CN115242756A

CN115242756A - Processing method, device, equipment and system for federated learning service

Info

Publication number: CN115242756A
Application number: CN202110357540.5A
Authority: CN
Inventors: 王钰; 张星; 耿慧拯; 张晶
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Priority date: 2021-04-01
Filing date: 2021-04-01
Publication date: 2022-10-25
Anticipated expiration: 2041-04-01
Also published as: CN115242756B

Abstract

The invention provides a processing method, a device, equipment and a system for a federated learning service, wherein the method comprises the following steps: receiving a request for a target object; and carrying out interactive processing of the federal learning service with the target object according to the request. The scheme of the invention can realize that the terminal opening the federal learning service carries out signaling interaction and service management; meanwhile, the advantages of the existing IP multimedia subsystem network and data network of an operator and the advantages of the vast terminal clients are fully exerted.

Description

Processing method, device, equipment and system for federated learning service

Technical Field

The present invention relates to the field of communications technologies, and in particular, to a method, an apparatus, a device, and a system for processing a federated learning service.

Background

Federal Learning (Federal Learning) is an emerging artificial intelligence basic technology, and aims to carry out efficient machine Learning among multiple parties or multiple computing nodes on the premise of guaranteeing information safety during big data exchange, protecting terminal data and personal data privacy and guaranteeing legal compliance.

Federal learning is now being applied to B2C applications, mainly for deployment on mobile devices trained in collaborative machine learning models, such as google input methods. With federal learning for mobile terminal applications, federal learning has been applied to a variety of occasions, but the development of federal learning is still in the infancy and still faces many challenges.

The development of federal learning faces a number of challenges, such as at least one of the following:

communication overhead and communication latency are challenges faced by federal learning systems;

many enterprises cannot develop federal learning applications because of the difficulty in collecting and sharing data;

in the data sharing and data use scenes of cross-domain and cross-enterprise, federal learning has great advantages and commercial application potential, but a bridge connecting all parties is lacked, namely, no collaborator is lacked.

The operators have inherent advantages when solving the problems, but no operators are involved in the field to develop the federal learning business at present.

Disclosure of Invention

The technical problem to be solved by the invention is how to provide a method, a device, equipment and a system for processing the federated learning service. The scheme of the invention can meet the requirements of operators for providing the federal learning service and managing the federal learning service, fully exert the function of data assets on the premise of protecting the personal privacy of terminal users, and realize the win-win of users, enterprises and operators.

In order to solve the technical problems, the technical scheme of the invention is as follows:

a processing method of a federated learning service is applied to a federated learning service application server, and comprises the following steps:

receiving a request of a first target object through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and according to the request, carrying out interaction processing of the federal learning service with the target object.

Optionally, the receiving, by the IMS network, a request for the first target object according to a trigger rule of the IMS network includes:

and receiving a request of a terminal according to a trigger rule of a Call Session Control Function (CSCF) network element of the IMS network, wherein the trigger rule is used for positioning the request of the terminal to the Federal learning service application server.

Optionally, according to the request, performing interaction processing of a federal learning service with the target object includes:

the CSCF network element receives a registration request of a terminal, and initiates a signing request to a Home Subscriber Server (HSS) of the IMS network according to the registration request;

receiving the subscription information of the terminal returned by the HSS;

and after registering the federal learning service of the terminal according to the subscription information of the terminal, returning a registration completion message to the CSCF.

Optionally, after the registration completion message is returned to the CSCF, the method further includes:

the CSCF network element receives a subscription request message of a terminal; the head field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

feeding back a confirmation message to the terminal through the CSCF network element according to the subscription request;

after confirming that the terminal opens the federal learning service, feeding back a successful subscription message to the terminal through the CSCF network element, wherein a subscription event field of a header field of the successful subscription message is the same as a subscription event field of a header field of the subscription request message, a subscription state field of the successful subscription message is a first value, and the first value indicates that the subscription is successful.

Optionally, the second target object is a third-party service platform, and receiving a request of the second target object includes:

and receiving a federal learning request sent by the third-party service platform, and processing the federal learning request to generate a federal learning model.

sending a first notification message to the terminal according to the request, wherein the first notification message is used for notifying the terminal that a new federal learning task can learn;

the receiving terminal finishes the downloading of the federal learning model according to the first notification message and carries out model training to obtain and send a training result;

obtaining an aggregation result according to the training result;

and after sending a second notification message to the terminal, sending the aggregation result to the third-party service platform.

Optionally, a subscription event field of a header field of the first notification message is the same as the subscription event field in the subscription request message;

a subscription status field of a header field of the first notification message is a second value, and the second value represents that a federal learning task is in progress;

the content type field of the first notification message includes: the download address of the federal learning model and/or the level of the federal learning task.

Optionally, a subscription event field of a header field of the second notification message is the same as a subscription event field in the subscription request;

and the subscription status field of the header field of the second notification message is a third value, and the third value indicates that the federal learning task is completed.

The embodiment of the invention also provides a processing method of the federated learning service, which is applied to a terminal and comprises the following steps:

sending a request of federal learning service interaction to a federal learning service application server through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server;

and receiving information fed back by the Federal learning service application server according to the request.

Optionally, sending a request for federation learning service interaction to a federation learning service application server through an IMS network according to a trigger rule of the IP multimedia subsystem IMS network, where the request includes:

sending a registration request to a CSCF network element of the IMS network; and sending the registration request to a Federal learning service application server through a trigger rule of a CSCF network element of the IMS network.

Optionally, after the registration is completed, the method further includes:

sending a subscription request message to a CSCF network element of the IMS network; and the head field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

Optionally, the receiving information fed back by the federal learning service application server according to the request includes:

receiving a confirmation message and a successful subscription message fed back by the CSCF network element; and the subscription event field of the header field of the subscription success message is the same as the subscription event field of the header field of the subscription request message, the subscription state field of the subscription success message is a first value, and the first value indicates that the subscription is successful.

receiving a first notification message sent by the CSCF network element, wherein the first notification message is used for notifying a terminal that a new federated learning task can learn;

downloading a federal learning model according to the first notification message;

carrying out model training according to the federal learning model to obtain a training result, and sending the training result to the federal learning service application server through the CSCF network element;

receiving an aggregation result fed back by the federal learning service application server through the CSCF network element;

and receiving a second notification message sent by the Federal learning service application server through the CSCF network element.

Optionally, downloading the federal learning model according to the first notification message includes:

and downloading the federal learning model from the federal learning service application server through a wireless network or an operator data network according to the first notification message.

The embodiment of the invention also provides a processing device of the federated learning service, which is applied to a federated learning service application server, and the device comprises:

a receiving and sending module, configured to receive, via an IP multimedia subsystem IMS network, a request for a first target object according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and the processing module is used for carrying out interactive processing of the federal learning service with the target object according to the request.

The embodiment of the invention also provides a federated learning service application server, which comprises:

a transceiver for receiving a request for a first target object via an IP multimedia subsystem, IMS, network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and the processor is used for performing interactive processing of the federal learning service with the target object according to the request.

The embodiment of the invention also provides a processing device of the federated learning service, which is applied to a terminal and comprises the following steps:

the system comprises a receiving and sending module, a processing module and a processing module, wherein the receiving and sending module is used for sending a request of federal learning service interaction to a federal learning service application server through an IP Multimedia Subsystem (IMS) according to a trigger rule of the IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server; and receiving the information fed back by the Federal learning service application server according to the request.

An embodiment of the present invention further provides a terminal, including:

the system comprises a transceiver and a federated learning service application server, wherein the transceiver is used for sending a request of federated learning service interaction to the federated learning service application server through an IP Multimedia Subsystem (IMS) according to a trigger rule of the IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server; and receiving information fed back by the Federal learning service application server according to the request.

An embodiment of the present invention further provides a communication device, including: a processor, a memory storing a computer program which, when executed by the processor, performs the method as described above.

Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method as described above.

The scheme of the invention at least comprises the following beneficial effects:

by receiving a request for a target object; and carrying out interactive processing of the federal learning service with the target object according to the request. Carrying out signaling interaction and service management with a terminal for opening the federal learning service; the advantages of the IMS network and the data network of the existing IP multimedia subsystem of the operator and the advantages of the vast terminal clients are fully exerted, and the operator is helped to preempt in the emerging field of federal learning; the method fully plays the role of data assets on the premise of protecting the personal privacy of the terminal user, and realizes the win-win of the user, the enterprise and the operator.

Drawings

FIG. 1 is a flow chart illustrating a method for processing a federated learning service in accordance with an embodiment of the present invention;

FIG. 2 is a flow diagram of federated learning service registration in accordance with an embodiment of the present invention;

FIG. 3 is a flow diagram of a federated learning service subscription according to an embodiment of the present invention;

FIG. 4 is a flow diagram illustrating the execution of a federated learning task in accordance with an embodiment of the present invention;

FIG. 5 is a general schematic diagram of a method of processing federated learning services in an embodiment of the present invention;

FIG. 6 is a flow chart illustrating a method for processing federated learning services in accordance with an embodiment of the present invention;

FIG. 7 is a flow chart illustrating a method for processing federated learning services in accordance with an embodiment of the present invention;

fig. 8 is a block diagram of a processing apparatus of the federal learning service according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

As shown in fig. 1, an embodiment of the present invention provides a method for processing a federated learning service, which is applied to a federated learning service application server, and the method includes:

step 11, receiving a request of a target object; specifically, a request of a first target object is received through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and 12, carrying out interaction processing of the federal learning service with the target object according to the request.

The embodiment is a processing method of the federal learning service applied to a federal learning service application server; the target object specifically comprises a third-party service platform and a terminal, and interaction processing of the federal learning service is carried out according to a request; the method can meet the requirements of operators for providing the federal learning service and managing the federal learning service.

It should be noted that the federal learning service application server is used as a cooperative party in the federal learning service, and is used for realizing the management of the federal learning service and processing the service requirements and models provided by a third-party service platform; the method comprises the steps of finishing service registration and service management interaction on a terminal, and providing data required by the federal learning service and feedback data of a receiving terminal for the terminal; the federal learning service application server performs signaling interaction with the terminal through an IMS network (an IP multimedia subsystem), and performs service data interaction with the terminal through an operator data network or a Wi-Fi (wireless fidelity) network; in addition, in particular embodiments, data traffic charging may be exempted from service charging rules when data is exchanged over the operator data network.

And the federal learning service application server and the terminal complete the federal learning service registration and other signaling interaction through the IMS network. After the terminal signs the federal learning service, when the terminal is started to register the IMS network, the IMS network preferably triggers through an iFC triggering rule and finishes the registration to the federal learning service application server. After the registration is completed, signaling interaction between the terminal and the federal learning service AS, such AS subscription request information SUBSCRIBE, confirmation information NOTIFY, etc., is completed by the IMS network. The IMS network does not need to be modified, and only an iFC trigger rule needs to be configured on a CSCF network element, so that the CSCF network element can send a signaling to a Federal learning service application server when registering and processing a related signaling, and normal interaction of the process is completed.

The first target object is a terminal, and step 111 includes:

In this embodiment, a CSCF (Call Session Control Function) network element is a functional entity inside an IMS network (IP Multimedia Subsystem), and is a core of the entire IMS network. Receiving a request of a terminal according to a trigger rule of a CSCF network element, where the trigger rule is used to locate the request of the terminal to a federal learning service application server, and it should be noted that the trigger rule of the CSCF network element is preferably an Initial Filter Criteria (iFC).

In an optional embodiment of the present invention, the receiving a request from a terminal includes:

and the CSCF network element receives a registration request of a terminal.

In an optional embodiment of the present invention, the performing, according to the request of the terminal, an interactive processing of a federated learning service with the target object includes:

initiating a signing request to a Home Subscriber Server (HSS) of the IMS network according to the registration request;

receiving the subscription information of the terminal returned by the HSS;

In the embodiment, an HSS (Home Subscriber Server) is used for receiving a signing request of a federal learning service application Server and sending signing information of a terminal to the federal learning service application Server; after receiving the subscription information of the terminal, the federal learning service application server completes register of the federal learning service of the terminal, and returns a register completion message to the CSCF network element, so as to implement the register process of the federal learning service, which is further described in the following by a specific embodiment 1:

as shown in fig. 2, the terminal, the CSCF network element, and the HSS together complete the procedure of federal learning service registration, and the specific steps are as follows:

step a1, a terminal initiates a network attachment request, initiates IMS network registration after completing EPC (Evolved Packet Core;4G Core network)/5 GC (5G Core network) authentication, and a CSCF network element completes the authentication of the terminal;

step a2, the CSCF network element initiates third-party registration to each application server according to the trigger rule analysis of the initial filter criterion iFC; when the terminal signs a federal learning service, the CSCF network element initiates third-party registration to the federal learning service application server and sends a registration message;

step a3, the federal learning service application server sends a request of UDR (Unified Data repository function) to HSS after receiving the registration message, and the HSS returns the related signing information to the federal learning service application server through UDA (Unified Data repository function);

and a4, the Federal learning service application server completes the Federal learning service registration of the terminal and returns a registration completion message to the CSCF network element.

In this embodiment, the trigger rule configured by the IMS network at the CSCF network element is an iFC trigger rule, so that the CSCF network element can trigger the federal learning service application server to complete registration when registering.

In an optional embodiment of the present invention, after returning the registration completion message to the CSCF, the method further includes:

the CSCF network element receives a subscription request message of a terminal; the header field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

In an alternative embodiment of the present invention, step 12 includes:

after confirming that the terminal opens the federal learning service, feeding back a successful subscription message to the terminal through the CSCF network element, wherein a subscription event field of a header field of the successful subscription message is the same as a subscription event field of a header field of the subscription request message, a subscription state field of the successful subscription message is a first value, and the first value indicates that subscription is successful.

In this embodiment, after the registration of the federal learning service is completed, the CSCF network element receives the subscription request information of the terminal, and feeds back confirmation information to the terminal through the CSCF network element, and after the terminal is confirmed to open the federal learning service, feeds back a successful subscription message to the terminal through the CSCF network element, thereby implementing a federal learning service subscription process. This is further illustrated by a specific example 2 below:

as shown in fig. 3, the terminal, the CSCF network element, and the federal learning service application server together complete the federal learning service subscription process, and the specific steps are as follows:

step b1, after the terminal completes the registration process of the federal learning service, the terminal initiates subscription request information SUBSCRIBE to the CSCF network element;

step b2, after receiving the subscription request message SUBSCRIBE, the CSCF network element performs trigger analysis according to the trigger rule of the initial filter criterion iFC, and forwards the subscription request message to the Federal learning service application server;

b3, after successfully receiving the SUBSCRIBE request message SUBSCRIBE, the federal learning service application server replies a confirmation message to the terminal;

b4, the Federal learning service application server confirms that the terminal opens the Federal learning service, and then replies a subscription success message NOTIFY to confirm that the subscription is successful; the successful subscription message NOTIFY is sent to the terminal through the CSCF network element;

and b5, after the terminal successfully receives the subscription success message NOTIFY, returning a response message to the Federal learning service application server.

In this embodiment, the trigger rule configured by the IMS network at the CSCF network element is an iFC trigger rule, so that the CSCF network element can send a subscription request message SUBSCRIBE to a federal learning service application server when processing the message;

it should be noted that the subscription Event field "Event" in the header field of the subscription request message SUBSCRIBE indicates that this message is a subscription request message for applying for the federal learning task, that is, the Event header field is: "Event: federatedLearning"; confirming that an Event header field in the information NOTIFY is consistent with an Event header field in the subscription request message SUBSCRIBE; and confirming that the description-State header field in the information NOTIFY is 'active', which indicates that the Subscription is successful.

In an optional embodiment of the present invention, the second target object is a third party service platform, and step 112 includes:

and receiving a federal learning request sent by the third-party service platform, processing the federal learning request and generating a federal learning model.

In an optional embodiment of the present invention, the performing, according to the request of the third-party service platform, interaction processing of a federal learning service with the target object includes:

obtaining an aggregation result according to the training result;

In this embodiment, the third-party service platform may be a service platform owned by an operator, or may be a third-party enterprise that provides application services to a terminal or serves as a federal study participant.

The first notification message is used for notifying a terminal that a new federal learning task can learn, and further, a subscription event field of a header field of the first notification message is the same as a subscription event field in the subscription request message; the subscription state field of the header field of the first notification message is a second value, and the second value represents that a federal learning task is in progress; the content type field of the first notification message includes: the download address of the federal learning model and/or the level of the federal learning task.

The second notification message is used for notifying the terminal that the federal learning task is finished, and further, a subscription event field of a header field of the second notification message is the same as a subscription event field in the subscription request; and the subscription status field of the header field of the second notification message is a third value, and the third value indicates that the federal learning task is completed.

This is further illustrated by a specific example 3:

as shown in fig. 4, the process of executing the federal learning task is completed by the terminal, the CSCF network element, the federal learning service application server, and the third party service platform, and the specific steps are as follows:

step c1, a third-party service platform provides a federal learning task request, and a federal learning service application server processes the request to generate a federal learning model capable of being issued to a terminal;

step c2, the federal learning service application server sends a first notification message NOTIFY to the terminal, wherein the first notification message NOTIFY is used for notifying the terminal of a new federal learning task and informing an address for downloading a federal learning model; it should be noted that:

an Event header field in the first notification message NOTIFY is consistent with an Event header field in the subscription request message SUBSCRIBE;

the description-State header field in the first notification message NOTIFY is "pending", which indicates that the federated learning task is in progress;

the Content-Type message body comprises an address for downloading the federal learning model, and the address points to the federal learning service application server or is connected with the federal learning service application server;

the Content-Type message body can contain the level of the learning task for reference when the terminal downloads and uploads data;

step c3, the first notification message NOTIFY is sent to the terminal through the CSCF network element, and the terminal returns a response message to the federal learning service application server after successfully receiving the first notification message NOTIFY;

step c4, the terminal initiates a model downloading request according to the address for downloading the Federal learning model in the message body of the Content-Type, and downloads the Federal learning model data; it should be noted that, the terminal will preferentially use Wi-Fi connection to download data, if there is no Wi-Fi connection, it can determine whether to use the operator data network to download data according to the level of the learning task; if the data is downloaded through the operator data network, the federal learning service application server can distinguish or reduce or avoid the part of data traffic according to the service rule;

c5, after the terminal finishes model downloading, executing local model training; after the training is finished, masking the gradient information by using an encryption technology, and uploading the masked result to a Federal learning service application server; and the federal learning service application server aggregates the learning results of the terminal and sends the aggregated results to the terminal, and the terminal updates the model parameters after receiving the aggregated results. The process from downloading to training to uploading to aggregating to downloading and updating the model parameters can be executed for multiple times and is carried out iteratively;

step c6, after the training task is completed, the federal learning service application server sends a second notification message NOTIFY to the terminal to NOTIFY the terminal that the federal learning task is finished, and what needs to be explained is that:

the Event header field in the second notification message NOTIFY is consistent with the Event header field in the subscription request message SUBSCRIBE;

the description-State header field in the second notification message NOTIFY is "terminated", indicating that the task is completed;

the second notification message NOTIFY is sent to the terminal through the CSCF network element, and the terminal returns a response message to the Federal learning service application server after successfully receiving the second notification message NOTIFY; the terminal obtains the final model parameters and provides the final model parameters for related applications;

and c7, the Federal learning service application server sends the final aggregation result to a third-party service platform according to service requirements.

As shown in FIG. 5, the processing method of the federal learning service of the scheme of the invention is to realize the processing of the federal learning service by the terminal of the participant, the federal learning application server of the cooperative party and the third-party service platform together, wherein the terminal and the federal learning application server can carry out signaling interaction and service management to the IMS Network, the ERC/5GC to the CMNET (China Mobile Internet) and directly through the CMNET, thereby fully playing the advantages of the existing IMS Network and data Network of the operator and the advantages of the vast terminal clients, and being beneficial to the operator to preempt the preempting in the new field of federal learning; meanwhile, by developing the federal learning service, the data asset function is fully played on the premise of protecting the personal privacy of the terminal user, and the win-win situation of the user, the enterprise and the operator is realized.

As shown in fig. 6, the present invention provides a method for processing a federated learning service, which is applied to a terminal, and the method includes:

step 61, according to the trigger rule of the IP multimedia subsystem IMS network, sending a request of federal learning service interaction to a federal learning service application server through the IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server;

and step 62, receiving the information fed back by the federal learning service application server according to the request.

The embodiment is a processing method applied to the federal learning service of a terminal; the terminal is used as a party participating in federal learning in the federal learning service and participates in a federal learning task. The terminal opens the federal learning service, and user experience of related applications can be optimized through the federal learning model; and optimizing the model by using the local data resources on the premise of not revealing personal privacy data. When the business is developed, the corresponding terminal can be considered to provide additional business service or tariff reward according to business rules such as contribution ratio in the federal learning task.

It should be noted that, after the IMS network registration and the federal learning service registration are completed, the terminal initiates a SIP Protocol (Session Initiation Protocol) subscription request information SUBSCRIBE process to the federal learning service application server, the federal learning service application server carries an address for downloading the federal learning model in a subscription success message NOTIFY returned to the terminal, the terminal downloads the federal learning model data from the address, trains the model by using local data, and uploads the updated content of the model. The Federal learning service application server fuses the model data of each terminal, optimizes the prediction model, and then sends the updated model to the terminal, so that the terminal completes model optimization and update, and the process is repeated continuously. In the whole process, the terminal sensitive data are always stored locally, and the risk of data leakage does not exist. After the model is optimized, the result can be applied to relevant applications on the terminal, and user experience is improved. In addition, in a specific embodiment, the terminal needs to be functionally modified to support the protocol processing flow.

sending a registration request to a CSCF network element of the IMS network; and sending the registration request to a federated learning service application server through a trigger rule of a Call Session Control Function (CSCF) network element of the IMS network.

Optionally, after the registration is completed, the method further includes:

receiving a confirmation message and a successful subscription message fed back by the CSCF network element; the subscription event field of the head domain of the subscription success message is the same as the subscription event field of the head domain of the subscription request message, the subscription state field of the subscription success message is a first value, and the first value indicates that the subscription is successful.

model training is carried out according to the federal learning model, a training result is obtained, and the training result is sent to the federal learning service application server through the CSCF network element;

and downloading the federal learning model from the federal learning service application server through a wifi network or an operator data network according to the first notification message.

It should be noted that the method of the terminal is a method corresponding to the method of the federal learning service application server, and all implementation manners in the method embodiment of the federal learning service application server are applicable to the embodiment of the method of the terminal, so that the same technical effect can be achieved.

As shown in fig. 7, the present invention further provides a method for processing a federated learning service, which is applied to a third-party service platform, and the method includes:

step 71, sending a federal learning request to a federal learning service application server;

and 72, receiving a learning result fed back by the federal learning service application server according to the federal learning request.

The embodiment is a processing method of a federal learning service applied to a third-party service platform; the third-party service platform may be a service platform owned by an operator, or a third-party enterprise providing application services to a terminal, or serving as a federal learning participant. The third-party service platform is in butt joint with the federal learning service application server, and provides a federal learning requirement or a federal learning model for the federal learning service application server, and the federal learning service application server checks and processes the federal learning model. After the model is checked and processed, the Federal learning service application server initiates a NOTIFY process of SIP protocol second notification information to the terminal, starts a Federal learning task, and feeds back the aggregated machine learning result to an enterprise demand side. Wherein, the third-party service platform should support the interface with the federal learning service AS.

Optionally, the learning result is an aggregation result obtained by the federal learning service application server according to a training result after the federal learning model is trained according to a terminal.

It should be noted that the method of the third-party service platform is a method corresponding to the method of the federal learning service application server and the method of the terminal, and all implementation manners in the embodiments of the method of the federal learning service application server and the method of the terminal are applicable to the embodiments of the method of the third-party service platform, and the same technical effect can be achieved.

As shown in fig. 8, an apparatus 80 for processing a federated learning service is applied to a federated learning service application server, the apparatus 80 includes:

a transceiving module 81 for receiving a request of a target object; specifically, a request of a first target object is received through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and the processing module 82 is used for performing interaction processing of the federal learning service with the target object according to the request.

receiving the subscription information of the terminal returned by the HSS;

Optionally, after returning a registration completion message to the CSCF, the CSCF network element receives a subscription request message of a terminal; the head field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

obtaining an aggregation result according to the training result;

It should be noted that the apparatus is an apparatus corresponding to the method, and all implementation manners in the method embodiments are applicable to the embodiment of the apparatus, and the same technical effects can be achieved.

a transceiver for receiving a request for a target object; specifically, a request of a first target object is received through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

and the processor is used for carrying out interaction processing of the federal learning service with the target object according to the request.

receiving the subscription information of the terminal returned by the HSS;

Optionally, after returning the registration completion message to the CSCF, the method further includes:

obtaining an aggregation result according to the training result;

Optionally, the subscription event field of the header field of the second notification message is the same as the subscription event field in the subscription request;

It should be noted that the federal learning service application server is a federal learning service application server corresponding to the method, and all implementation manners in the method embodiment are applicable to the embodiment of the federal learning service application server, and the same technical effect can be achieved.

The invention provides a processing device of a federated learning service, which is applied to a terminal and comprises the following steps:

Optionally, after the registration is completed, the method further includes:

sending a subscription request message to a CSCF network element of the IMS network; the header field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

Optionally, the receiving information fed back by the federal learning service application server according to the request includes: receiving a confirmation message and a successful subscription message fed back by the CSCF network element; and the subscription event field of the header field of the subscription success message is the same as the subscription event field of the header field of the subscription request message, the subscription state field of the subscription success message is a first value, and the first value indicates that the subscription is successful.

Optionally, the receiving information fed back by the federal learning service application server according to the request includes: receiving a first notification message sent by the CSCF network element, wherein the first notification message is used for notifying a terminal that a new federated learning task can learn;

An embodiment of the present invention further provides a terminal, including:

Optionally, after the registration is completed, the method further includes:

Optionally, the receiving information fed back by the federal learning service application server according to the request includes: receiving a confirmation message and a successful subscription message fed back by the CSCF network element; the subscription event field of the head domain of the subscription success message is the same as the subscription event field of the head domain of the subscription request message, the subscription state field of the subscription success message is a first value, and the first value indicates that the subscription is successful.

Optionally, the receiving information fed back by the federal learning service application server according to the request includes: receiving a first notification message sent by the CSCF network element, wherein the first notification message is used for notifying a terminal that a new federal learning task can learn;

It should be noted that the terminal is a terminal corresponding to the method, and all the implementation manners in the above method embodiment are applicable to the embodiment of the terminal, and the same technical effect can be achieved.

The invention also provides a processing device of the federated learning service, which is applied to a third-party service platform, and the device comprises:

the receiving and sending module is used for sending a federal learning request to the federal learning service application server; and receiving a learning result fed back by the Federal learning service application server according to the Federal learning request.

Optionally, the learning result is an aggregation result obtained by the federal learning service application server according to the training result after the federal learning model is trained by the terminal.

It should be noted that the apparatus is an apparatus corresponding to the method, and all the implementations in the embodiments of the method are applicable to the embodiments of the apparatus, and the same technical effects can be achieved.

The embodiment of the present invention further provides a third party service platform, including:

a transceiver for sending a federal learning request to a federal learning service application server; and receiving a learning result fed back by the Federal learning service application server according to the Federal learning request.

It should be noted that the third-party service platform is a third-party service platform corresponding to the method, and all implementation manners in the embodiment of the method are applicable to the embodiment of the third-party service platform, and the same technical effect can be achieved.

The invention provides a processing system of a federated learning service, which comprises: a federal learning business application server;

the federal learning service application server receives a request of a target object; according to the request, carrying out interaction processing of a federal learning service with the target object; the target object is a terminal and/or a third service platform.

Optionally, receiving a request of a target object includes:

receiving a request of a first target object through an IP Multimedia Subsystem (IMS) network according to a trigger rule of the IMS network; and/or receiving a request for a second target object.

receiving the subscription information of the terminal returned by the HSS;

the receiving terminal finishes the Federal learning model downloading according to the first notification message and performs model training to obtain and send a training result;

obtaining an aggregation result according to the training result;

the content type field of the first notification message includes: the download address of the federal learning model and/or the level of the present federal learning task.

It should be noted that, the system is similar to the above method system, and all the implementation manners in the above method embodiment are applicable to the embodiment of the system, and the same technical effect can be achieved.

An embodiment of the present invention further provides a communication device, including: a processor, a memory storing a computer program which, when executed by the processor, performs the method as described above. All the implementation manners in the method embodiment are applicable to the embodiment, and the same technical effect can be achieved.

Embodiments of the present invention also provide a computer-readable storage medium comprising instructions that, when executed on a computer, cause the computer to perform the method as described above. All the implementation manners in the above method embodiment are applicable to this embodiment, and the same technical effect can be achieved.

The method and the system for the federal learning service carry out signaling interaction and service management with the terminal of the open federal learning service by utilizing the existing IMS network; the advantages of the existing IMS network and the existing data network of the operator and the advantages of the vast terminal customers are fully exerted, and the operator is helped to preempt in the emerging field of federal study; by developing the federal learning service, the data asset function is fully played on the premise of protecting the personal privacy of the terminal user, and the win-win situation of users, enterprises and operators is realized.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the various embodiments described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed coupling or direct coupling or communication connection between each other may be through some interfaces, indirect coupling or communication connection between devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: u disk, removable hard disk, ROM, RAM, magnetic disk or optical disk, etc.

Furthermore, it is to be noted that in the device and method of the invention, it is obvious that the individual components or steps can be decomposed and/or recombined. These decompositions and/or recombinations are to be considered equivalents of the present invention. Also, the steps of performing the above-described series of processes may naturally be performed in chronological order in the order described, but need not necessarily be performed in chronological order, and some steps may be performed in parallel or independently of each other. It will be understood by those skilled in the art that all or any of the steps or elements of the method and apparatus of the present invention may be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or any combination thereof, which can be implemented by those skilled in the art using their basic programming skills after reading the description of the present invention.

The object of the invention is thus also achieved by a program or a set of programs running on any computing device. The computing device may be a general purpose device as is well known. The object of the invention is thus also achieved solely by providing a program product comprising program code for implementing the method or device. That is, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. It is to be understood that such storage media can be any known storage media or any storage media developed in the future. It is further noted that in the apparatus and method of the present invention, it is apparent that each component or step can be decomposed and/or recombined. These decompositions and/or recombinations are to be regarded as equivalents of the present invention. Also, the steps of executing the series of processes described above may naturally be executed chronologically in the order described, but need not necessarily be executed chronologically. Some steps may be performed in parallel or independently of each other.

The foregoing is a preferred embodiment of the present invention, and it should be noted that it is obvious to those skilled in the art that various modifications and improvements can be made without departing from the principle of the present invention, and these modifications and improvements should be construed as the protection scope of the present invention.

Claims

1. A processing method of a federated learning service is characterized in that the method is applied to a federated learning service application server, and the method comprises the following steps:

2. The method for processing a federal learning service as claimed in claim 1, wherein the first target object is a terminal, and the receiving of the request of the first target object through the IMS network according to the trigger rule of the IMS network comprises:

3. The processing method of the federal learning service of claim 2, wherein the interactive processing of the federal learning service with the target object according to the request comprises:

receiving the subscription information of the terminal returned by the HSS;

4. The method for processing federal learning service as claimed in claim 3, wherein after returning a registration complete message to the CSCF, the method further comprises:

5. The processing method of the federal learning business as claimed in claim 4, wherein the interactive processing of the federal learning business with the target object according to the request comprises:

6. The method for processing a federal learning service as claimed in claim 4 or 5, wherein the second target object is a third party service platform, and the receiving of the request of the second target object includes:

7. The method for processing the federal learning service as claimed in claim 6, wherein the interactive processing of the federal learning service with the target object according to the request comprises:

sending a first notification message to the terminal according to the request, wherein the first notification message is used for notifying the terminal that a new federated learning task can learn;

obtaining an aggregation result according to the training result;

8. The method of processing a federal learning service as in claim 7,

a subscription event field of a header field of the first notification message is the same as a subscription event field in the subscription request message;

the subscription state field of the header field of the first notification message is a second value, and the second value represents that a federal learning task is in progress;

9. The method of processing a federal learning service as in claim 7,

a subscription event field of a header field of the second notification message is the same as the subscription event field in the subscription request;

10. A processing method of a federated learning service is characterized in that the method is applied to a terminal, and the method comprises the following steps:

sending a request of federal learning service interaction to a federal learning service application server through an IMS network according to a trigger rule of an IP multimedia subsystem IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server;

and receiving the information fed back by the Federal learning service application server according to the request.

11. The method for processing a federal learning service as claimed in claim 10, wherein the sending a request for federal learning service interaction to a federal learning service application server through an IP multimedia subsystem IMS network according to a trigger rule of the IMS network comprises:

12. The method for handling a federal learning service as claimed in claim 11, further comprising, after the registration is completed:

sending a subscription request message to a CSCF network element of the IMS network; the head field of the subscription request message carries a subscription event field, and the subscription event field indicates that the subscription request message is a message for applying for a federal learning task.

13. The method for processing the federal learning service as claimed in claim 12, wherein the step of receiving the information fed back by the federal learning service application server according to the request comprises:

14. The method for processing federal learning service of claim 10, wherein the receiving of the information fed back by the federal learning service application server according to the request comprises:

receiving a first notification message sent by the CSCF network element, wherein the first notification message is used for notifying a terminal that a new federal learning task can learn;

15. The method of processing a federal learning service as claimed in claim 14, wherein downloading a federal learning model in response to the first notification message comprises:

16. The utility model provides a processing apparatus of federal study business, is applied to federal study business application server, the device includes:

and the processing module is used for carrying out interaction processing of the federal learning service with the target object according to the request.

17. A Federation learning service application server, comprising:

a transceiver for receiving a request for a first target object over an IP multimedia subsystem, IMS, network according to a trigger rule of the IMS network; and/or, receiving a request for a second target object;

18. The utility model provides a processing apparatus of bang study business which characterized in that, is applied to the terminal, the device includes:

the system comprises a receiving and sending module, a processing module and a processing module, wherein the receiving and sending module is used for sending a request of federal learning service interaction to a federal learning service application server through an IP Multimedia Subsystem (IMS) according to a trigger rule of the IMS network; the triggering rule is used for positioning a request of a terminal to the Federal learning service application server; and receiving information fed back by the Federal learning service application server according to the request.

19. A terminal, comprising:

20. A communication device, comprising: a processor, a memory storing a computer program which, when executed by the processor, performs the method of any of claims 1 to 9 or the method of any of claims 10 to 15.

21. A computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method of any one of claims 1 to 9 or the method of any one of claims 10 to 15.