Federated learning based ubiquitous computing power providing method and related device

Abstract

The application discloses a ubiquitous computing power providing method based on federated learning and a related device thereof. The method comprises the following steps: when a request of a client ubiquitous computing power resource is detected, acquiring request information of the request; inputting the request information into a target ubiquitous computing power federated learning model, processing the request information based on the target ubiquitous computing power federated learning model, obtaining a target correlation degree between the client and different preset ubiquitous computing power resources, and the target ubiquitous computing power federated learning model is obtained by performing federated learning on a preset basic model based on target training data with a preset correlation degree label; determining a ubiquitous computing power resource allocation strategy suitable for the client based on the target correlation degree; and selecting target ubiquitous computing power resources suitable for the client based on the ubiquitous computing power resource allocation strategy, so that the client uses the target ubiquitous computing power resources to complete a pre-designed computing task. In the application, the required ubiquitous computing power resources are accurately provided for the client.

Description

Technical Field

[0001] This application relates to the field of communication computers, and in particular to a method, apparatus, device and storage medium for providing ubiquitous computing power based on federated learning. Background Technology

[0002] Edge computing provides users with basic computing resources such as computing and storage close to the data source or user. Compared to the basic topology of the traditional Internet architecture, i.e., the end-to-end model, which is used by centralized computing, edge computing is gradually becoming a way of providing ubiquitous computing resources. Specifically, edge computing has changed from the basic topology of the existing end-to-end model to a network topology that dynamically forms ubiquitous computing resources from the perspective of completing edge computing tasks (considering factors such as spatial distance and network transmission latency). This has led to computing gradually shifting from the center to the edge, becoming more lightweight, dynamic, serverless, and function-based.

[0003] However, in the process of providing ubiquitous computing resources to clients, the existing edge computing involves the computing platform layer abstracting computing resources through computing power modeling to form computing power capability templates. Based on these templates, ubiquitous computing resources are provided to clients. However, these templates are not specific to the client's needs and are prone to leaking client privacy. This makes it difficult to accurately and securely provide clients with the ubiquitous computing resources they need. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method, apparatus, device and storage medium for providing ubiquitous computing power based on federated learning, aiming to solve the technical problem in the prior art that it is difficult to accurately and securely provide the ubiquitous computing power resources needed by clients.

[0005] This application provides a method for providing ubiquitous computing power based on federated learning, the method comprising:

[0006] When a request for ubiquitous computing resources from a client is detected, the request information of the request is obtained;

[0007] The request information is input into the target ubiquitous computing power federated learning model, and the request information is processed based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources.

[0008] The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels.

[0009] Based on the target correlation, a ubiquitous computing resource allocation strategy adapted to the client is determined;

[0010] Based on the ubiquitous computing resource allocation strategy, a target ubiquitous computing resource is selected to suit the client, so that the client can use the target ubiquitous computing resource to complete a preset computing task.

[0011] In one possible implementation of this application, before the step of processing the request information based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources, the method includes:

[0012] The target training data with preset correlation labels of the first ubiquitous computing power provider is determined, and the second training data of the second ubiquitous computing power provider is determined. The second ubiquitous computing power provider is one or more, and the users in the training data of the first ubiquitous computing power provider and the second ubiquitous computing power provider are different but the training features are the same.

[0013] Based on the target training data and the second training data, the preset base model is subjected to encrypted horizontal federated iterative training until a target model that meets the preset training completion conditions is obtained.

[0014] The target model is used as the target ubiquitous computing power federated learning model.

[0015] In one possible implementation of this application, the step of inputting the request information into the target ubiquitous computing power federated learning model and processing the request information based on the target ubiquitous computing power federated learning model includes:

[0016] Determine the request type of the request information, wherein the request type includes one or more of the following: calculation request type, network request type, and service request type;

[0017] Based on the request type, the target ubiquitous computing power federated learning model is selected from a preset set of ubiquitous computing power federated learning models;

[0018] The request information is input into the target ubiquitous computing power federated learning model, and the request information is processed based on the target ubiquitous computing power federated learning model.

[0019] In one possible implementation of this application, after the step of selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task, the method includes:

[0020] Based on the target correlation, the client's preference for each ubiquitous computing resource is determined;

[0021] Based on the degree of preference, the cache weight of each ubiquitous computing resource in the first preset cache resource pool of the client is determined;

[0022] Based on the cache weight, the first ubiquitous computing resource that needs to be reused by the first preset cache resource pool is determined.

[0023] In one possible implementation of this application, after the step of determining the client's preference for each ubiquitous computing resource based on the target correlation, the method further includes:

[0024] Based on the preferences of each client, predict the frequency and weight of the reuse of each ubiquitous computing resource;

[0025] Based on the frequency and weight of repeated use, the shared weight of each ubiquitous computing resource in the second preset cache resource pool is determined.

[0026] Based on the shared weight, the second ubiquitous computing resources that the second preset cache resource pool needs to share are determined.

[0027] In one possible implementation of this application, after the step of determining the first ubiquitous computing resource that needs to be reused by the first preset cache resource pool based on the cache weight, the method includes:

[0028] If the client's ubiquitous computing power request is detected again, ubiquitous computing power resources are selected from the first ubiquitous computing power resources that need to be used again, so that the client can complete the preset computing task again based on the selected ubiquitous computing power resources.

[0029] In one possible implementation of this application, the request includes at least one of the following: video traffic ubiquitous computing resource request, plaza traffic ubiquitous computing resource request, shopping mall traffic ubiquitous computing resource request, and game traffic ubiquitous computing resource request.

[0030] This application also provides a ubiquitous computing power provision device based on federated learning, the device comprising:

[0031] The acquisition module is used to acquire request information from clients requesting ubiquitous computing resources;

[0032] The input module is used to input the request information into the target ubiquitous computing power federated learning model, process the request information based on the target ubiquitous computing power federated learning model, and obtain the target correlation degree between the client and different preset ubiquitous computing power resources.

[0033] The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels.

[0034] The first module is used to determine a ubiquitous computing resource allocation strategy that is suitable for the client based on the target correlation degree.

[0035] The selection module is used to select a target ubiquitous computing resource that is compatible with the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task.

[0036] This application also provides a ubiquitous computing power providing device based on federated learning. The ubiquitous computing power providing device based on federated learning is a physical node device. The ubiquitous computing power providing device based on federated learning includes: a memory, a processor, and a program of the ubiquitous computing power providing method based on federated learning stored in the memory and executable on the processor. When the program of the ubiquitous computing power providing method based on federated learning is executed by the processor, it can implement the steps of the ubiquitous computing power providing method based on federated learning as described above.

[0037] To achieve the above objectives, a storage medium is also provided, on which a ubiquitous computing power provider based on federated learning is stored, wherein when the ubiquitous computing power provider based on federated learning is executed by a processor, the ubiquitous computing power provider based on federated learning implements the steps of any of the ubiquitous computing power provision methods based on federated learning described above.

[0038] This application provides a method, apparatus, device, and storage medium for providing ubiquitous computing power based on federated learning. Compared with the current method where the computing power platform layer only provides the required ubiquitous computing power resources to the client through ubiquitous computing power matching templates, which makes it difficult to accurately provide the required ubiquitous computing power resources to the client, this application, upon detecting a client's request for ubiquitous computing power resources, obtains the request information of the request; inputs the request information into a target ubiquitous computing power federated learning model; processes the request information based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources; wherein, the target ubiquitous computing power federated learning model is obtained by federated learning of a preset base model based on target training data with preset correlation degree labels; based on the target correlation degree, a ubiquitous computing power resource allocation strategy adapted to the client is determined; based on the ubiquitous computing power resource allocation strategy, a target ubiquitous computing power resource adapted to the client is selected so that the client can use the target ubiquitous computing power resource to complete a preset computing task. It is understood that in this application, when a client's request for ubiquitous computing resources is detected, the target correlation degree between the client and different preset ubiquitous computing resources can be accurately obtained based on the target ubiquitous computing federated learning model. That is, based on the target ubiquitous computing federated learning model, ubiquitous computing resources that match the client's needs can be discovered. Thus, the ubiquitous computing resource allocation strategy that adapts to the client's needs can be accurately determined, thereby accurately providing the client with the ubiquitous computing resources it needs. In addition, the target ubiquitous computing federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels. Due to the characteristics of federated learning, client privacy will not be leaked. Therefore, this application achieves accurate and secure provision of the client with the ubiquitous computing resources it needs. Attached Figure Description

[0039] Figure 1 A flowchart illustrating the first embodiment of the method for providing ubiquitous computing power based on federated learning in this application;

[0040] Figure 2 A flowchart illustrating steps S01-S03 in one embodiment of the method for providing ubiquitous computing power based on federated learning in this application.

[0041] Figure 3 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of this application;

[0042] Figure 4 This is a schematic diagram of the first scenario involved in the embodiments of this application;

[0043] Figure 5 This is a schematic diagram of the second scenario involved in the embodiments of this application. Detailed Implementation

[0044] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0045] This application provides a method for providing ubiquitous computing power based on federated learning. In the first embodiment of this method, referring to... Figure 1 The method for providing ubiquitous computing power based on federated learning includes:

[0046] Step S10: When a request for ubiquitous computing resources from a client is detected, obtain the request information of the request;

[0047] Step S20: Input the request information into the target ubiquitous computing power federated learning model, process the request information based on the target ubiquitous computing power federated learning model, and obtain the target correlation degree between the client and different preset ubiquitous computing power resources;

[0048] The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels.

[0049] Step S30: Based on the target correlation, determine the ubiquitous computing resource allocation strategy that is suitable for the client;

[0050] Step S40: Based on the ubiquitous computing power resource allocation strategy, select a target ubiquitous computing power resource that is compatible with the client, so that the client can use the target ubiquitous computing power resource to complete a preset computing task.

[0051] In this embodiment, the ubiquitous computing power provision method based on federated learning is applied to a ubiquitous computing power provision device based on federated learning, which is applied to a ubiquitous computing power provision equipment based on federated learning. The ubiquitous computing power provision equipment based on federated learning is applied to a ubiquitous computing power provision system based on federated learning. The ubiquitous computing power provision system based on federated learning belongs to a ubiquitous computing power provision platform based on federated learning, which can specifically be a computing power platform such as a computing power computer.

[0052] In this embodiment, the ubiquitous computing power federated learning model is used to determine the ubiquitous computing power resources required by the client's current business needs, thereby accurately providing the client with the necessary ubiquitous computing power resources.

[0053] In this embodiment, the ubiquitous computing power federated learning model is used to mine the ubiquitous computing power resources required by user terminals with different business needs, and then recommends ubiquitous computing power resources that the client is interested in in a targeted manner.

[0054] In this embodiment, the target ubiquitous computing power federated learning model avoids the repeated creation and destruction of ubiquitous computing power resources, thus avoiding excessive overhead and network bandwidth consumption.

[0055] In this embodiment, updating the target ubiquitous computing federated learning model helps optimize the ubiquitous computing resources of the client.

[0056] In this embodiment, the client can be a client of an enterprise, a company, an organization, etc., which has a need for ubiquitous computing resources.

[0057] The specific steps are as follows:

[0058] Step S10: When a request for ubiquitous computing resources from a client is detected, obtain the request information of the request;

[0059] In this embodiment, the ubiquitous computing power provision platform based on federated learning (hereinafter referred to as the platform) has different interfaces. On the corresponding application interface, the client can apply for ubiquitous computing power resources. Then, when the platform detects the client's request for ubiquitous computing power resources, it obtains the request information of the request, which includes ubiquitous computing power resource requests for video traffic, ubiquitous computing power resources for public squares, ubiquitous computing power resources for shopping malls, and ubiquitous computing power resources for games, etc.

[0060] In this embodiment, the request information may include information such as calculation type, calculation time, calculation amount, request location, and calculation object.

[0061] Step S20: Input the request information into the target ubiquitous computing power federated learning model, process the request information based on the target ubiquitous computing power federated learning model, and obtain the target correlation degree between the client and different preset ubiquitous computing power resources;

[0062] The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels.

[0063] In this embodiment, it should be noted that the target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels, and is a model that can accurately determine the target correlation between the client and different preset ubiquitous computing power resources.

[0064] Among them, reference Figure 2 Before the step of processing the request information based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources, the method includes:

[0065] Step S01: Determine the target training data with preset correlation labels from the first ubiquitous computing power provider, and determine the second training data from the second ubiquitous computing power provider. The second ubiquitous computing power provider may be one or more, and the users in the training data of the first ubiquitous computing power provider and the second ubiquitous computing power provider are different but the training features are the same.

[0066] Step S02: Based on the target training data and the second training data, perform encrypted horizontal federated iterative training on the preset basic model until a target model that meets the preset training completion conditions is obtained.

[0067] Step S03: Use the target model as the target ubiquitous computing power federated learning model.

[0068] This embodiment illustrates how to obtain the target ubiquitous computing power federated learning model.

[0069] This embodiment first describes privacy computing, which refers to a technology and system in which two or more participants (ubiquitous computing power providers, including a first and a second ubiquitous computing power provider in this embodiment) jointly perform computations. These participants collaborate to perform joint machine learning and joint analysis on their data without disclosing their own data. The participants in privacy computing can be different departments within the same organization or different organizations. Under the privacy computing framework, the participants' data remains in plaintext on their local machines, enabling cross-domain collaboration of multi-source data while protecting data security, thus solving the challenges of data protection and integrated application.

[0070] Currently, privacy computing mainly includes federated learning computing. Federated learning refers to a method of machine learning by collaborating different participants. The parameter update stage of federated learning is divided into two steps: (a) each participant uses only the data it owns to train the machine learning model and sends the model parameter update to a central coordinator (in this embodiment, the federated training modeling pipeline); (b) the coordinator merges the received model updates from different participants (e.g., takes the average) and redistributes the merged model parameter updates to each participant. In federated learning, participants do not need to expose their data to other participants or the coordinator. Therefore, federated learning can effectively protect user privacy and ensure data security.

[0071] Federated learning includes vertical federated learning computation and horizontal federated learning computation. Vertical federated learning, also known as sample-aligned federated learning, is suitable for situations where the training sample IDs of the participating parties overlap significantly, while the data features overlap less. It requires multiple parties to collaborate to complete the training and optimization of the model within a secure and confidential framework.

[0072] Horizontal Federated Learning is suitable for situations where there is a lot of overlap in the features of the participants' training sample data, but there are few user IDs. It requires multiple parties to collaborate to complete the training and optimization of the model within a secure and confidential framework.

[0073] In this embodiment, the target ubiquitous computing power federated learning model is obtained through horizontal federated learning training.

[0074] Specifically, there can be one primary provider of ubiquitous computing power, namely ubiquitous computing power provider A, and one secondary provider of ubiquitous computing power, namely ubiquitous computing power provider B, such as... Figure 4 As shown.

[0075] In this embodiment, there can be one first provider of ubiquitous computing power, namely ubiquitous computing power provider A, and multiple second providers of ubiquitous computing power, namely ubiquitous computing power provider B and ubiquitous computing power provider C. For example... Figure 5 As shown.

[0076] In this embodiment, the users of different ubiquitous computing power providers are different, but their training features are consistent or similar. The training data of different ubiquitous computing power providers are processed by encrypted sample data alignment and other methods.

[0077] Specifically, based on the target training data and the second training data, the iterative process of performing encrypted horizontal federated iterative training on the preset base model until a target model that meets the preset training completion conditions can be as follows:

[0078] like Figure 5 As shown, 1. Each ubiquitous computing power provider sends encrypted sample data (target training data with preset correlation labels from the first ubiquitous computing power provider, and second training data from the second ubiquitous computing power provider) to the platform's federated learning modeling pipeline service.

[0079] 2. The platform's federated learning modeling pipeline service aggregates data from various ubiquitous computing power providers for federated learning training, encrypts and updates model parameters;

[0080] 3. The Federated Learning Modeling Pipeline service provides updated models to various ubiquitous computing power providers;

[0081] 4. Each ubiquitous computing power provider updates its own model.

[0082] Iterate through steps 1-5 until a target model that meets the preset training completion conditions is obtained.

[0083] The above iteration process involves aligning encrypted sample data, performing encrypted federated training, and deriving a target ubiquitous computing power federated learning model that can be deployed and applied. This target ubiquitous computing power federated learning model is continuously updated.

[0084] Specifically, in this embodiment, if the request information is a request for ubiquitous computing power resources for video traffic, the target ubiquitous computing power federated learning model is a video-related ubiquitous computing power federated learning model; if the request information is a request for ubiquitous computing power resources for shopping mall traffic, the target ubiquitous computing power federated learning model is a shopping mall-related ubiquitous computing power federated learning model; if the request information is a request for ubiquitous computing power resources for square traffic, the target ubiquitous computing power federated learning model is a square-related ubiquitous computing power federated learning model; and if the request information is a request for ubiquitous computing power resources for game traffic, the target ubiquitous computing power federated learning model is a game-related ubiquitous computing power federated learning model.

[0085] In this embodiment, different preset ubiquitous computing resources can be storage ubiquitous computing resources, computing ubiquitous computing resources, network ubiquitous computing resources, business process ubiquitous computing resources, etc.

[0086] This includes ubiquitous computing resources for storage, ubiquitous computing resources for computing, ubiquitous computing resources for networks, and ubiquitous computing resources for business processes, which can be further subdivided.

[0087] In this embodiment, based on the request information, it is determined which type of ubiquitous computing resources the client needs (storage ubiquitous computing resources, computing ubiquitous computing resources, network ubiquitous computing resources, business process ubiquitous computing resources, etc.). Specifically, it can be further subdivided to determine more specific ubiquitous computing resources for the client.

[0088] If the target correlation is: storage ubiquitous computing resources: 0.8;

[0089] Ubiquitous computing power resources: 0.3;

[0090] Network ubiquitous computing resources: 0.1 etc.

[0091] The client then needs ubiquitous computing resources for storage.

[0092] Step S40: Based on the target correlation, determine the ubiquitous computing resource allocation strategy adapted to the client.

[0093] After determining the target correlation, a ubiquitous computing resource allocation strategy adapted to the client is determined. Specifically, if the client has a higher correlation with the storage ubiquitous computing resources, the ubiquitous computing resource allocation strategy is to determine the resources to be allocated to the client from the preset storage ubiquitous computing resources. The specific allocation strategy is related to the preset more detailed strategy determination rules.

[0094] Step S50: Based on the ubiquitous computing power resource allocation strategy, select a target ubiquitous computing power resource that is compatible with the client, so that the client can use the target ubiquitous computing power resource to complete a preset computing task.

[0095] In this embodiment, after determining the ubiquitous computing power resource allocation strategy, a target ubiquitous computing power resource that is compatible with the client is selected based on the computing power resource allocation strategy, so that the client can use the target ubiquitous computing power resource to complete the preset computing task.

[0096] The preset calculation task can be a video stream calculation task or a game traffic calculation task, etc., and there is no specific limitation.

[0097] This application provides a method, apparatus, device, and storage medium for providing ubiquitous computing power based on federated learning. Compared with the current method where the computing power platform layer only provides the required ubiquitous computing power resources to the client through ubiquitous computing power matching templates, which makes it difficult to accurately provide the required ubiquitous computing power resources to the client, this application, upon detecting a client's request for ubiquitous computing power resources, obtains the request information of the request; inputs the request information into a target ubiquitous computing power federated learning model; processes the request information based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources; wherein, the target ubiquitous computing power federated learning model is obtained by federated learning of a preset base model based on target training data with preset correlation degree labels; based on the target correlation degree, a ubiquitous computing power resource allocation strategy adapted to the client is determined; based on the ubiquitous computing power resource allocation strategy, a target ubiquitous computing power resource adapted to the client is selected so that the client can use the target ubiquitous computing power resource to complete a preset computing task. It is understood that in this application, when a client's request for ubiquitous computing resources is detected, the target correlation degree between the client and different preset ubiquitous computing resources can be accurately obtained based on the target ubiquitous computing federated learning model. That is, based on the target ubiquitous computing federated learning model, ubiquitous computing resources that match the client's needs can be discovered. Thus, the ubiquitous computing resource allocation strategy that adapts to the client's needs can be accurately determined, thereby accurately providing the client with the ubiquitous computing resources it needs. In addition, the target ubiquitous computing federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels. Due to the characteristics of federated learning, client privacy will not be leaked. Therefore, this application achieves accurate and secure provision of the client with the ubiquitous computing resources it needs.

[0098] Furthermore, based on the first embodiment of this application, another embodiment of this application is provided. In this embodiment, the step of inputting the request information into the target ubiquitous computing power federated learning model and processing the request information based on the target ubiquitous computing power federated learning model includes:

[0099] Step M1: Determine the request type of the request information, wherein the request type includes one or more of the following: calculation request type, network request type, and service request type;

[0100] In this embodiment, the request type of the request information is determined. The request type includes a calculation request type, a network request type, and a service request type. In addition, the request type can be combined with specific request information. For example, the request type can be a video stream calculation request type, a video stream network request type, and a video stream service request type, etc.

[0101] Step M2: Based on the request type, select the target ubiquitous computing power federated learning model from the preset ubiquitous computing power federated learning model set;

[0102] Step M3: Input the request information into the target ubiquitous computing power federated learning model, and process the request information based on the target ubiquitous computing power federated learning model.

[0103] In this embodiment, different request types have targeted ubiquitous computing power federated learning models. That is, in this embodiment, multiple ubiquitous computing power federated learning models are trained, and these multiple ubiquitous computing power federated learning models constitute a preset ubiquitous computing power federated learning model set.

[0104] In this embodiment, based on the request type, a target ubiquitous computing power federated learning model is selected from a preset set of ubiquitous computing power federated learning models. The request information is then input into the target ubiquitous computing power federated learning model, and the request information is processed based on the target ubiquitous computing power federated learning model. This embodiment improves the targeting of the processing.

[0105] Furthermore, based on the first embodiment of this application, another embodiment of this application is provided. In this embodiment, after the step of selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task, the method includes:

[0106] Step N1: Based on the target correlation, determine the client's preference for each ubiquitous computing resource;

[0107] Step N2: Based on the preference level, determine the cache weight of each ubiquitous computing resource in the first preset cache resource pool of the client;

[0108] Step N3: Based on the cache weight, determine the first ubiquitous computing resource that the first preset cache resource pool needs to use again.

[0109] In this embodiment, the platform's federated learning modeling pipeline service distributes ubiquitous computing resource pool services to each client. The ubiquitous computing resource pool service updates and optimizes the resource pool resources for the client, and in particular, updates and optimizes the cached resources of the resource pool for the client.

[0110] In this embodiment, the resource pool includes a first preset cache resource pool, which corresponds to each client.

[0111] In this embodiment, based on the target correlation degree, the client's preference for each ubiquitous computing resource is determined. Based on the preference degree, the cache weight of each ubiquitous computing resource in the client's first preset cache resource pool is determined. A higher preference degree results in a higher cache weight. There is a preset mapping relationship between the preference degree and the cache weight, which can be set.

[0112] Based on the cache weight, the first ubiquitous computing power resource that needs to be used again in the first preset cache resource pool is determined, thereby avoiding repeated destruction and creation of the first ubiquitous computing power resource and avoiding resource consumption.

[0113] After the step of determining the first ubiquitous computing resource that needs to be reused by the first preset cache resource pool based on the cache weight, the method includes:

[0114] Step H1: If the client's ubiquitous computing power request is detected again, ubiquitous computing power resources are selected from the first ubiquitous computing power resources that need to be used again, so that the client can complete the preset computing task again based on the selected ubiquitous computing power resources.

[0115] In this embodiment, if the client's ubiquitous computing power request is detected again, ubiquitous computing power resources are selected from the first ubiquitous computing power resources that need to be used again. Therefore, it avoids rebuilding the ubiquitous computing power resources required by the client in a short period of time and avoids excessive consumption of material resources.

[0116] After the step of determining the client's preference for each ubiquitous computing resource based on the target correlation, the following is included:

[0117] Step G1: Based on the preference levels of each client, predict the frequency and weight of the reuse of each ubiquitous computing resource;

[0118] Step G2: Based on the frequency and weight of repeated use, determine the shared weight of each ubiquitous computing resource in the second preset cache resource pool;

[0119] Step G3: Based on the shared weight, determine the second ubiquitous computing resources that the second preset cache resource pool needs to share.

[0120] By using a pool of ubiquitous computing resources (the second preset cache resource pool), and combining the relevance between each ubiquitous computing resource and the client output by the application target ubiquitous computing federated learning model, the frequency and weight of repeated use of a certain type of ubiquitous computing resource in a related region are predicted and analyzed. Based on this, the corresponding ubiquitous computing resources are cached in the second preset cache resource pool. By using a shared resource caching mechanism, the frequency of creation and destruction of ubiquitous computing resources is reduced, and the user experience of using ubiquitous computing resources is optimized.

[0121] Specifically, based on the preference levels of each client, the frequency and weight of reuse of each ubiquitous computing resource are predicted. Based on the frequency and weight of reuse, the shared weight of each ubiquitous computing resource in the second preset cache resource pool is determined (wherein, the frequency and weight of reuse are respectively associated with the shared weight). Based on the shared weight, the second ubiquitous computing resource that needs to be shared by the second preset cache resource pool is determined.

[0122] In this embodiment, the frequency of creating and destroying ubiquitous computing resources is reduced, optimizing the user experience of using ubiquitous computing resources (because it is faster to obtain ubiquitous computing resources from the preset cache resource pool).

[0123] Reference Figure 3 , Figure 3 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of this application.

[0124] like Figure 3 As shown, the ubiquitous computing power providing device based on federated learning may include: a processor 1001, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the processor 1001 and the memory 1005.

[0125] Optionally, the ubiquitous computing power provider based on federated learning may also include a user interface, a network interface, a camera, RF (Radio Frequency) circuitry, sensors, a WiFi module, etc. The user interface may include a display screen and an input submodule such as a keyboard; optional user interfaces may also include standard wired or wireless interfaces. The network interface may include standard wired or wireless interfaces (such as a Wi-Fi interface).

[0126] Those skilled in the art will understand that Figure 3 The structure of the ubiquitous computing power provider based on federated learning shown in the figure does not constitute a limitation on the ubiquitous computing power provider based on federated learning. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0127] like Figure 3As shown, the memory 1005, serving as a storage medium, may include an operating system, a network communication module, and a federated learning-based ubiquitous computing power provider. The operating system is a program that manages and controls the hardware and software resources of the federated learning-based ubiquitous computing power provider, supporting the operation of the federated learning-based ubiquitous computing power provider and other software and / or programs. The network communication module is used to enable communication between the various components within the memory 1005, as well as communication with other hardware and software in the federated learning-based ubiquitous computing power provider system.

[0128] exist Figure 3 In the ubiquitous computing power provision device based on federated learning shown, the processor 1001 is used to execute the ubiquitous computing power provision program based on federated learning stored in the memory 1005 to implement the steps of the ubiquitous computing power provision method based on federated learning described above.

[0129] The specific implementation method of the ubiquitous computing power provision device based on federated learning in this application is basically the same as the embodiments of the ubiquitous computing power provision method based on federated learning described above, and will not be repeated here.

[0130] This application also provides a ubiquitous computing power provision device based on federated learning, the device comprising:

[0131] The acquisition module is used to acquire request information from clients requesting ubiquitous computing resources;

[0132] The input module is used to input the request information into the target ubiquitous computing power federated learning model, process the request information based on the target ubiquitous computing power federated learning model, and obtain the target correlation degree between the client and different preset ubiquitous computing power resources.

[0133] The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels.

[0134] The first module is used to determine a ubiquitous computing resource allocation strategy that is suitable for the client based on the target correlation degree.

[0135] The selection module is used to select a target ubiquitous computing resource that is compatible with the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task.

[0136] In one possible embodiment of this application, the apparatus further includes:

[0137] The second determining module is used to determine the target training data with preset correlation labels of the first ubiquitous computing power provider, and to determine the second training data of the second ubiquitous computing power provider, wherein the second ubiquitous computing power provider is one or more, and the users in the training data of the first ubiquitous computing power provider and the second ubiquitous computing power provider are different but the training features are the same.

[0138] An iterative module is used to perform encrypted horizontal federated iterative training on a preset base model based on the target training data and the second training data until a target model that meets the preset training completion conditions is obtained.

[0139] The configuration module is used to set the target model as the target ubiquitous computing power federated learning model.

[0140] In one possible implementation of this application, the input module includes:

[0141] The first determining unit is configured to determine the request type of the request information, wherein the request type includes one or more of the following: calculation request type, network request type, and service request type.

[0142] The selection unit is used to select the target ubiquitous computing power federated learning model from a preset ubiquitous computing power federated learning model set based on the request type.

[0143] The input unit is used to input the request information into the target ubiquitous computing power federated learning model, and to process the request information based on the target ubiquitous computing power federated learning model.

[0144] In one possible embodiment of this application, the apparatus further includes:

[0145] The third determining module is used to determine the client's preference for each ubiquitous computing resource based on the target correlation degree.

[0146] The fourth determining module is used to determine the cache weight of each ubiquitous computing resource in the first preset cache resource pool of the client based on the preference level.

[0147] The fifth determining module is used to determine, based on the cache weight, the first ubiquitous computing power resource that the first preset cache resource pool needs to be used again.

[0148] In one possible embodiment of this application, the apparatus further includes:

[0149] The prediction module is used to predict the frequency and weight of the reuse of each ubiquitous computing resource based on the preference level of each client.

[0150] The sixth determining module is used to determine the shared weight of each ubiquitous computing resource in the second preset cache resource pool based on the frequency and weight of repeated use;

[0151] The seventh determining module is used to determine the second ubiquitous computing power resource that the second preset cache resource pool needs to share based on the shared weight.

[0152] In one possible embodiment of this application, the apparatus further includes:

[0153] The priority selection module is used to select a ubiquitous computing resource from the first ubiquitous computing resource that needs to be used again if the ubiquitous computing power request of the client is detected again, so that the client can complete the preset computing task again based on the selected ubiquitous computing power resource.

[0154] In one possible implementation of this application, the request includes at least one of the following: video traffic ubiquitous computing resource request, plaza traffic ubiquitous computing resource request, shopping mall traffic ubiquitous computing resource request, and game traffic ubiquitous computing resource request.

[0155] The specific implementation of the ubiquitous computing power provision device based on federated learning in this application is basically the same as the embodiments of the ubiquitous computing power provision method based on federated learning described above, and will not be repeated here.

[0156] This application provides a storage medium that stores one or more programs, which can be executed by one or more processors to implement the steps of the ubiquitous computing power provision method based on federated learning described above.

[0157] The specific implementation of the storage medium in this application is basically the same as the embodiments of the above-described method for providing ubiquitous computing power based on federated learning, and will not be repeated here.

[0158] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for providing ubiquitous computing power based on federated learning.

[0159] The specific implementation of the computer program product in this application is basically the same as the embodiments of the above-described method for providing ubiquitous computing power based on federated learning, and will not be repeated here.

[0160] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0161] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0162] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of a software plus hardware platform, or by hardware, but in many cases the former is a better implementation. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0163] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for providing ubiquitous computing power based on federated learning, characterized in that, The method includes: When a request for ubiquitous computing resources from a client is detected, the request information of the request is obtained; The request information is input into the target ubiquitous computing power federated learning model, and the request information is processed based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources. The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels. Based on the target correlation, a ubiquitous computing resource allocation strategy adapted to the client is determined; Based on the ubiquitous computing resource allocation strategy, a target ubiquitous computing resource is selected that is compatible with the client, so that the client can use the target ubiquitous computing resource to complete a preset computing task; The step of inputting the request information into the target ubiquitous computing power federated learning model and processing the request information based on the target ubiquitous computing power federated learning model includes: Determine the request type of the request information, wherein the request type includes one or more of the following: calculation request type, network request type, and service request type; Based on the request type, the target ubiquitous computing power federated learning model is selected from a preset set of ubiquitous computing power federated learning models; The request information is input into the target ubiquitous computing power federated learning model, and the request information is processed based on the target ubiquitous computing power federated learning model.

2. The method for providing ubiquitous computing power based on federated learning as described in claim 1, characterized in that, Before the step of processing the request information based on the target ubiquitous computing power federated learning model to obtain the target correlation degree between the client and different preset ubiquitous computing power resources, the method includes: The target training data with preset correlation labels of the first ubiquitous computing power provider is determined, and the second training data of the second ubiquitous computing power provider is determined. The second ubiquitous computing power provider is one or more, and the users in the training data of the first ubiquitous computing power provider and the second ubiquitous computing power provider are different but the training features are the same. Based on the target training data and the second training data, the preset base model is subjected to encrypted horizontal federated iterative training until a target model that meets the preset training completion conditions is obtained. The target model is used as the target ubiquitous computing power federated learning model.

3. The method for providing ubiquitous computing power based on federated learning as described in claim 1, characterized in that, After the step of selecting a target ubiquitous computing resource that is compatible with the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task, the method includes: Based on the target correlation, the client's preference for each ubiquitous computing resource is determined; Based on the degree of preference, the cache weight of each ubiquitous computing resource in the first preset cache resource pool of the client is determined; Based on the cache weight, the first ubiquitous computing resource that needs to be reused by the first preset cache resource pool is determined.

4. The method for providing ubiquitous computing power based on federated learning as described in claim 3, characterized in that, After the step of determining the client's preference for each ubiquitous computing resource based on the target correlation, the following is included: Based on the preferences of each client, predict the frequency and weight of the reuse of each ubiquitous computing resource; Based on the frequency and weight of repeated use, the shared weight of each ubiquitous computing resource in the second preset cache resource pool is determined. Based on the shared weight, the second ubiquitous computing resources that the second preset cache resource pool needs to share are determined.

5. The method for providing ubiquitous computing power based on federated learning as described in claim 3, characterized in that, After the step of determining the first ubiquitous computing resource that needs to be reused by the first preset cache resource pool based on the cache weight, the method includes: If the client's ubiquitous computing power request is detected again, a ubiquitous computing power resource is selected from the first ubiquitous computing power resource that needs to be used again, so that the client can complete the preset computing task again based on the selected ubiquitous computing power resource.

6. The method for providing ubiquitous computing power based on federated learning as described in any one of claims 1-5, characterized in that, The request includes at least one of the following: video traffic ubiquitous computing resource request, plaza traffic ubiquitous computing resource request, shopping mall traffic ubiquitous computing resource request, and game traffic ubiquitous computing resource request.

7. A ubiquitous computing power provision device based on federated learning, characterized in that, The device includes: The acquisition module is used to acquire request information from clients requesting ubiquitous computing resources; The input module is used to input the request information into the target ubiquitous computing power federated learning model, process the request information based on the target ubiquitous computing power federated learning model, and obtain the target correlation degree between the client and different preset ubiquitous computing power resources. The target ubiquitous computing power federated learning model is obtained by federated learning of a preset basic model based on target training data with preset correlation labels. The first module is used to determine a ubiquitous computing resource allocation strategy that is suitable for the client based on the target correlation degree. The selection module is used to select a target ubiquitous computing resource that is compatible with the client based on the ubiquitous computing resource allocation strategy, so that the client can use the target ubiquitous computing resource to complete a preset computing task; The input module includes: The first determining unit is configured to determine the request type of the request information, wherein the request type includes one or more of the following: calculation request type, network request type, and service request type. The selection unit is used to select the target ubiquitous computing power federated learning model from a preset ubiquitous computing power federated learning model set based on the request type. The input unit is used to input the request information into the target ubiquitous computing power federated learning model, and to process the request information based on the target ubiquitous computing power federated learning model.

8. A ubiquitous computing power provision device based on federated learning, characterized in that, The method includes a memory, a processor, and a federated learning-based ubiquitous computing power provider stored in the memory and executable on the processor. When the processor executes the federated learning-based ubiquitous computing power provider, it implements the steps of the federated learning-based ubiquitous computing power provision method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a ubiquitous computing power provider based on federated learning, which, when executed by a processor, implements the steps of the ubiquitous computing power provider method based on federated learning as described in any one of claims 1 to 6.

Images