRef CM T F F T trainingProcessRef M T F F T lastTrainingRef CM T F F T mLEnityGeneratedRef M T F F T mLEnityCoordinationGroupGene rated Ref CM T F F T In a step 4 of FIG. 15, according to example embodiments, a management service (MnS) consumer may request the AI / ML training producer to load the trained AI / ML entity to the AI / ML inference producer also indicating the 5 inference type and the inference output type that the AI / ML entity had to be trained for. Additionally, the MnS consumer may provide the policy for loading the AI / ML entity, e.g., AI / ML entity should meet certain performance requirements. 10 The AI / ML entity loading request IOC may be specified as in the following table: Attribute name Support Qualifier Is-Read-able Is- Writ-able isln-variant is-Noti-fyable inferenceType M T T F T inferenceOutputType M T T F T policyForLoading M T T F T requeststatus M T T F T cancel Request 0 T T F T Attribute related to role mLEntityToLoadRef M T F F T In a step 5 of FIG. 15, according to example embodiments, the AI / ML training producer checks if there is any AI / ML entity that satisfies the requirements (inference type, inference output type and policy for loading) received in step 4. If yes, the AI / ML training producer loads the AI / ML entity to the AI / ML inference producer and if not, (re)trains the AI / ML entity to satisfy those requirements. In a step 6 of FIG. 15, according to example embodiments, the supported capabilities are discovered by the AI / ML inference consumer. Step 6 of FIG. 15 may be implemented in two alternatives. In a first alternative, a single AI / ML inference producer hosts AI / ML entities supporting multiple AI / ML inference types (e.g., RAN energy saving, mobility load balancing, mobility robust optimization, etc.) and multiple AI / ML inference output types for each AI / ML inference type. In such case, the AI / ML inference function IOC may be specified as in the following table: Attribute name Support Qualifier is- Read-able is- Writ-able isln-variant is-Noti-fyable AIMLInferenceFunctionlD M T T F T mLEntityld M T T F T activationstatus M T T F T supportedAiMIInferenceType M T T F T supportedAiMIESInferenceOu tputType CM T T F T The AI / ML inference function IOC (attribute constraints) may be specified as in the following table: Name Definition supportedAiMIESInferenceOutputType Support Qualifier Condition: supportedAiMIInferenceType should be of type ES The attribute definitions may be specified as in the following table: Attribute Name Documentation and Allowed Values Properties supportedAiMIInferenceType It indicates the type of inference. type: inferenceType multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: True Attribute Name Documentation and Allowed Values Properties supportedAiMIESInferenceOutp utType It indicates the type of ES inference outputs. allowedValues: BEAMONOFF, CELLONOFF, CANDIDATEBEAMTOHA N DOVER, CANDIDATECELLTOHA N DOVER, CELLENERGYCOSTPRE DICTION, GNODEBENERGYCOSTP REDICTION, CELLLOADPREDICTION / ENERGYEFFICIENCYPR EDICTION. type: Enum multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: True activationstatus It describes the activation status. allowedValues: ACTIVATED, DEACTIVATED. Type: Enum multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: False Attribute Name Documentation and Allowed Values Properties inferenceTypeActivationStatus It describes the activation status of specific inferenceType. allowedValues: ACTIVATED, DEACTIVATED. Type: Enum multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: False In a second alternative, a dedicated AI / ML inference producer hosts AI / ML entities supporting a particular AI / ML inference type (e.g., RAN energy saving) only. Alternatively, dedicated AI / ML inference producer capabilities 5 may be derived from the common AI / ML inference producer capabilities. In such case, the (exemplary) AI / ML energy saving (ES) inference function IOC may be specified as in the following table: Attribute name Support Qualifier is-Read-able is- Writ-able isln-variant is-Noti-fyable AIMLESInferenceFunctionlD M T T F T activationstatus M T T F T supportedAiMIESInferenceO utputType M T T F T The attribute definitions may be specified as in the following table: Attribute Name Documentation and Allowed Values Properties supportedAiMIESInference OutputType It indicates the type of ES inference outputs. allowedValues: BEAMONOFF, CELLONOFF, CANDIDATEBEAMTOHANDOVER, CANDIDATECELLTOHANDOVER, CELLENERGYCOSTPREDICTION, GNODEBENERGYCOSTPREDICTIO N, CELLLOADPREDICTION, ENERGYEFFICIENCYPREDICTION. type: Enum multiplicity: 1 isOrdered: N / A isUnique: N / A defaultvalue: None isNullable: True Similar definitions would be specified for the other uses cases, e.g. load balancing (LB), mobility optimization (MO), which might include "supportedAiMlLBInferenceOutputType", 5 "supportedAiMIMOInferenceOutputType". In a step 7 of FIG. 15, according to example embodiments, once the AI / ML inference consumer discovers the supported AI / ML inference function capabilities or the supported AI / ML ES inference function capabilities, it may 10 request the AI / ML inference producer to generate and provide inferences for a particular inference type and inference output type. If the AI / ML inference request is for the common AI / ML inference function, inferenceType attribute is present, and if the AI / ML inference request is for the dedicated AI / ML inference function (e.g., RAN energy saving) inferenceType attribute is not 15 present. The AI / ML inference request IOC may be specified as in the following table: Attribute name Support Qualifier is- Read-able is- Writ-able isln-variant is-Noti-fyable mLInferenceRequestld M T F F F inferenceTimeWindow 0 T F F F inferenceType M T F F F inferenceOutputType M T F F F The attribute definitions may be specified as in the following table: Attribute Name Documentation and Allowed Values Properties inferenceTimeWindow It indicates the time window over which the generated AI / ML inferences for a particular inference type and inference output type from a common or dedicated AI / ML inference function should be provided. Type: TimeWindow multiplicity: 1 isOrdered: False isUnique: True defaultvalue: None isNullable: False 5 In a step 8 of FIG. 15, according to example embodiments, the AI / ML inference producer generates the required AI / ML inference outputs based on the requirements from step 7. In a step 9 of FIG. 15, according to example embodiments, the AI / ML 10 inference producer provides the report of generated AI / ML inference outputs for the indicated inference type and inference output type in step 7. The MLInferenceReport may include inherited attributes from Top IOC and may be specified as including the following attributes: 15 Attribute name Support Qualifier is- Read-able is-Writ-able isln-variant is-Noti-fyable m LI nferenceReportl D M T F F F mLEntityld M T F F F m Li nferenceOutcomes M T F F F model Performanceinference M T F F T The datatype "MLOutcome" represents properties of a single MLOutcome. The MLOutcome may be specified as including the following attributes: Attribute name Support Qualifier is- Read-able is- Writ-able isln-variant is-Noti-fyable mLOutcomelD M T F F F inferenceType M T F F F inferenceOutputType M T F F F variableName M T F F F variableType M T F F F mLOutcomeValue M T F F F Principles according to example embodiments may be specified as follows: Use cases, potential requirements and possible solutions AI / ML Performance Management 10 Description Use cases Training an ML entity to support a specific subset of multiple AI / ML inference outputs supported by the ML entity The ML entity (composed of one or more ML models) may be trained to provide multiple AI / ML inference outputs. For example, an ML entity for RAN energy saving may be trained to provide multiple AI / ML inference outputs, including among others (i) Beam on-off prediction, (ii) Cell on-off prediction, (iii) Candidate beam to handover the traffic before switching off the serving beam, (iv) Candidate cell to handover the traffic before switching off the serving cell, (v) Cell energy cost prediction, (vi) gNodeB energy cost prediction, (vii) Cell load prediction, (viii) Energy efficiency prediction, etc. Currently, the ML training consumer can only request the ML training producer to train an ML entity but cannot request the ML training producer to train an ML entity to support a specific subset of AI / ML inference outputs. For example, the MnS consumer cannot request the ML training producer to train an ML entity to support specific subset of AI / ML inference outputs for RAN energy saving. Similarly, there is no means for the AI / ML inference consumer to request AI / ML inference producer to generate and provide AI / ML inference reports with specific subset AI / ML inference outputs, e.g., for RAN energy saving. The AIML MnS training producer and AIML inference producer should support capabilities enabling the MnS consumers to request training or inference for a specific subset AI / ML inference outputs. Potential requirements REQ-AI / MLUPDATE-1 the AIML MnS training producer should have a capabilities enabling the AIML MnS consumers to request AIML MnS training producer to train an ML entity that can support only a subset of AI / ML inference outputs for a given use case. REQ-AI / MLUPDATE-1 the AIML MnS inference producer should have capabilities enabling the AIML MnS consumers to request AIML MnS inference producer to generate and provide AI / ML inference reports on only a subset of AI / ML inference outputs for a given AIML use case. Possible solutions Introduce an attribute for supported inference types on the ML entity or on the AIML MnS training and AIML inference producers. The attribute may be called supportedAiMIInferenceTypes. Introduce an attribute for the required inference types on the ML training request report. The attribute may be called inferenceOutputType. It may be a multiplicity to allow for multiple types to be requested in one request. Introduce an attribute for an inference type on the ML inference report. The attribute may be called supportedAiMIInferenceTypes. It may be a multiplicity to allow for multiple types to be reported at once. The above-described procedures and functions may be implemented by respective functional elements, processors, or the like, as described below. In the foregoing exemplary description of the network entity, only the units that are relevant for understanding the principles of the disclosure have been described using functional blocks. The network entity may comprise further units that are necessary for its respective operation. However, a description of these units is omitted in this specification. The arrangement of the functional blocks of the devices is not construed to limit the disclosure, and the functions may be performed by one block or further split into sub-blocks. When in the foregoing description it is stated that the apparatus, i.e. network node or entity (or some other means) is configured to perform some function, this is to be construed to be equivalent to a description stating that a (i.e. at least one) processor or corresponding circuitry, potentially in cooperation with computer program code stored in the memory of the respective apparatus, is configured to cause the apparatus to perform at least the thus mentioned function. Also, such function is to be construed to be equivalently implementable by specifically configured circuitry or means for performing the respective function (i.e. the expression "unit configured to" is construed to be equivalent to an expression such as "means for"). In FIG. 16, an alternative illustration of apparatuses according to example embodiments is depicted. As indicated in FIG. 16, according to example embodiments, the apparatus (network node or entity) 10' (corresponding to the network node or entity 10) comprises a processor 161, a memory 162 and an interface 163, which are connected by a bus 164 or the like. Further, according to example embodiments, the apparatus (network node or entity) 30' (corresponding to the network node or entity 30) comprises a processor 161, a memory 162 and an interface 163, which are connected by a bus 164 or the like. Further, according to example embodiments, the apparatus (network node or entity) 50' (corresponding to the network node or entity 50) comprises a processor 161, a memory 162 and an interface 163, which are connected by a bus 164 or the like. Further, according to example embodiments, the apparatus (network node or entity) 70' (corresponding to the network node or entity 70) comprises a processor 161, a memory 162 and an interface 163, which are connected by a bus 164 or the like. The apparatuses 10', 30', 50', 70' may be connected via link 165 with another apparatus (e.g. an interface of the another apparatus). The another apparatus may be any of apparatuses 10', 30', 50', 70'. The processor 161 and / or the interface 163 may also include a modem or the like to facilitate communication over a (hardwire or wireless) link, respectively. The interface 163 may Include a suitable transceiver coupled to one or more antennas or communication means for (hardwire or wireless) communications with the linked or connected device(s), respectively. The interface 163 is generally configured to communicate with at least one other apparatus, i.e. the interface thereof. The memory 162 may store respective programs assumed to include program instructions or computer program code that, when executed by the respective processor, enables the respective electronic device or apparatus to operate in accordance with the example embodiments. In general terms, the respective devices / apparatuses (and / or parts thereof) may represent means for performing respective operations and / or exhibiting respective functionalities, and / or the respective devices (and / or parts thereof) may have functions for performing respective operations and / or exhibiting respective functionalities. When in the subsequent description it is stated that the processor (or some other means) is configured to perform some function, this is to be construed to be equivalent to a description stating that at least one processor, potentially in cooperation with computer program code stored in the memory of the respective apparatus, is configured to cause the apparatus to perform at least the thus mentioned function. Also, such function is to be construed to be equivalently implementable by specifically configured means for performing the respective function (i.e. the expression "processor configured to [cause the apparatus to] perform xxx-ing" is construed to be equivalent to an expression such as "means for xxx-ing"). According to example embodiments, an apparatus representing the network node or entity 10 comprises at least one processor 161, at least one memory 162 including computer program code, and at least one interface 163 configured for communication with at least another apparatus. The processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform receiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect (thus the apparatus comprising corresponding means for receiving), and to perform training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect (thus the apparatus comprising corresponding means for training). Alternatively, the processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform receiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing (thus the apparatus comprising corresponding means for receiving), and to perform training, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect (thus the apparatus comprising corresponding means for training). According to example embodiments, an apparatus representing the network node or entity 30 comprises at least one processor 161, at least one memory 162 including computer program code, and at least one interface 163 configured for communication with at least another apparatus. The processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform determining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect (thus the apparatus comprising corresponding means for determining), and to perform transmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect (thus the apparatus comprising corresponding means for transmitting). Alternatively, the processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform determining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect (thus the apparatus comprising corresponding means for determining), and to perform transmitting a machine learning training configuration for training said machine leaning entity to support machine 50 learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing (thus the apparatus comprising corresponding means for transmitting). According to example embodiments, an apparatus representing the network node or entity 50 comprises at least one processor 161, at least one memory 162 including computer program code, and at least one interface 163 configured for communication with at least another apparatus. The processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect (thus the apparatus comprising corresponding means for generating), and to perform transmitting said machine learning inference capability report (thus the apparatus comprising corresponding means for transmitting). According to example embodiments, an apparatus representing the network node or entity 70 comprises at least one processor 161, at least one memory 162 including computer program code, and at least one interface 163 configured for communication with at least another apparatus. The processor (i.e. the at least one processor 161, with the at least one memory 162 and the computer program code) is configured to perform receiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect (thus the apparatus comprising corresponding means for receiving), and to perform considering said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision (thus the apparatus comprising corresponding means for considering). For further details regarding the operability / functionality of the individual apparatuses, reference is made to the above description in connection with any one of FIGs. 1 to 15, respectively. For the purpose of the present disclosure as described herein above, it should be noted that - method steps likely to be implemented as software code portions and being run using a processor at a network server or network entity (as examples of devices, apparatuses and / or modules thereof, or as examples of entities including apparatuses and / or modules therefore), are software code independent and can be specified using any known or future developed programming language as long as the functionality defined by the method steps is preserved; - generally, any method step is suitable to be implemented as software or by hardware without changing the idea of the embodiments and its modification in terms of the functionality implemented; - method steps and / or devices, units or means likely to be implemented as hardware components at the above-defined apparatuses, or any module(s) thereof, (e.g., devices carrying out the functions of the apparatuses according to the embodiments as described above) are hardware independent and can be implemented using any known or future developed hardware technology or any hybrids of these, such as MOS (Metal Oxide Semiconductor), CMOS (Complementary MOS), BiMOS (Bipolar MOS), BiCMOS (Bipolar CMOS), ECL (Emitter Coupled Logic), TTL (Transistor-Transistor Logic), etc., using for example ASIC (Application Specific IC (Integrated Circuit)) components, FPGA (Field-programmable Gate Arrays) components, CPLD (Complex Programmable Logic Device) components or DSP (Digital Signal Processor) components; - devices, units or means (e.g. the above-defined network entity or network register, or any one of their respective units / means) can be implemented as individual devices, units or means, but this does not exclude that they are implemented in a distributed fashion throughout the system, as long as the functionality of the device, unit or means is preserved; - an apparatus like the user equipment and the network entity / network register may be represented by a semiconductor chip, a chipset, or a (hardware) module comprising such chip or chipset; this, however, does not exclude the possibility that a functionality of an apparatus or module, instead of being hardware implemented, be implemented as software in a (software) module such as a computer program or a computer program product comprising executable software code portions for execution / being run on a processor; - a device may be regarded as an apparatus or as an assembly of more than one apparatus, whether functionally in cooperation with each other or functionally independently of each other but in a same device housing, for example. In general, it is to be noted that respective functional blocks or elements according to above-described aspects can be implemented by any known means, either in hardware and / or software, respectively, if it is only adapted to perform the described functions of the respective parts. The mentioned method steps can be realized in individual functional blocks or by individual devices, or one or more of the method steps can be realized in a single functional block or by a single device. Generally, any method step is suitable to be implemented as software or by hardware without changing the idea of the present disclosure. Devices and means can be implemented as individual devices, but this does not exclude that they are implemented in a distributed fashion throughout the system, as long as the functionality of the device is preserved. Such and similar principles are to be considered as known to a skilled person. Software in the sense of the present description comprises software code as such comprising code means or portions or a computer program or a computer program product for performing the respective functions, as well as software (or a computer program or a computer program product) embodied on a tangible medium such as a computer-readable (storage) medium having stored thereon a respective data structure or code means / portions or embodied in a signal or in a chip, potentially during processing thereof. The present disclosure also covers any conceivable combination of method steps and operations described above, and any conceivable combination of nodes, apparatuses, modules or elements described above, as long as the above-described concepts of methodology and structural arrangement are applicable. In view of the above, there are provided measures for control of machine learning training and inference. Such measures exemplarily comprise receiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one subaspect of a network control aspect, and training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Even though the disclosure is described above with reference to the examples according to the accompanying drawings, it is to be understood that the disclosure is not restricted thereto. Rather, it is apparent to those skilled in the art that the present disclosure can be modified in many ways without departing from the scope of the inventive idea as disclosed herein. The above disclosure covers at least the following Items: Item 1. A method, comprising receiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 2. A method, comprising receiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one subaspect of said network control aspect. Item 3. The method according to Item 1 or 2, further comprising generating a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect, and transmitting said machine learning training report. Item 4. The method according to Item 3, further comprising evaluating said inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. Item 5. The method according to any of Items 1 to 4, further comprising transmitting information on said machine leaning entity to a machine learning entity repository entity or to a machine learning inference producer entity. Item 6. The method according to Item 5, further comprising receiving a machine leaning entity transmission request for transmitting said machine leaning entity to said machine learning inference producer entity. Item 7. The method according to any of Items 1 to 6, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 8. The method according to any of Items 1 to 7, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 9. The method according to any of Items 1 to 8, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 10. A method, comprising determining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. Item 11. A method, comprising determining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. Item 12. The method according to Item 10 or 11, further comprising receiving a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect. Item 13. The method according to any of Items 10 to 12, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 14. The method according to any of Items 10 to 13, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 15. The method according to any of Items 10 to 14, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 16. A method, comprising generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and transmitting said machine learning inference capability report. Item 17. The method according to Item 16, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 18. The method according to Item 16 or 17, further comprising receiving a machine learning inference capability report request. Item 19. The method according to any of Items 16 to 18, further comprising receiving a machine learning inference request for machine learning inference related to at least one sub-aspect of a network control aspect, and generating a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 20. The method according to any of Items 16 to 18, further comprising receiving a machine learning inference configuration for machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning inference configuration is indicative of a machine learning inference timing, and generating, meeting said machine learning inference timing, a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 21. The method according to Item 19 or 20, further comprising transmitting said machine learning inference report. Item 22. The method according to any of Items 19 to 21, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 23. The method according to any of Items 19 to 22, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 24. The method according to any of Items 19 to 23, wherein said machine learning inference request is indicative of an inference time window. Item 25. The method according to any of Items 19 to 24, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 26. The method according to any of Items 16 to 25, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 27. The method according to any of Items 16 to 26, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 28. The method according to any of Items 16 to 27, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 29. A method, comprising receiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and considering said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. Item 30. The method according to Item 29, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 31. The method according to Item 29 or 30, further comprising transmitting a machine learning inference capability report request. Item 32. The method according to any of Items 29 to 31, further comprising determining necessity for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning inference request for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 33. The method according to any of Items 29 to 31, further comprising determining necessity, at a machine learning inference timing, for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning inference configuration for machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning inference configuration is indicative of said machine learning inference timing. Item 34. The method according to Item 32 or 33, further comprising receiving a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 35. The method according to any of Items 32 to 34, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 36. The method according to any of Items 32 to 35, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 37. The method according to any of Items 32 to 36, wherein said machine learning inference request is indicative of an inference time window. Item 38. The method according to any of Items 32 to 37, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 39. The method according to any of Items 29 to 38, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 40. The method according to any of Items 29 to 39, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 41. The method according to any of Items 29 to 40, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 42. An apparatus, comprising receiving circuitry configured to receive a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and training circuitry configured to train said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. receiving circuitry configured to receive a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training circuitry configured to train, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 44. The apparatus according to Item 42 or 43, further comprising generating circuitry configured to generate a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect, and transmitting circuitry configured to transmit said machine learning training report. Item 45. The apparatus according to Item 44, further comprising evaluating circuitry configured to evaluate said inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. Item 46. The apparatus according to any of Items 42 to 45, further comprising transmitting circuitry configured to transmit information on said machine leaning entity to a machine learning entity repository entity or to a machine learning inference producer entity. Item 47. The apparatus according to Item 46, further comprising receiving circuitry configured to receive a machine leaning entity transmission request for transmitting said machine leaning entity to said machine learning inference producer entity. Item 48. The apparatus according to any of Items 42 to 47, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 49. The apparatus according to any of Items 42 to 48, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 50. The apparatus according to any of Items 42 to 49, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 51. An apparatus, comprising determining circuitry configured to determine necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting circuitry configured to transmit a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. Item 52. An apparatus, comprising determining circuitry configured to determine necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control And transmitting circuitry configured to transmit a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. Item 53. The apparatus according to Item 51 or 52, further comprising receiving circuitry configured to receive a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. Item 54. The apparatus according to any of Items 51 to 53, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 55. The apparatus according to any of Items 51 to 54, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 56. The apparatus according to any of Items 51 to 55, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. generating circuitry configured to generate a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and transmitting circuitry configured to transmit said machine learning inference capability report. Item 58. The apparatus according to Item 57, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 59. The apparatus according to Item 57 or 58, further comprising receiving circuitry configured to receive a machine learning inference capability report request. Item 60. The apparatus according to any of Items 57 to 59, further comprising receiving circuitry configured to receive a machine learning inference request for machine learning inference related to at least one sub-aspect of a network control aspect, and generating circuitry configured to generate a machine learning inference report indicative of an inference result for said at least one subaspect of said network control aspect. Item 61. The apparatus according to any of Items 57 to 59, further comprising receiving circuitry configured to receive a machine learning inference configuration for machine learning inference related to at least one sub aspect of a network control aspect, wherein said machine learning inference configuration is indicative of a machine learning inference timing, and generating circuitry configured to generate, meeting said machine learning inference timing, a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 62. The apparatus according to Item 60 or 61, further comprising transmitting circuitry configured to transmit said machine learning inference report. Item 63. The apparatus according to any of Items 60 to 62, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 64. The apparatus according to any of Items 60 to 63, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 65. The apparatus according to any of Items 60 to 64, wherein said machine learning inference request is indicative of an inference time window. Item 66. The apparatus according to any of Items 61 to 65, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 67. The apparatus according to any of Items 57 to 66, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 68. The apparatus according to any of Items 57 to 67, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 69. The apparatus according to any of Items 57 to 68, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 70. An apparatus, comprising receiving circuitry configured to receive a machine learning inference capability report indicative of support of machine learning inference related to at least one supported sub-aspect of a supported network control aspect, and considering circuitry configured to consider said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. Item 71. The apparatus according to Item 70, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 72. The apparatus according to Item 70 or 71, further comprising transmitting circuitry configured to transmit a machine learning inference capability report request. Item 73. The apparatus according to any of Items 70 to 72, further comprising determining circuitry configured to determine necessity for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting circuitry configured to transmit a machine learning inference request for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 74. The apparatus according to any of Items 70 to 73, further comprising determining circuitry configured to determine necessity, at a machine learning inference timing, for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting circuitry configured to transmit a machine learning inference configuration for machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning inference configuration is indicative of said machine learning inference timing. Item 75. The apparatus according to Item 73 or 74, further comprising receiving circuitry configured to receive a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 76. The apparatus according to any of Items 73 to 75, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 77. The apparatus according to any of Items 73 to 76, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 78. The apparatus according to any of Items 73 to 77, wherein said machine learning inference request is indicative of an inference time window. Item 79. The apparatus according to any of Items 70 to 78, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 80. The apparatus according to any of Items 70 to 79, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 81. The apparatus according to any of Items 70 to 80, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 82. The apparatus according to any of Items 70 to 81, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 83. An apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, and training said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 84. An apparatus, comprising at least one processor, and at least one memory storing Instructions that, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, and training, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one subaspect of said network control aspect. Item 85. The apparatus according to Item 83 or 84, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: generating a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect, and transmitting said machine learning training report. Item 86. The apparatus according to Item 85, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: evaluating said inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect. Item 87. The apparatus according to any of Items 83 to 86, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: transmitting information on said machine leaning entity to a machine learning entity repository entity or to a machine learning inference producer entity. Item 88. The apparatus according to Item 87, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine leaning entity transmission request for transmitting said machine leaning entity to said machine learning inference producer entity. Item 89. The apparatus according to any of Items 83 to 88, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 90. The apparatus according to any of Items 83 to 89, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 91. The apparatus according to any of Items 83 to 90, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 92. An apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform: determining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, ano transmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect. Item 93. An apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform: determining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing. Item 94. The apparatus according to Item 92 or 93, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect. Item 95. The apparatus according to any of Items 92 to 94, wherein said machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 96. The apparatus according to any of Items 92 to 95, wherein said machine learning training request includes a second identifier indicative of said network control aspect. Item 97. The apparatus according to any of Items 92 to 96, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 98. An apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform: generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and transmitting said machine learning inference capability report. Item 99. The apparatus according to Item 98, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 100. The apparatus according to Item 98 or 99, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning inference capability report request. Item 101. The apparatus according to any of Items 98 to 100, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning inference request for machine learning inference related to at least one sub-aspect of a network control aspect, and generating a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 102. The apparatus according to any of Items 98 to 100, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning inference configuration for machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning inference configuration is indicative of a machine learning inference timing, and generating, meeting said machine learning inference timing, a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 103. The apparatus according to Item 101 or 102, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: transmitting said machine learning inference report. Item 104. The apparatus according to any of Items 101 to 103, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 105. The apparatus according to any of Items 101 to 104, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 106. The apparatus according to any of Items 101 to 105, wherein said machine learning inference request is indicative of an inference time window. Item 107. The apparatus according to any of Items 101 to 106, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 108. The apparatus according to any of Items 98 to 107, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 109. The apparatus according to any of Items 98 to 108, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 110. The apparatus according to any of Items 98 to 109, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 111. An apparatus, comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, and considering said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision. Item 112. The apparatus according to Item 111, wherein said machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / or said machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect. Item 113. The apparatus according to Item 111 or 112, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: transmitting a machine learning inference capability report request. Item 114. The apparatus according to any of Items 111 to 113, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: determining necessity for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning inference request for machine learning inference related to said at least one sub-aspect of said network control aspect. Item 115. The apparatus according to any of Items 111 to 113, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: determining necessity, at a machine learning inference timing, for a machine leaning inference related to at least one sub-aspect of a network control aspect, and transmitting a machine learning inference configuration for machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning inference configuration is indicative of said machine learning inference timing. Item 116. The apparatus according to Item 114 or 115, wherein the instructions, when executed by the at least one processor, cause the apparatus at least to perform: receiving a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect. Item 117. The apparatus according to any of Items 114 to 116, wherein said machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect. Item 118. The apparatus according to any of Items 114 to 117, wherein said machine learning inference request includes a second identifier indicative of said network control aspect. Item 119. The apparatus according to any of Items 114 to 118, wherein said machine learning inference request is indicative of an inference time window. Item 120. The apparatus according to any of Items 111 to 119, wherein said network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 121. The apparatus according to any of Items 111 to 120, wherein said machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect. Item 122. The apparatus according to any of Items 111 to 121, wherein said machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect. Item 123. The apparatus according to any of Items 111 to 122, wherein said supported network control aspect is one of the following: energy saving, or load balancing, or mobility optimization, and / or said at least one supported sub-aspect of said network control aspect is at least one of the following: beam on / off prediction, or cell on / off prediction, or handover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, or base station energy cost prediction, or cell load prediction, or energy efficiency prediction, or said at least one sub-aspect of said network control aspect is at least one of the following: target cell prediction, or source cell resource status prediction, or neighboring cell resource status prediction, or prediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following: terminal trajectory prediction, or terminal traffic prediction, or prediction of handover target cell, or prediction of handover candidate cells. Item 124. A computer program product comprising computer-executable computer program code which, when the program is run on a computer, is configured to cause the computer to carry out the method according to any one of Items 1 to 9, 10 to 15, 16 to 28, or 29 to 41. Item 125. The computer program product according to Item 124, wherein the computer program product comprises a computer-readable medium on which the computer-executable computer program code is stored, and / or wherein the program is directly loadable into an internal memory of the computer or a processor thereof. List of acronyms and abbreviations 3GPP Third Generation Partnership Project AI artificial intelligence AnLF analytics logical function ES energy saving IOC information object class LB load balancing MDA management data analytics ML machine learning MnS management service MO mobility optimization RAN radio access network UE user equipment

Claims

1. A method, comprisingreceiving a machine learning training request for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, andtraining said machine leaning entity for machine learning inference related to said at least one sub-aspect of said network control aspect.

2. A method, comprisingreceiving a machine learning training configuration for training a machine leaning entity to support machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning training configuration is indicative of a machine learning training timing, andtraining, meeting said machine learning training timing, said machine leaning entity for machine learning inference related to said at least one subaspect of said network control aspect.

3. The method according to claim 1 or 2, further comprisinggenerating a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect, andtransmitting said machine learning training report.

4. The method according to claim 3, further comprisingevaluating said inference performance of said machine leaning entity for one of said at least one sub-aspect of said network control aspect.

5. The method according to any of claims 1 to 4, further comprisingtransmitting information on said machine leaning entity to a machine learning entity repository entity or to a machine learning inference producer entity.

6. The method according to claim 5, further comprisingreceiving a machine leaning entity transmission request for transmitting said machine leaning entity to said machine learning inference producer entity.

7. The method according to any of claims 1 to 6, whereinsaid machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect, and / or whereinsaid machine learning training request includes a second identifier indicative of said network control aspect, and / or whereinsaid network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, orhandover candidate cell determination, orcell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, orsaid at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, orsaid at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, or prediction of handover candidate cells.

8. A method, comprisingdetermining necessity for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, andtransmitting a machine learning training request for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect.

9. A method, comprisingdetermining necessity, at a machine learning training timing, for a machine leaning entity supporting machine learning inference related to at least one sub-aspect of a network control aspect, andtransmitting a machine learning training configuration for training said machine leaning entity to support machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning training configuration is indicative of said machine learning training timing.

10. The method according to claim 8 or 9, further comprisingreceiving a machine learning training report indicative of an inference performance of said machine leaning entity for one of said at least one subaspect of said network control aspect.

11. The method according to any of claims 8 to 10, whereinsaid machine learning training request includes a first identifier indicative of said at least one sub-aspect of said network control aspect, and / or whereinsaid machine learning training request includes a second identifier indicative of said network control aspect, and / or whereinsaid network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, orhandover candidate cell determination, orcell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, orsaid at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, orsaid at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, orprediction of handover candidate cells.generating a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, andtransmitting said machine learning inference capability report.

13. The method according to claim 12, whereinsaid machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / orsaid machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect.

14. The method according to claim 12 or 13, further comprising receiving a machine learning inference capability report request.

15. The method according to any of claims 12 to 14, further comprising receiving a machine learning inference request for machine learning inference related to at least one sub-aspect of a network control aspect, and generating a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect, and optionallytransmitting said machine learning inference report.

16. The method according to any of claims 12 to 14, further comprising receiving a machine learning inference configuration for machine learning inference related to at least one sub-aspect of a network control aspect, wherein said machine learning inference configuration is indicative of a machine learning inference timing, andgenerating, meeting said machine learning inference timing, a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect, and optionallytransmitting said machine learning inference report.

17. The method according to any of claims 15 to 16, whereinsaid machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect, and / or whereinsaid machine learning inference request includes a second identifier indicative of said network control aspect, and / or whereinsaid machine learning inference request is indicative of an inference time window, and / or whereinsaid network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, orhandover candidate cell determination, orcell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, orsaid at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, orsaid at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, orprediction of handover candidate cells.

18. The method according to any of claims 12 to 17, whereinsaid machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect, and / or whereinsaid machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect, and / or whereinsaid supported network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one supported sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, orhandover candidate cell determination, orcell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, orsaid at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, orsaid at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, orprediction of handover candidate cells.

19. A method, comprisingreceiving a machine learning inference capability report indicative of support of machine learning inference related to at least one supported subaspect of a supported network control aspect, andconsidering said support of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect for machine learning inference request decision.

20. The method according to claim 19, whereinsaid machine learning inference capability report is indicative of an activation status of machine learning inference related to said supported network control aspect, and / orsaid machine learning inference capability report is indicative of an activation status of machine learning inference related to said at least one supported sub-aspect of said supported network control aspect.

21. The method according to claim 19 or 20, further comprisingtransmitting a machine learning inference capability report request.

22. The method according to any of claims 19 to 21, further comprisingdetermining necessity for a machine leaning inference related to at least one sub-aspect of a network control aspect, andtransmitting a machine learning inference request for machine learning inference related to said at least one sub-aspect of said network control aspect, and optionallyreceiving a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect.

23. The method according to any of claims 19 to 21, further comprisingdetermining necessity, at a machine learning inference timing, for a machine leaning inference related to at least one sub-aspect of a network control aspect, andtransmitting a machine learning inference configuration for machine learning inference related to said at least one sub-aspect of said network control aspect, wherein said machine learning inference configuration is indicative of said machine learning inference timing, and optionallyreceiving a machine learning inference report indicative of an inference result for said at least one sub-aspect of said network control aspect.

24. The method according to any of claims 22 to 23, whereinsaid machine learning inference request includes a first identifier indicative of said at least one sub-aspect of said network control aspect, and / or whereinsaid machine learning inference request includes a second identifier indicative of said network control aspect, and / or whereinsaid machine learning inference request is indicative of an inference time window, and / or whereinsaid network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, or handover candidate cell determination, or cell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, orsaid at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, orprediction of handover candidate cells.

25. The method according to any of claims 19 to 24, whereinsaid machine learning inference capability report includes a third identifier indicative of said at least one supported sub-aspect of said supported network control aspect, and / or whereinsaid machine learning inference capability report includes a fourth identifier indicative of said supported network control aspect, and / or wherein said supported network control aspect is one of the following:energy saving, orload balancing, ormobility optimization, and / orsaid at least one supported sub-aspect of said network control aspect is at least one of the following:beam on / off prediction, orcell on / off prediction, orhandover candidate beam determination, orhandover candidate cell determination, orcell energy cost prediction, orbase station energy cost prediction, orcell load prediction, orenergy efficiency prediction, or10said at least one sub-aspect of said network control aspect is at least one of the following:target cell prediction, orsource cell resource status prediction, orneighboring cell resource status prediction, orprediction of terminals to be handed over to a target cell, or said at least one sub-aspect of said network control aspect is at least one of the following:terminal trajectory prediction, orterminal traffic prediction, orprediction of handover target cell, orprediction of handover candidate cells.

Images