CN117271103A - Ubiquitous computing power providing method based on federal learning and related equipment thereof - Google Patents
Ubiquitous computing power providing method based on federal learning and related equipment thereof Download PDFInfo
- Publication number
- CN117271103A CN117271103A CN202210467589.0A CN202210467589A CN117271103A CN 117271103 A CN117271103 A CN 117271103A CN 202210467589 A CN202210467589 A CN 202210467589A CN 117271103 A CN117271103 A CN 117271103A
- Authority
- CN
- China
- Prior art keywords
- ubiquitous computing
- target
- client
- computing power
- federal learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000012549 training Methods 0.000 claims abstract description 60
- 238000013468 resource allocation Methods 0.000 claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000004364 calculation method Methods 0.000 claims description 22
- 230000003044 adaptive effect Effects 0.000 abstract 1
- 238000004891 communication Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000004590 computer program Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 230000006378 damage Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/5011—Pool
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Power Sources (AREA)
Abstract
The application discloses a ubiquitous computing power providing method based on federal learning and relevant equipment thereof, wherein the method comprises the following steps: when a request of a ubiquitous computing resource of a client is detected, acquiring request information of the request; inputting the request information into a target ubiquitous computing power federation learning model, processing the request information based on the target ubiquitous computing power federation learning model to obtain target association degrees between a client and different preset ubiquitous computing power resources, wherein the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with preset association degree labels; determining a ubiquitous computing power resource allocation strategy of the adaptive client based on the target association degree; and selecting a target ubiquitous computing resource adapted to the client based on a ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource. In the application, the needed ubiquitous computing resource is accurately provided for the client.
Description
Technical Field
The present application relates to the field of communications computers, and in particular, to a ubiquitous computing power providing method, device, apparatus and storage medium based on federal learning.
Background
Edge computing is the provision of basic computing power resources, such as computation, storage, etc., to a user in close proximity to the data source or user. Compared with the traditional internet architecture adopted by the center calculation, namely the basic topological structure of the end-to-end model, the edge calculation is gradually changed into a providing mode of ubiquitous computing resources, namely specifically, the basic topological structure of the existing end-to-end model is changed into a network topological structure of ubiquitous computing resources from the perspective of completing the task of the edge calculation (considering factors such as space distance and network transmission delay), and further, the calculation is gradually changed from the center to the edge, and the directions of light weight, dynamic, no service, function calculation and the like are changed.
However, in the existing edge computing process of providing ubiquitous computing resources for clients, computing resources are abstracted by a computing platform layer through computing modeling, after a computing capability template is formed, the ubiquitous computing resources are provided for the clients based on the computing capability template, however, the computing capability template does not have the pertinence of client requirements, and the privacy of the clients is easy to leak, so that the required ubiquitous computing resources are difficult to accurately and safely provide for the client pertinence.
Disclosure of Invention
In view of this, the embodiments of the present application provide a ubiquitous computing power providing method, apparatus, device and storage medium based on federal learning, which aims to solve the technical problem that it is difficult to accurately and safely provide a client with a required ubiquitous computing power resource in the prior art.
The embodiment of the application provides a ubiquitous computing power providing method based on federal learning, which comprises the following steps:
when a request of a ubiquitous computing resource of a client is detected, acquiring request information of the request;
inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree;
and selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
In a possible implementation manner of the present application, before the step of processing the request information based on the target ubiquitous computing power federal learning model to obtain the target association degree between the client and different preset ubiquitous computing power resources, the method includes:
determining target training data with preset relevance labels of a ubiquitous computing power first provider and second training data of a ubiquitous computing power second provider, wherein one or more ubiquitous computing power second providers are provided, and users in the training data of the ubiquitous computing power first provider and the ubiquitous computing power second provider are different but have consistent training characteristics;
based on the target training data and the second training data, performing encryption transverse federal iterative training on a preset basic model until a target model meeting preset training completion conditions is obtained;
and taking the target model as the target ubiquitous computing power federal learning model.
In one possible implementation manner of the present application, the step of inputting the request information into a target ubiquitous computing federal learning model and processing the request information based on the target ubiquitous computing federal learning model includes:
Determining a request type of the request information, wherein the request type comprises one or more of a calculation request type, a network request type and a service request type;
selecting the target ubiquitous computing power federal learning model from a preset ubiquitous computing power federal learning model set based on the request type;
and inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model.
In one possible implementation manner of the present application, after the step of selecting, based on the ubiquitous computing resource allocation policy, a target ubiquitous computing resource adapted to the client for the client to complete a preset computing task using the target ubiquitous computing resource, the method includes:
determining the preference degree of the client on the ubiquitous computing resource based on the target association degree;
based on the preference degree, determining the cache weight of each ubiquitous computing resource in a first preset cache resource pool of the client;
and determining a first ubiquitous computing resource which is needed to be reused by the first preset cache resource pool based on the cache weight.
In a possible implementation manner of the present application, after the step of determining the preference degree of the client on the ubiquitous computing resource based on the target association degree, the method includes:
predicting the frequency and weight of repeated use of each ubiquitous computing resource based on the preference degree of each client;
determining the sharing weight of each ubiquitous computing resource in a second preset cache resource pool based on the reused frequency and weight;
and determining a second ubiquitous computing resource which needs to be shared by the second preset cache resource pool based on the sharing weight.
In a possible implementation manner of the present application, after the step of determining, based on the buffer weights, a first ubiquitous computing resource that needs to be reused by the first preset buffer resource pool, the method includes:
and if the ubiquitous computing power request of the client is detected again, preferentially selecting ubiquitous computing power resources from the first ubiquitous computing power resources needing to be reused, so that the client can finish the preset computing task again based on the selected ubiquitous computing power resources.
In one possible embodiment of the present application, the request includes at least one of a video traffic ubiquitous computing resource request, a square traffic ubiquitous computing resource request, a mall traffic ubiquitous computing resource request, and a game traffic ubiquitous computing resource request.
The application also provides a ubiquitous computing power providing device based on federal learning, the device comprising:
the acquisition module is used for acquiring request information of a client for requesting ubiquitous computing resource;
the input module is used for inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
the first module determining module is used for determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree;
and the selection module is used for selecting and adapting the target ubiquitous computing resource of the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
The application also provides a ubiquitous computing power providing device based on federal learning, wherein the ubiquitous computing power providing device based on federal learning is entity node equipment, and the ubiquitous computing power providing device based on federal learning comprises: the computer-readable storage medium comprises a memory, a processor and a program stored on the memory and executable on the processor for performing the method for providing ubiquitous computing force based on federal learning.
In order to achieve the above object, there is also provided a storage medium having stored thereon a ubiquitous computing force providing program based on federal learning, which when executed by a processor, implements the steps of any one of the above-described ubiquitous computing force providing methods based on federal learning.
Compared with the existing method, device, equipment and storage medium for providing ubiquitous computing resources for a client through a ubiquitous computing matching template only by a computing platform layer, which causes difficulty in accurately providing the client with the required ubiquitous computing resources, the method, device and storage medium for providing ubiquitous computing resources based on federal learning, in the method, device and storage medium for providing ubiquitous computing resources based on federal learning, acquire request information of a request when the request of the ubiquitous computing resources of the client is detected; inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources; the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label; determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree; and selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource. It can be understood that in the application, when a request of a client ubiquitous computing resource is detected, the target association degree between the client and different preset ubiquitous computing resources can be accurately obtained based on a target ubiquitous computing federal learning model, that is, the ubiquitous computing resource matched with the client demand can be mined based on the target ubiquitous computing federal learning model, and then the ubiquitous computing resource allocation strategy adapting to the client demand can be accurately determined, so that the required ubiquitous computing resource can be accurately provided for the client.
Drawings
FIG. 1 is a flow chart of a first embodiment of a ubiquitous computing force providing method based on federal learning according to the present application;
fig. 2 is a schematic flow chart corresponding to step S01-step S03 in an embodiment of a ubiquitous computing power providing method based on federal learning;
FIG. 3 is a schematic diagram of a device architecture of a hardware operating environment according to an embodiment of the present application;
fig. 4 is a schematic view of a first scenario involved in an embodiment of the present application;
fig. 5 is a schematic diagram of a second scenario according to an embodiment of the present application.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
In a first embodiment of a ubiquitous computing force providing method based on federal learning according to the present application, referring to fig. 1, the ubiquitous computing force providing method based on federal learning includes:
step S10, when a request of a ubiquitous computing resource of a client is detected, request information of the request is obtained;
step S20, inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
The target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
step S30, determining a ubiquitous computing power resource allocation strategy adapting to the client based on the target association degree;
step S40, selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
In this embodiment, the ubiquitous computing power providing method based on federal learning is applied to a ubiquitous computing power providing device based on federal learning, the ubiquitous computing power providing device based on federal learning is applied to a ubiquitous computing power providing system based on federal learning, the ubiquitous computing power providing system based on federal learning belongs to a ubiquitous computing power providing platform based on federal learning, and the platform can be a computer, etc.
In this embodiment, the ubiquitous computing power resource required by the current service requirement of the client is determined through the target ubiquitous computing power federal learning model, so as to accurately provide the required ubiquitous computing power resource for the client.
In this embodiment, the target ubiquitous computing power federal learning model is used to mine the ubiquitous computing power resources required by the clients with different service requirements, so as to pertinently recommend the ubiquitous computing power resources of interest to the clients.
In the embodiment, through the target ubiquitous computing power federal learning model, repeated creation and destruction of ubiquitous computing power resources are avoided, and excessive overhead cost and network bandwidth occupation are avoided.
In the embodiment, assistance is provided for optimizing ubiquitous computing resources of the client through updating of the target ubiquitous computing federal learning model.
In this embodiment, the client may be a client of an enterprise, a company, an organization, etc., which has the requirement of ubiquitous computing resources.
The method comprises the following specific steps:
step S10, when a request of a ubiquitous computing resource of a client is detected, request information of the request is obtained;
in this embodiment, the ubiquitous computing resource providing platform (hereinafter referred to as platform) based on federal learning has different interfaces, and on the corresponding application interfaces, the client may apply for a ubiquitous computing resource, and further, when the platform detects a request for the ubiquitous computing resource of the client, obtain request information of the request, where the request information includes a video traffic ubiquitous computing resource request, a square traffic ubiquitous computing resource request, a market traffic ubiquitous computing resource request, and a game traffic ubiquitous computing resource request.
In this embodiment, the request information may include information of calculation type, calculation time, calculation amount, request place, calculation object, and the like.
Step S20, inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
in this embodiment, it should be noted that, the target ubiquitous computing power federal learning model is obtained after federal learning is performed on a preset basic model based on target training data with a preset association degree label, and can accurately determine a model of target association degrees between a client and different preset ubiquitous computing power resources.
Referring to fig. 2, before the step of processing the request information based on the target ubiquitous computing power federal learning model to obtain the target association degree between the client and different preset ubiquitous computing power resources, the method includes:
Step S01, determining target training data with a preset relevance label of a ubiquitous computing power first provider, and determining second training data of a ubiquitous computing power second provider, wherein one or more ubiquitous computing power second providers are provided, and users in the training data of the ubiquitous computing power first provider and the ubiquitous computing power second provider are different but have consistent training characteristics;
step S02, based on the target training data and the second training data, performing encryption transverse federal iterative training on a preset basic model until a target model meeting preset training completion conditions is obtained;
and S03, taking the target model as the target ubiquitous computing power federal learning model.
In this embodiment, it is explained how to obtain a target ubiquitous computing federal learning model.
In this embodiment, the privacy computation is first described, and the privacy computation (Privacy Computing) refers to a technique and a system for joint computation by two or more participants (ubiquitous computing force providers, in this embodiment, including a ubiquitous computing force first provider and a ubiquitous computing force second provider), which perform joint machine learning and joint analysis on their data through collaboration without revealing the respective data. The participants in the privacy calculations may be different departments of the same organization or different organizations. Under the privacy computing framework, the data plaintext of the participant cannot be out of the local area, so that multi-source data cross-domain cooperation is realized while the data security is protected, and the difficult problem of data protection and fusion application is solved.
Currently, privacy calculation (Privacy Computing) mainly includes federal learning calculation, federal learning refers to a method of machine learning by combining different participants, and a cycle of parameter updating in federal learning is divided into two steps: (a) Each participant trains the machine learning model using only its own data and sends model parameter updates to a central coordinator (modeling Pipeline for federal training in this embodiment); (b) The coordinator fuses (e.g., averages) the received model updates from different participants, and redistributes the fused model parameter updates to each participant, and in federal learning, the participants do not need to expose own data to other participants or the coordinator, so that federal learning can well protect user privacy and ensure data security.
The federation learning comprises longitudinal federation learning calculation and transverse federation learning calculation, and the longitudinal federation learning is Sample alignment federation learning (Sample-Aligned Federated Learning) which is suitable for the situations that the IDs of the training samples of the participators overlap more and the data features overlap less, and the training and the optimization of the model are completed in a multiparty cooperative manner under the framework of security and confidentiality.
The horizontal federal learning (Horizontal Federated Learning) is suitable for the situation that the characteristics of the training sample data of the participants overlap more and the user IDs are less, and the training and the optimization of the model are completed cooperatively by multiple parties under the framework of security and confidentiality.
In this embodiment, the target ubiquitous computing power federal learning model is obtained through horizontal federal learning training.
Specifically, the ubiquitous computing force first provider may be one, i.e., the ubiquitous computing force provider a, and the ubiquitous computing force second provider is one, i.e., the ubiquitous computing force provider B, as shown in fig. 4.
In this embodiment, the ubiquitous computing force first provider may be one, that is, the ubiquitous computing force provider a, and the ubiquitous computing force second provider is a plurality of, that is, the ubiquitous computing force provider B and the ubiquitous computing force provider C. As shown in fig. 5.
In this embodiment, users of different ubiquitous computing providers are different, but training features are consistent, or similar, wherein training data of different ubiquitous computing providers is processed by encrypted sample data alignment, etc.
The method specifically performs encryption transverse federal iterative training on the preset basic model based on the target training data and the second training data until an iterative process for obtaining the target model meeting the preset training completion condition can be as follows:
As shown in fig. 5, 1, each ubiquitous computing force provider sends encrypted sample data (target training data with preset association label of a ubiquitous computing force first provider and second training data of a ubiquitous computing force second provider) to a federal learning modeling Pipeline service of a platform;
2. federal learning modeling Pipeline service of the platform aggregates data of all ubiquitous computing power providers to perform federal learning training, encrypts and updates model parameters;
3. the federal learning modeling Pipeline service turns over the updated model to each ubiquitous computing provider;
4. each ubiquitous computing force provider updates a respective model.
And continuously iterating the steps 1-5 until a target model meeting the preset training completion condition is obtained.
In the iteration process, the encrypted sample data are aligned, encryption federation training is carried out, and a target ubiquitous computing federation learning model which can be deployed and applied is obtained, wherein the target ubiquitous computing federation learning model is continuously updated.
Specifically, in this embodiment, if the request information is a video traffic ubiquitous computing power resource request, the target ubiquitous computing power federal learning model is a video-associated ubiquitous computing power federal learning model, if the request information is a market traffic ubiquitous computing power resource request, the target ubiquitous computing power federal learning model is a market-associated ubiquitous computing power federal learning model, if the request information is a square traffic ubiquitous computing power resource request, the target ubiquitous computing power federal learning model is a square-associated ubiquitous computing power federal learning model, and if the request information is a game traffic ubiquitous computing power resource request, the target ubiquitous computing power federal learning model is a game-associated ubiquitous computing power federal learning model.
In this embodiment, the different preset ubiquitous computing resources may be a storage ubiquitous computing resource, a calculation ubiquitous computing resource, a network ubiquitous computing resource, a business process ubiquitous computing resource, and the like.
The system stores ubiquitous computing power resources, calculates ubiquitous computing power resources, network ubiquitous computing power resources, and can subdivide business flow ubiquitous computing power resources and the like.
In this embodiment, based on the request information, it is determined what kind of ubiquitous computing resource (storage ubiquitous computing resource, calculation ubiquitous computing resource, network ubiquitous computing resource, business flow ubiquitous computing resource, etc.) is needed by the client, specifically, it may be further subdivided, and it is determined that the client is more specific ubiquitous computing resource.
If the target association degree is: storing ubiquitous computing resources: 0.8;
calculating ubiquitous computing power resources: 0.3;
network ubiquitous computing resource: 0.1, etc.
The client needs to store ubiquitous computing resources.
Step S40, based on the target association degree, determining a ubiquitous computing resource allocation strategy adapting to the client
After determining the target association degree, determining a ubiquitous computing resource allocation policy adapting to the client, specifically, if the association degree between the client and the storage ubiquitous computing resource is higher, determining the resource needing to be allocated to the client from the preset storage ubiquitous computing resource by the ubiquitous computing resource allocation policy, wherein the specific allocation policy is related to a preset finer policy determination rule.
Step S50, selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
In this embodiment, after determining the ubiquitous computing resource allocation policy, selecting, based on the computing resource allocation policy, a target ubiquitous computing resource adapted to the client so that the client can complete a preset computing task by using the target ubiquitous computing resource.
The preset computing task may be a video stream computing task or a game flow computing task, which is not specifically limited.
Compared with the existing method, device, equipment and storage medium for providing ubiquitous computing resources for a client through a ubiquitous computing matching template only by a computing platform layer, which causes difficulty in accurately providing the client with the required ubiquitous computing resources, the method, device and storage medium for providing ubiquitous computing resources based on federal learning, in the method, device and storage medium for providing ubiquitous computing resources based on federal learning, acquire request information of a request when the request of the ubiquitous computing resources of the client is detected; inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources; the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label; determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree; and selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource. It can be understood that in the application, when a request of a client ubiquitous computing resource is detected, the target association degree between the client and different preset ubiquitous computing resources can be accurately obtained based on a target ubiquitous computing federal learning model, that is, the ubiquitous computing resource matched with the client demand can be mined based on the target ubiquitous computing federal learning model, and then the ubiquitous computing resource allocation strategy adapting to the client demand can be accurately determined, so that the required ubiquitous computing resource can be accurately provided for the client.
Further, based on the first embodiment in the present application, another embodiment in the present application is provided, in this embodiment, the step of inputting the request information into a target ubiquitous computing federal learning model, and processing the request information based on the target ubiquitous computing federal learning model includes:
step M1, determining a request type of the request information, wherein the request type comprises one or more of a calculation request type, a network request type and a service request type;
in this embodiment, the request type of the request information is determined, and the request type includes a calculation request type, a network request type, and a service request type, and in addition, the request type may be combined with specific request information, for example, the request type may be a video stream calculation request type, a video stream network request type, a video stream service request type, and the like.
Step M2, selecting the target ubiquitous computing power federal learning model from a preset ubiquitous computing power federal learning model set based on the request type;
and step M3, inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model.
In this embodiment, different request types have targeted ubiquitous computing power federal learning models, that is, in this embodiment, a plurality of ubiquitous computing power federal learning models are trained, and the plurality of ubiquitous computing power federal learning models form a preset ubiquitous computing power federal learning model set.
In this embodiment, based on the request type, the target ubiquitous computing power federal learning model is selected from a preset set of ubiquitous computing power federal learning models, the request information is input into the target ubiquitous computing power federal learning model, and the request information is processed based on the target ubiquitous computing power federal learning model. In this embodiment, pertinence of the processing is promoted.
Further, based on the first embodiment of the present application, another embodiment of the present application is provided, in this embodiment, after the step of selecting, based on the ubiquitous computing resource allocation policy, a target ubiquitous computing resource adapted to the client for the client to use the target ubiquitous computing resource to complete a preset computing task, the method includes:
step N1, determining the preference degree of the client on the ubiquitous computing resource based on the target association degree;
Step N2, based on the preference degree, determining the cache weight of each ubiquitous computing resource in a first preset cache resource pool of the client;
and step N3, determining a first ubiquitous computing resource which is needed to be reused by the first preset cache resource pool based on the cache weight.
In this embodiment, the federal learning modeling Pipeline service of the platform issues a ubiquitous computing resource pool service for each client, and the ubiquitous computing resource pool service updates the optimized resource pool resources for the client, and in particular updates the cached resources of the optimized resource pool for the client.
In this embodiment, the resource pool includes a first preset cache resource pool, where the first preset cache resource pool corresponds to each client.
In this embodiment, based on the target association degree, a preference degree of the client for each ubiquitous computing resource is determined, based on the preference degree, a cache weight of each ubiquitous computing resource in a first preset cache resource pool of the client is determined, the preference degree is high, the cache weight is high, a preset mapping relationship exists between the preference degree and the cache weight, and the preset mapping relationship can be set.
And determining a first ubiquitous computing resource which is needed to be reused in the first preset cache resource pool based on the cache weight, thereby avoiding repeated destruction and establishment of the first ubiquitous computing resource for a plurality of times and avoiding consumption of the resource.
After the step of determining, based on the buffer weight, a first ubiquitous computing resource that needs to be reused by the first preset buffer resource pool, the method includes:
and step H1, if the ubiquitous computing power request of the client is detected again, preferentially selecting ubiquitous computing power resources from the first ubiquitous computing power resources needing to be reused, so that the client can finish the preset computing task again based on the selected ubiquitous computing power resources.
In this embodiment, if the ubiquitous computing resource request of the client is detected again, the ubiquitous computing resource is preferentially selected from the first ubiquitous computing resource that needs to be reused, so that the reconstruction of the needed ubiquitous computing resource for the client in a short time is avoided, and the consumption of excessive material resources is avoided.
After the step of determining the preference degree of the client on the ubiquitous computing resource based on the target association degree, the method comprises the following steps:
step G1, predicting the frequency and weight of repeated use of each ubiquitous computing resource based on the preference degree of each client;
step G2, determining the sharing weight of each ubiquitous computing resource in a second preset cache resource pool based on the reused frequency and weight;
And G3, determining a second ubiquitous computing resource which needs to be shared by the second preset cache resource pool based on the sharing weight.
The method comprises the steps of predicting and analyzing the frequency and weight of the ubiquitous computing resource of a certain type in a certain area in a repeated mode by combining the correlation degree of each ubiquitous computing resource and a client output by an application target ubiquitous computing federal learning model through a ubiquitous computing resource object pool (a second preset cache resource pool), caching the corresponding ubiquitous computing resource in the second preset cache resource pool according to the frequency and weight of the ubiquitous computing resource repeatedly used in a certain area, reducing the frequency of creating and destroying the ubiquitous computing resource by utilizing a shared resource caching mechanism, and optimizing the experience of using the ubiquitous computing resource by a user.
Specifically, based on the preference degree of each client, predicting the frequency and weight of repeated use of each ubiquitous computing resource, determining the sharing weight of each ubiquitous computing resource in a second preset cache resource pool based on the repeated use frequency and weight (wherein the repeated use frequency and weight respectively have a preset mapping relation with the sharing weight), and determining the second ubiquitous computing resource which needs to be shared by the second preset cache resource pool based on the sharing weight.
In this embodiment, the frequency of creating and destroying the ubiquitous computing resource is reduced, and the experience of using the ubiquitous computing resource by the user is optimized (because the speed of acquiring the ubiquitous computing resource from the preset cache resource pool is faster).
Referring to fig. 3, fig. 3 is a schematic device structure diagram of a hardware running environment according to an embodiment of the present application.
As shown in fig. 3, the ubiquitous computing force providing device based on federal learning may include: a processor 1001, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to enable connected communication between the processor 1001 and the memory 1005.
Optionally, the ubiquitous computing force providing device based on federal learning may further include a user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, a WiFi module, and the like. The user interface may include a Display, an input sub-module such as a Keyboard (Keyboard), and the optional user interface may also include a standard wired interface, a wireless interface. The network interface may include a standard wired interface, a wireless interface (e.g., WI-FI interface).
It will be appreciated by those skilled in the art that the federal learning-based ubiquitous computing force-providing device structure shown in fig. 3 does not constitute a limitation of the federal learning-based ubiquitous computing force-providing device, and may include more or less components than those illustrated, or certain components may be combined, or a different arrangement of components.
As shown in fig. 3, an operating system, a network communication module, and a ubiquitous computing power providing program based on federal learning may be included in the memory 1005 as one type of storage medium. The operating system is a program that manages and controls the ubiquitous computing force providing device hardware and software resources based on federal learning, supporting the operation of the ubiquitous computing force providing program based on federal learning, as well as other software and/or programs. The network communication module is used to enable communication between components within the memory 1005 and with other hardware and software in the federal learning based ubiquitous computing power provisioning system.
In the federal learning-based ubiquitous computing power providing apparatus shown in fig. 3, a processor 1001 is configured to execute a federal learning-based ubiquitous computing power providing program stored in a memory 1005, to implement the steps of the federal learning-based ubiquitous computing power providing method according to any one of the above.
The specific implementation manner of the ubiquitous computing force providing device based on federal learning is basically the same as that of each embodiment of the ubiquitous computing force providing method based on federal learning, and is not repeated here.
The application also provides a ubiquitous computing power providing device based on federal learning, the device comprising:
The acquisition module is used for acquiring request information of a client for requesting ubiquitous computing resource;
the input module is used for inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
the first module determining module is used for determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree;
and the selection module is used for selecting and adapting the target ubiquitous computing resource of the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
In a possible embodiment of the present application, the apparatus further comprises:
the second determining module is used for determining target training data with preset relevance labels of the ubiquitous computing power first provider and determining second training data of the ubiquitous computing power second provider, wherein one or more ubiquitous computing power second providers are provided, and users in the training data of the ubiquitous computing power first provider and the ubiquitous computing power second provider are different but have consistent training characteristics;
The iteration module is used for carrying out encryption transverse federal iteration training on a preset basic model based on the target training data and the second training data until a target model meeting preset training completion conditions is obtained;
and the setting module is used for taking the target model as the target ubiquitous computing power federal learning model.
In one possible embodiment of the present application, the input module includes:
a first determining unit, configured to determine a request type of the request information, where the request type includes one or more of a calculation request type, a network request type, and a service request type;
the selection unit is used for selecting the target ubiquitous computing power federal learning model from a preset ubiquitous computing power federal learning model set based on the request type;
the input unit is used for inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model.
In a possible embodiment of the present application, the apparatus further comprises:
the third determining module is used for determining the preference degree of the client on the ubiquitous computing resource based on the target association degree;
A fourth determining module, configured to determine, based on the preference degree, a cache weight of each ubiquitous computing resource in a first preset cache resource pool of the client;
and a fifth determining module, configured to determine, based on the buffer weight, a first ubiquitous computing resource that needs to be reused by the first preset buffer resource pool.
In a possible embodiment of the present application, the apparatus further comprises:
the prediction module is used for predicting the frequency and weight of repeated use of each ubiquitous computing resource based on the preference degree of each client;
a sixth determining module, configured to determine, based on the frequency and the weight of the repeated use, a shared weight of each ubiquitous computing resource in a second preset buffer resource pool;
and a seventh determining module, configured to determine, based on the sharing weight, a second ubiquitous computing resource that needs to be shared by the second preset cache resource pool.
In a possible embodiment of the present application, the apparatus further comprises:
and the priority selection module is used for preferentially selecting the ubiquitous computing resource from the first ubiquitous computing resource needing to be reused if the ubiquitous computing request of the client is detected again, so that the client can finish the preset computing task again based on the selected ubiquitous computing resource.
In one possible embodiment of the present application, the request includes at least one of a video traffic ubiquitous computing resource request, a square traffic ubiquitous computing resource request, a mall traffic ubiquitous computing resource request, and a game traffic ubiquitous computing resource request.
The specific implementation manner of the ubiquitous computing force providing device based on federal learning is basically the same as that of each embodiment of the ubiquitous computing force providing method based on federal learning, and is not repeated here.
Embodiments of the present application provide a storage medium, where one or more programs are stored, where the one or more programs are further executable by one or more processors to implement the steps of the federal learning-based ubiquitous computing power providing method according to any one of the above.
The specific implementation manner of the storage medium is basically the same as that of the above embodiments of the ubiquitous computing power providing method based on federal learning, and is not repeated here.
The present application also provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the federal learning-based ubiquitous computing force providing method described above.
The specific embodiments of the computer program product of the present application are substantially the same as the embodiments of the ubiquitous computing power providing method based on federal learning, and are not described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be apparent to those skilled in the art that the above embodiment method may be implemented by means of a software-and-hardware platform, or may be implemented by hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. A ubiquitous computing force providing method based on federal learning, the method comprising:
when a request of a ubiquitous computing resource of a client is detected, acquiring request information of the request;
inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree;
and selecting a target ubiquitous computing resource adapted to the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
2. The method for providing ubiquitous computing force based on federal learning according to claim 1, wherein before the step of processing the request information based on the target ubiquitous computing force federal learning model to obtain the target association degree between the client and different preset ubiquitous computing force resources, the method comprises:
determining target training data with preset relevance labels of a ubiquitous computing power first provider and second training data of a ubiquitous computing power second provider, wherein one or more ubiquitous computing power second providers are provided, and users in the training data of the ubiquitous computing power first provider and the ubiquitous computing power second provider are different but have consistent training characteristics;
based on the target training data and the second training data, performing encryption transverse federal iterative training on a preset basic model until a target model meeting preset training completion conditions is obtained;
and taking the target model as the target ubiquitous computing power federal learning model.
3. The federal learning-based ubiquitous computing power providing method according to claim 1, wherein the step of inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model, comprises:
Determining a request type of the request information, wherein the request type comprises one or more of a calculation request type, a network request type and a service request type;
selecting the target ubiquitous computing power federal learning model from a preset ubiquitous computing power federal learning model set based on the request type;
and inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model.
4. The method for providing ubiquitous computing force based on federal learning according to claim 1, wherein after the step of selecting a target ubiquitous computing force resource adapted to the client for the client to complete a preset computing task using the target ubiquitous computing force resource based on the ubiquitous computing force resource allocation policy, the method comprises:
determining the preference degree of the client on the ubiquitous computing resource based on the target association degree;
based on the preference degree, determining the cache weight of each ubiquitous computing resource in a first preset cache resource pool of the client;
and determining a first ubiquitous computing resource which is needed to be reused by the first preset cache resource pool based on the cache weight.
5. The federally learned based ubiquitous computing force providing method according to claim 4, wherein after the step of determining the preference degree of the client for each ubiquitous computing force resource based on the target association degree, it comprises:
predicting the frequency and weight of repeated use of each ubiquitous computing resource based on the preference degree of each client;
determining the sharing weight of each ubiquitous computing resource in a second preset cache resource pool based on the reused frequency and weight;
and determining a second ubiquitous computing resource which needs to be shared by the second preset cache resource pool based on the sharing weight.
6. The federally learned based ubiquitous computing force providing method according to claim 4, wherein after the step of determining a first ubiquitous computing force resource that the first preset cache resource pool needs to be reused based on the cache weight, the method comprises:
and if the ubiquitous computing power request of the client is detected again, preferentially selecting ubiquitous computing power resources from the first ubiquitous computing power resources needing to be reused, so that the client can finish the preset computing task again based on the selected ubiquitous computing power resources.
7. The federal learning-based ubiquitous computing force providing method according to any one of claims 1 to 6, wherein the request includes at least one of a video traffic ubiquitous computing force resource request, a square traffic ubiquitous computing force resource request, a market traffic ubiquitous computing force resource request, and a game traffic ubiquitous computing force resource request.
8. A ubiquitous computing force providing device based on federal learning, the device comprising:
the acquisition module is used for acquiring request information of a client for requesting ubiquitous computing resource;
the input module is used for inputting the request information into a target ubiquitous computing power federal learning model, and processing the request information based on the target ubiquitous computing power federal learning model to obtain target association degrees between the client and different preset ubiquitous computing power resources;
the target ubiquitous computing power federation learning model is obtained after federation learning is carried out on a preset basic model based on target training data with a preset association degree label;
the first module determining module is used for determining a ubiquitous computing resource allocation strategy adapting to the client based on the target association degree;
and the selection module is used for selecting and adapting the target ubiquitous computing resource of the client based on the ubiquitous computing resource allocation strategy so as to enable the client to complete a preset computing task by using the target ubiquitous computing resource.
9. A federal learning-based ubiquitous computing power providing apparatus comprising a memory, a processor and a federal learning-based ubiquitous computing power providing program stored on the memory and executable on the processor, the processor implementing the steps of the federal learning-based ubiquitous computing power providing method according to any one of claims 1 to 7 when executing the federal learning-based ubiquitous computing power providing program.
10. A computer readable storage medium, characterized in that it has stored thereon a federal learning-based ubiquitous computing force providing program, which when executed by a processor, implements the steps of the federal learning-based ubiquitous computing force providing method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210467589.0A CN117271103A (en) | 2022-04-24 | 2022-04-24 | Ubiquitous computing power providing method based on federal learning and related equipment thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210467589.0A CN117271103A (en) | 2022-04-24 | 2022-04-24 | Ubiquitous computing power providing method based on federal learning and related equipment thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117271103A true CN117271103A (en) | 2023-12-22 |
Family
ID=89209186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210467589.0A Pending CN117271103A (en) | 2022-04-24 | 2022-04-24 | Ubiquitous computing power providing method based on federal learning and related equipment thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117271103A (en) |
-
2022
- 2022-04-24 CN CN202210467589.0A patent/CN117271103A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110633806B (en) | Longitudinal federal learning system optimization method, device, equipment and readable storage medium | |
| CN110263936B (en) | Horizontal federal learning method, device, equipment and computer storage medium | |
| CN110245510B (en) | Method and apparatus for predicting information | |
| CN111860864A (en) | Longitudinal federal modeling optimization method, device and readable storage medium | |
| JP6644800B2 (en) | Determination of area to be superimposed with image, image superimposition, image display method and apparatus | |
| CN111061956B (en) | Method and apparatus for generating information | |
| US20210297323A1 (en) | Method for gathering traffic analytics data about a communication network | |
| CN110751294A (en) | Model prediction method, device, equipment and medium combining multi-party characteristic data | |
| US10263858B2 (en) | Environment simulator for user percentile | |
| CN110837653B (en) | Label prediction method, apparatus and computer readable storage medium | |
| US9910724B2 (en) | Fast and accurate identification of message-based API calls in application binaries | |
| CN111258602A (en) | Information updating method and device | |
| JP2016508321A (en) | System architecture, subsystems, and methods for opening telecommunication network functions | |
| US20220385739A1 (en) | Method and apparatus for generating prediction information, electronic device, and computer readable medium | |
| CN110691042A (en) | Resource allocation method and device | |
| CN110909521A (en) | Synchronous processing method and device for online document information and electronic equipment | |
| CN111652381A (en) | Data set contribution degree evaluation method, device and equipment and readable storage medium | |
| CN110309142B (en) | Method and device for rule management | |
| CN112434620A (en) | Scene character recognition method, device, equipment and computer readable medium | |
| CN112257582A (en) | Foot posture determination method, device, equipment and computer readable medium | |
| CN112115217A (en) | Data processing method and device for high-precision map, electronic equipment and storage medium | |
| CN111767411A (en) | Knowledge graph representation learning optimization method and device and readable storage medium | |
| CN111652382B (en) | Data processing method, device and equipment based on block chain and storage medium | |
| CN106131109A (en) | The method and apparatus of any cost caching in the embedded webpage of APP is realized based on http protocol | |
| CN113259353A (en) | Information processing method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |