WO2020259717A1 - 联邦学习系统的控制方法、装置、终端设备及存储介质 - Google Patents

联邦学习系统的控制方法、装置、终端设备及存储介质 Download PDF

Info

Publication number
WO2020259717A1
WO2020259717A1 PCT/CN2020/107327 CN2020107327W WO2020259717A1 WO 2020259717 A1 WO2020259717 A1 WO 2020259717A1 CN 2020107327 W CN2020107327 W CN 2020107327W WO 2020259717 A1 WO2020259717 A1 WO 2020259717A1
Authority
WO
WIPO (PCT)
Prior art keywords
connection
server
sub
central node
client
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.)
Ceased
Application number
PCT/CN2020/107327
Other languages
English (en)
French (fr)
Inventor
黄安埠
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
WeBank Co Ltd
Original Assignee
WeBank Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by WeBank Co Ltd filed Critical WeBank Co Ltd
Publication of WO2020259717A1 publication Critical patent/WO2020259717A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/06Management of faults, events, alarms or notifications
    • H04L41/0654Management of faults, events, alarms or notifications using network fault recovery
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/14Session management
    • H04L67/141Setup of application sessions
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/50Network services
    • H04L67/52Network services specially adapted for the location of the user terminal
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/104Peer-to-peer [P2P] networks
    • H04L67/1044Group management mechanisms 
    • H04L67/1046Joining mechanisms

Definitions

  • This application relates to the field of Fintech (financial technology) technology, and in particular to a control method, device, terminal device and storage medium of a federated learning system.
  • federated learning technology is based on the protection of user privacy and data security, and is gradually receiving more and more attention. Pay attention.
  • Federated learning refers to a method of machine learning modeling by uniting different participants (participants, or parties, also known as data owners or clients). In federated learning, participants do not need to expose their own data to other participants and coordinators (also called servers, parameter servers, or aggregation servers). Federated learning can well protect user privacy and ensure data security, and can solve the problem of data islands.
  • the main purpose of this application is to provide a control method, device, terminal device and storage medium of a federated learning system, aiming to avoid hierarchical horizontal federated learning, if any node fails, it cannot be used to establish a connection with the current node
  • the data owned by each participant is the technical problem of training the federated learning model to improve the overall efficiency of federated learning.
  • the present application provides a control method for a federated learning system.
  • the federated learning control method is applied to a federated learning system.
  • the federated learning system includes: a main central node server, a sub-central node server, and clients Terminal, wherein the client is connected to the sub-center node server, the sub-center node server is connected to the main center node server, and the sub-center node server and the client are Multiple,
  • the control method of the federated learning system includes the following steps:
  • each of the clients is controlled to establish a connection with each of the sub-center node servers, or each of the clients is controlled to directly establish a connection with the main central node server.
  • the present application also provides a control device for a federated learning system.
  • the federated learning control device is applied to a federated learning system.
  • the federated learning system includes: a main central node server, a sub-central node server, and a client, Wherein, the client is connected to the sub-central node server, the sub-central node server is connected to the main central node server, and the sub-central node server and the client are multiple ,
  • the control device of the federated learning system includes:
  • An obtaining module configured to obtain geographic location information of the main central node server, each of the sub-central node servers, and each of the clients;
  • the control module is configured to control each client to establish a connection with each of the sub-central node server based on the geographic location information, or control each of the clients to directly establish a connection with the main central node server.
  • Each functional module of the control device of the federated learning system implements the steps of the control method of the federated learning system as described above when running.
  • this application also provides a terminal device, the terminal device comprising: a memory, a processor, and a control program of a federated learning system stored on the memory and running on the processor, the federated learning system
  • the control program is executed by the processor, the steps of the control method of the federated learning system as described above are realized.
  • the present application also provides a storage medium applied to a computer.
  • the storage medium stores a control program of the federated learning system.
  • the control program of the federated learning system is executed by a processor, the federated learning system as described above is realized The steps of the control method.
  • the control method of the federated learning system of this application is applied to a federated learning system.
  • the federated learning system includes: a main central node server, a sub-central node server, and a client, wherein the client is connected to the sub-central node server,
  • the sub-center node server is connected to the main center node server, and there are multiple sub-center node servers and clients.
  • the control method of the federated learning system of this application obtains the main center node server and each sub-center Node server and geographic location information of each client; based on the geographic location information, control each client to establish a connection with each sub-center node server, or control each client to directly communicate with the The main central node server establishes a connection.
  • each client is calculated based on the collected geographic location information
  • the connection distance between the server of each sub-central node and the server of the main central node, and based on the connection distance, each client in the current federated learning system is controlled to establish a connection with each sub-central node server or directly connect with the main
  • the central node establishes a connection, and when any node (such as any one of the multiple sub-central node servers or the main central node server) fails, the client connected to the current node that has failed is connected to other similar nodes ,
  • any node such as any one of the multiple sub-central node servers or the main central node server
  • Figure 1 is a schematic structural diagram of a hardware operating environment involved in a solution of an embodiment of the present application
  • FIG. 2 is a schematic flowchart of the first embodiment of the control method of the federated learning system according to the application;
  • FIG. 3 is a schematic diagram of an application scenario in an embodiment of a control method for a federated learning system according to this application;
  • FIG. 4 is a schematic diagram of modules of the control device of the federal learning system of this application.
  • Fig. 1 is a schematic structural diagram of a hardware operating environment involved in a solution of an embodiment of the present application.
  • Fig. 1 can be a structural diagram of the hardware operating environment of the terminal device.
  • the terminal device in the embodiment of the present application may be a terminal device such as a PC and a portable computer.
  • the terminal device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002.
  • the communication bus 1002 is used to implement connection and communication between these components.
  • the user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface.
  • the network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).
  • the memory 1005 may be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as a magnetic disk memory.
  • the memory 1005 may also be a storage device independent of the foregoing processor 1001.
  • the structure of the terminal device shown in FIG. 1 does not constitute a limitation on the terminal device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange different components.
  • a memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a control program of the federated learning system.
  • the operating system is a program that manages and controls the hardware and software resources of the sample terminal equipment, and supports the operation of the control program of the federated learning system and other software or programs.
  • the terminal device shown in FIG. 1 is combined with other terminals to form a federated learning system.
  • the federated learning system includes at least one server and multiple clients.
  • the user interface 1003 is mainly used for Perform data communication with various terminals;
  • the network interface 1004 is mainly used to connect to the background server and perform data communication with the background server;
  • the processor 1001 can be used to call the control program of the federated learning system stored in the memory 1005 and perform the following operations:
  • each of the clients is controlled to establish a connection with each of the sub-center node servers, or each of the clients is controlled to directly establish a connection with the main central node server.
  • processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005 and execute the following steps:
  • processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005 and execute the following steps:
  • the client When it is detected that the current connection distance of the first target is the connection distance between the client and the sub-central node server, controlling the client to establish a connection with the sub-central node server;
  • the client when it is detected that the current first target connection distance is the connection distance between the client and the main central node server, control the client to directly establish the connection with the main central node server connection.
  • processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005 and execute the following steps:
  • the processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005, and to detect the server of the main central node, the server of each of the sub-central nodes, and each After the step of the connection distance between the clients, the following steps are performed:
  • server of the sub-center node If the server of the sub-center node is down, obtain the information between each target client connected to the server of the target sub-center node that is currently down and the server of each of the sub-center nodes and the server of the main center node. Connection distance
  • the processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005, and to control the target client corresponding to the second target connection distance and the corresponding child center node After the server establishes a connection or directly establishes a connection with the main central node server, the following steps are performed:
  • processor 1001 may also be used to call the control program of the federated learning system stored in the memory 1005 and execute the following steps:
  • FIG. 2 is a schematic flowchart of the first embodiment of the control method of the federated learning system of this application.
  • the embodiment of the application provides an embodiment of the control method of the federated learning system. It should be noted that although the logical sequence is shown in the flowchart, in some cases, the sequence shown here can be executed in a different order. Steps out or described.
  • the control method of the federated learning system of the embodiment of this application is applied to the above-mentioned terminal equipment.
  • the terminal device of the embodiment of this application may be a terminal device such as a PC, a portable computer, etc., which is not specifically limited here.
  • the control method of the federated learning system of this application Applied to a federated learning system, as shown in Figure 3, the control method of the federated learning system of this application is applied to an application scenario of the federated learning system.
  • the federated learning system shown in this scenario includes at least one main central node server and multiple sub-nodes. A central node server and multiple clients, wherein the client is connected to the sub-central node server, and the sub-central node server is connected to the main central node server.
  • Step S100 Obtain geographic location information of the main central node server, each of the sub-central node servers, and each of the clients.
  • the geographic location information of the server of the main central node, the server of each sub-center node, and the clients are obtained at the same time, and whether there is a client newly connected to the current federated learning system is always checked.
  • the geographic location information of the new client is obtained.
  • step S100 includes:
  • the geographic location information of the client is acquired.
  • the current federated learning system obtains the geographic location information of the main central node server, each sub-central node server, and each client.
  • LBS Location Based Service: geographic location information service technology
  • the geographic location information of the server of the main central node is obtained based on LBS (for example, the geographic coordinates of the server of the primary central node are obtained as: M main (x, y)), and then the geographic location of the server of each sub-central node is obtained one by one Location information (for example, the geographic coordinates of the server of each sub-center node are obtained as: Mi(x i , y i ), where i is an integer greater than or equal to zero), and each node connected to the server of each sub-center node is obtained one by one.
  • the geographic location information of the client terminal (the geographic coordinates of each client acquired are: Where j is an integer greater than or equal to zero).
  • control method of the federated learning system of the present application can also first obtain the geographic location information of the server of the main central node and the server of each sub-central node, and detect each client in the current federated learning system one by one Whether it has been connected to the federated learning system, and when it is detected that the current client is connected to the federated learning system, the geographic location information of the current client is obtained.
  • Step S200 based on the geographic location information, control each of the clients to establish a connection with each of the sub-central node server, or control each of the clients to directly establish a connection with the main central node server.
  • the geographic location information of the main central node server, each sub-central node server, and each client calculate the connection distance between the three one by one, and calculate the connection distance between the three.
  • the distance value of the line distance controls each client in the current hierarchical horizontal federated learning system to establish a connection with one of the sub-central node servers, or directly establish a connection with the main central node server.
  • each sub-central node server after obtaining the geographic location information (ie geographic location coordinates) of the main central node server, each sub-central node server, and each client based on LBS, calculate the relationship between each client and each sub-central node server one by one.
  • geographic location information ie geographic location coordinates
  • connection distance between each client and the server of the main central node Calculate the connection distance between each client and the server of the main central node, and detect the minimum value of the distance between each client and the server of each sub-node, so as to follow
  • This connection controls the connection between each client and the server of each sub-central node, or the distance value of the connection distance between the client's main central node and server is detected, which is greater than the current client and a certain sub-node server If the distance between the connection distances is even smaller, the current client is controlled to directly connect with the main central node server.
  • step S200 includes:
  • Step S201 Detect the connection distance between the server of the main central node, the server of each sub-central node, and the clients according to the geographic location information.
  • each sub-center node server, and each client based on LBS calculate the distance between each client and each client one by one according to the calculation formula of the coordinate distance between two points in the plane.
  • Step S202 Detect a first target connection distance that meets a preset condition among the connection distances between each of the client and each of the sub-center nodes and the server of the main center node.
  • the preset condition is that the distance value of the connection distance between the current client and the server of each sub-central node or between the service of the main central node is the smallest.
  • the connection distance between each client and the server of each sub-center node is calculated one by one to obtain the distance value: And calculate the connection distance between each client and the main central node server and get the distance value: After D j , extract the connection between one of the clients and each sub-central node server one by one Distance values of distance: And detect the smallest distance value among the extracted distance values: Then, obtain the distance value of the connection distance between the current client and the main central node server: D j , and further compare the current minimum distance value: Is it less than the distance value: D j .
  • Step S203 controlling the client corresponding to the first target connection distance to establish a connection with the corresponding sub-central node server or directly establish a connection with the main central node server.
  • the distance value that meets the preset conditions identifies the first target
  • the connection corresponding to the connection distance controls each client to establish a connection with each sub-central node server or directly with the main central node server.
  • step S203 includes:
  • Step S2031 When it is detected that the current connection distance of the first target is the connection distance between the client and the server of the sub-central node, control the client to establish a connection with the server of the sub-central node connection.
  • each distance value of the connection distance between the current client and each sub-center node server is extracted And detect the smallest distance value among the extracted distance values And further detect when the minimum distance value Get the current distance value when it is less than the distance value D j of the connection distance between the current client and the main central node server According to the corresponding connection, control the current client to establish a connection with the sub-center node server on the connection.
  • Step S2032 when it is detected that the current first target connection distance is the connection distance between the client and the main central node server, control the client to directly communicate with the main central node server establish connection.
  • each distance value of the connection distance between the current client and each sub-center node server is extracted And detect the smallest distance value among the extracted distance values And further detect that the distance value D j of the connection distance between the current client and the main central node server is less than the current minimum distance value When, obtain the connection corresponding to the current distance value D j , and control the current client to establish a connection with the main central node server according to the connection.
  • control method of the federal learning system of this application also includes:
  • Step A dynamically detecting the connection distance of the first target in each connection distance, and executing the control of the client corresponding to the connection distance of the first target and the corresponding child center node
  • the server establishes a connection or directly establishes a connection with the main central node server.
  • the connection distance between each client and the main central node server and each sub-central node server is always checked. Dynamically monitor, and according to the first target connection distance that meets the preset conditions among the detected connection distances, control the client corresponding to the first target connection distance to establish a connection with the corresponding sub-center node server or Establish a connection directly with the server of the main central node. Therefore, in the process of the client's geographic location change, the connection line with the shortest connection distance between the client and the sub-central node server or the main central node server is always used to connect the client to the current joint learning system, which improves The overall performance of the federated learning system.
  • connection distance between the central node servers to obtain the distance value: And after calculating the connection distance between each client and the server of the main central node and obtaining the distance value D j , continue to calculate the distance value of the connection distance between each client and the server of each sub-central node
  • the connection distance between each client and the server of the main central node and the minimum distance value in the distance value D j is obtained minD j to monitor and always follow the minimum distance value detected Or the minimum distance value D j to control the minimum distance value between each client
  • the corresponding connection's sub-central node server establishes a connection, or a connection with the main central node server of the connection corresponding to the minimum distance value D j .
  • control method of the federal learning system of this application further includes:
  • Step S204 If the server of the sub-center node is down, obtain the relationship between each target client connected to the server of the target sub-center node that is currently down and each of the other sub-center node servers and the main center node server. The distance between the connections.
  • the working status of the main central node server, each sub-central node server, and each client is monitored at all times.
  • the machine that is, a failure
  • the target client connected to the server of the target child center node that has failed and the other child center nodes of all the child center node servers in the current horizontal federated learning system
  • the connection distance between the server and the server of the main central node is the maximum distance between the server and the server of the main central node.
  • the sub-center node server 4 out of the sub-center node server terminals 1 to 5 is down (that is, there is a failure)
  • the sub-center node server 4 obtains one by one the clients 5 to 7 that have established connections with the current target sub-center node server 4, and connect with the other sub-center node servers in the current horizontal federated learning system (that is, the sub-center node server 1 to 3 and the sub-center node service The distance between terminals 4) and the connection distance between the main central node server in the current horizontal federated learning system.
  • Step S205 Select a second target connection distance that meets a preset condition among the connection distances between each of the target clients and each of the sub-central nodes and the server of the main central node.
  • each distance value of each connection distance between terminal 4 is detected Minimum distance in Is the distance value of the connection distance between the current target client 5 and the sub-center node server 3, and the current minimum distance value Is less than the distance value D j , so that the connection distance between the current target client 5 and the sub-center node server 3 is determined as the second target connection distance.
  • Step S206 controlling the target client corresponding to the second target connection distance to establish a connection with the corresponding sub-central node server or directly establish a connection with the main central node server.
  • each target client is controlled to establish a connection with each target sub-center node server, or directly establish a connection with the main center node server.
  • the control target client 5 after determining the connection distance between the target client 5 and the sub-center node server 3 as the second target connection distance , According to the connection corresponding to the current second target connection distance, control the target client 5 to establish a connection with the child center node server 3, and determine the connection between the target client 6 and the main center node server based on the same operation described above
  • the connection distance between the target client 7 and the sub-center node server 5 is the second target connection distance, respectively according to the second target connection distance
  • the control target client 6 directly establishes a connection with the main central node server, and the control target client 7 establishes a connection with the child central node server 5.
  • This application obtains the geographic information of the main central node server, each sub-central node server, and each client in the current federal learning system based on LBS (Location Based Service) in a hierarchical horizontal federated learning system Location information, according to the obtained current hierarchical horizontal federated learning system, the geographic location information of the main central node server, each sub-central node server, and each client is calculated, and the connection distances between the three are calculated one by one.
  • LBS Location Based Service
  • the distance value that meets the preset conditions identifies the first target connection distance, and according to the connection corresponding to the first target connection distance, control each client and each sub-center node server or directly Establish a connection with the server of the main central node, and, in the hierarchical horizontal federated learning system, monitor the working status of the server of the main central node, the server of each sub-central node, and each client at all times.
  • each sub-center is detected When any one or more of the node servers are down (that is, there is a failure), the target client and all sub-nodes connected to the server of the failed target sub-central node in the current horizontal federated learning system are immediately obtained.
  • connection distance between the other sub-central node servers in the central node server and between the main central node server, and the distance value of each connection distance is detected, which meets the preset condition (ie the connection distance The distance value is the smallest) to identify the second target connection distance, and control each target client and each target sub-center node according to the connection corresponding to the second target connection distance identified by the distance value that meets the preset conditions
  • the server or directly establish a connection with the server of the main central node.
  • each client's subordinates are calculated based on the collected geographic location information.
  • the connection distance between the central node server and the main central node server, and based on the connection distance, each client in the current federated learning system is controlled with each sub-central node server or directly establishes a connection with the main central node ,
  • any node such as any one of the multiple sub-central node server or the main central node server
  • the participant connected to the current node that has the failure is connected to other similar nodes to avoid
  • the system will eliminate the data owned by the participants on the current node and cause data waste, and also avoid the entire federated learning system from being unable to use this part of the participants
  • the technical problem of training the federated learning model with the possessed data improves the performance of federated learning
  • control method of the federal learning system of this application further includes:
  • Step S207 after the target sub-center node server is restored, control each client to re-establish a connection with the target sub-center node server.
  • the working status of the main central node server, each sub-central node server, and each client is monitored at all times.
  • the target sub-center node server controls each client in the current federated learning system and re-establishes a connection with the current target sub-center node server, thereby reducing other sub-centers currently connected to each target client
  • the operating pressure of the node server further ensures the overall performance of the horizontal federated learning system.
  • step S207 includes:
  • the target sub-center node server 4 that detects a failure when the target sub-center node server 4 that detects a failure is detected to resume normal operation (for example, the staff completes the modification and the test runs normally) Re-check the connection distance between client 1 to client 9 in the current federated learning system and the restored target from the central node server 4, and extract the distance between client 1 to client 9 and the restored target. From the connection distance between the central node server 4, the first target connection distance with the smallest distance value, and then control the client 5 to 7 corresponding to the first target connection distance with the smallest distance value, and reconnect with the current The target child central node server 4 establishes a connection.
  • the server connected to the main central node in the current horizontal federated learning system is then controlled
  • the server and client of the sub-center node establish a connection with the server of the sub-center node closest to the server of the main center node, thereby avoiding the current entire federated learning system from being paralyzed due to the failure of the main center node server, and improving the horizontal The overall performance of the federated learning system.
  • the main central node server After detecting that the main central node server resumes normal operation, control the sub-central node server or client connected to the main central node server before the failure, and the main central node The server re-establishes the connection, thereby reducing the operating pressure on the server of the child center node closest to the server of the main center node, and further ensuring the overall performance of the horizontal federated learning system.
  • the sub-central node server 2 to 5 and the distance from the main central node are immediately controlled.
  • the nearest child center node server 1 of the server establishes a connection, and the child center node server 1 charges the temporary main center node server, and after detecting that the main center node server resumes normal operation, the child center node server 2 is controlled To 5 re-establish a connection with the server of the main central node.
  • This application controls each target client connected to the current target sub-center node server before the failure occurs when the target sub-center node server detects that the server of each sub-center node is down. , To re-establish the connection with the server of the current target sub-center node, thereby reducing the operating pressure of the server of the other sub-center nodes that are currently connected to each target client, and further ensuring the overall performance of the horizontal federated learning system; on the other hand , When it detects that the server of the main central node in the current horizontal federated learning system is down, then control the server and client of the sub-central node connected to the server of the main central node in the current horizontal federated learning system, and the distance from the main center The node server establishes a connection with the nearest child center node server, thus avoiding the entire federated learning system from being paralyzed due to the failure of the main center node server, and improving the overall performance of the horizontal federated learning system.
  • an embodiment of the present application also proposes a control device for a federated learning system.
  • the federated learning control device is applied to a federated learning system.
  • the federated learning system includes: a main central node server, a sub-center Node server and client, wherein the client is connected to the sub-center node server, the sub-center node server is connected to the main central node server, and the sub-center node server is connected to There are multiple clients,
  • the control device of the federated learning system includes:
  • An obtaining module configured to obtain geographic location information of the main central node server, each of the sub-central node servers, and each of the clients;
  • the control module is configured to control each client to establish a connection with the main central node server through each of the sub-central node server based on the geographic location information, or control each client to directly communicate with the main central node server.
  • the main central node server establishes a connection.
  • control module includes:
  • the first detection unit is configured to detect the connection distance between the server of the main central node, the server of each sub-central node, and the clients according to the geographic location information;
  • the second detection unit is used to detect the first target connection distance that meets a preset condition among the connection distances between each of the client and each of the sub-central nodes and the server of the main central node , Wherein the preset condition is that the distance value of the connection distance is the smallest;
  • the first control unit is configured to control the client corresponding to the first target connection distance to establish a connection with the corresponding sub-central node server or directly establish a connection with the main central node server.
  • the first control unit includes:
  • the first control subunit is configured to control the client and the sub-center node when it is detected that the current connection distance of the first target is the connection distance between the client and the server of the sub-center node
  • the central node server establishes a connection
  • the second control subunit is configured to control the client to directly communicate with the server when the current connection distance of the first target is detected as the connection distance between the client and the server of the main central node
  • the main central node server establishes a connection.
  • control module is further configured to: dynamically detect the first target connection distance in each of the connection distances, and execute the control of the client corresponding to the first target connection distance The step of establishing a connection with the corresponding sub-central node server or directly establishing a connection with the main central node server.
  • control module further includes:
  • the first obtaining unit is configured to obtain each target client connected to the currently down target sub-central node server and each of the other sub-central node servers and the main The connection distance between the central node servers;
  • a selecting unit configured to select a second target connection distance that meets a preset condition among the connection distances between each of the target clients and each of the sub-central nodes and the main central node server;
  • the second control unit is configured to control the target client corresponding to the second target connection distance to establish a connection with the corresponding sub-central node server or directly establish a connection with the main central node server.
  • control module further includes:
  • the third control unit is configured to control each client to re-establish a connection with the target sub-central node server after the target sub-central node server is restored.
  • the third control unit includes:
  • the third detection unit is configured to detect the connection distance between each of the clients and the server of the target sub-center node after restoration
  • the third control subunit is used to extract the first target connection distance in each of the connection distances, and control the client corresponding to the first target connection distance to reconnect with the corresponding sub-center node
  • the server establishes a connection.
  • the embodiment of the present application also proposes a storage medium, which is applied to a computer, that is, the storage medium is a computer-readable storage medium, and the control program of the federated learning system is stored on the medium.
  • the steps of the control method of the federated learning system as described above are realized.
  • the method implemented when the control program of the federated learning system running on the processor is executed can refer to the various embodiments of the control method based on the federated learning system of this application, which will not be repeated here.
  • the method of the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is better. ⁇
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, The optical disc) includes several instructions to enable a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the method described in each embodiment of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Information Transfer Between Computers (AREA)
  • Telephonic Communication Services (AREA)

Abstract

本申请公开了一种联邦学习系统的控制方法、装置、终端设备及存储介质,应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,所述子中心节点服务端和客户端所述为多个,该联邦学习系统的控制方法包括:获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。

Description

联邦学习系统的控制方法、装置、终端设备及存储介质
优先权信息
本申请要求于2019年8月26日申请的、申请号为201910800297.2、名称为“联邦学习系统的控制方法、装置、终端设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及Fintech(金融科技)技术领域,尤其涉及一种联邦学习系统的控制方法、装置、终端设备及存储介质。
背景技术
伴随着金融科技,尤其是互联网金融科技的快速发展,已经有越来越多的技术应用于金融领域,其中,联邦学习技术基于对用户隐私和数据的安全保障,正逐渐受到越来越多的重视。
联邦学习(federated learning)是指,通过联合不同的参与者(participant,或者party,也称为数据拥有者(data owner)、或者客户(client))进行机器学习建模的方法。在联邦学习中,参与者不需要向其它参与者和协调者(coordinator,也称为服务器(server),参数服务器(parameter server),或者聚合服务器(aggregation server))暴露自己所拥有的数据,因而联邦学习可以很好的保护用户隐私和保障数据安全,并可以解决数据孤岛问题。
然而,在现有的联邦学习中,尤其是在分层的横向联邦学习(横向联邦学习是在不同机构样本重叠较少,但特征维度重叠较多时,通过提取多方用户特征相同而用户不完全相同的那部分数据进行训练)中,若任意节点发生故障,则与当前节点建立连接的各参与者所拥有的数据将随即被剔除,从而无法再利用该部分数据进行联邦学习模型的训练,严重拉低了联邦学习的性能。
发明内容
本申请的主要目的在于提供一种联邦学习系统的控制方法、装置、终端设备及存储介 质,旨在避免在分层的横向联邦学习中,若任意节点发生故障,则无法利用与当前节点建立连接的各参与者所拥有的数据进行联邦学习模型的训练的技术问题,提升联邦学习的整体效率。
为实现上述目的,本申请提供一种联邦学习系统的控制方法,所述联邦学习的控制方法应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,其中,所述客户端与所述子中心节点服务端相连接,所述子中心节点服务端与所述主中心节点服务端相连接,所述子中心节点服务端和所述客户端为多个,
所述联邦学习系统的控制方法包括以下步骤:
获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
此外,本申请还提供一种联邦学习系统的控制装置,所述联邦学习的控制装置应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,其中,所述客户端与所述子中心节点服务端相连接,所述子中心节点服务端与所述主中心节点服务端相连接,所述子中心节点服务端和所述客户端为多个,
所述联邦学习系统的控制装置包括:
获取模块,用于获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
控制模块,用于基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
所述联邦学习系统的控制装置各功能模块运行时实现如上所述的联邦学习系统的控制方法的步骤。
此外,本申请还提供一种终端设备,所述终端设备包括:存储器、处理器及存储在所述存储器上并可在所述处理器上运行的联邦学习系统的控制程序,所述联邦学习系统的控制程序被所述处理器执行时实现如上所述的联邦学习系统的控制方法的步骤。
此外,本申请还提供一种存储介质,应用于计算机,所述存储介质上存储有联邦学习系统的控制程序,所述联邦学习系统的控制程序被处理器执行时实现如上所述的联邦学习系统的控制方法的步骤。
本申请联邦学习系统的控制方法,应用于联邦学习系统,所述联邦学习系统包括:主 中心节点服务端、子中心节点服务端以及客户端,其中,客户端与子中心节点服务端相连接,子中心节点服务端与主中心节点服务端相连接,子中心节点服务端和客户端为多个,本申请联邦学习系统的控制方法,通过获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
本申请实现了,在分层的横向联邦学习系统中,通过采集主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息,从而根据采集到的地理位置信息计算出各客户端各子中心节点服务端、以及主中心节点服务端相互之间的连线距离,并基于该连线距离控制当前联邦学习系统中的各客户端与各子中心节点服务端建立连接或者直接与主中心节点建立连接,并在任意节点(例如多个子中心节点服务端中的任意一个或者主中心节点服务端)发生故障时,将发生故障的当前节点所连接客户端,连接至其他相近的节点上,从而避免了在任意节点发生故障时,系统将当前节点上的参与者所拥有的数据进行剔除造成的数据浪费,并且,还避免了整个联邦学习系统无法利用该部分参与者所拥有的数据进行联邦学习模型的训练的技术问题,提升了联邦学习进行模型训练从而为用户提供服务的性能。
附图说明
图1是本申请实施例方案涉及的硬件运行环境的结构示意图;
图2为本申请联邦学习系统的控制方法第一实施例的流程示意图;
图3为本申请联邦学习系统的控制方法一实施例中应用场景示意图;
图4为本申请联邦学习系统的控制装置的模块示意图。
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
应当理解,此处所描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
如图1所示,图1是本申请实施例方案涉及的硬件运行环境的结构示意图。
需要说明的是,图1即可为终端设备的硬件运行环境的结构示意图。本申请实施例终端设备可以是PC,便携计算机等终端设备。
如图1所示,该终端设备可以包括:处理器1001,例如CPU,网络接口1004,用户接口 1003,存储器1005,通信总线1002。其中,通信总线1002用于实现这些组件之间的连接通信。用户接口1003可以包括显示屏(Display)、输入单元比如键盘(Keyboard),可选用户接口1003还可以包括标准的有线接口、无线接口。网络接口1004可选的可以包括标准的有线接口、无线接口(如WI-FI接口)。存储器1005可以是高速RAM存储器,也可以是稳定的存储器(non-volatile memory),例如磁盘存储器。存储器1005可选的还可以是独立于前述处理器1001的存储装置。
本领域技术人员可以理解,图1中示出的终端设备结构并不构成对终端设备的限定,可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置。
如图1所示,作为一种计算机存储介质的存储器1005中可以包括操作系统、网络通信模块、用户接口模块以及联邦学习系统的控制程序。其中,操作系统是管理和控制样本终端设备硬件和软件资源的程序,支持联邦学习系统的控制程序以及其它软件或程序的运行。
图1所示的终端设备,与其他终端共同组建成联邦学习系统,该联邦学习系统中至少包含一个服务端和多个客户端,在图1所示的终端设备中,用户接口1003主要用于与各个终端进行数据通信;网络接口1004主要用于连接后台服务器,与后台服务器进行数据通信;而处理器1001可以用于调用存储器1005中存储的联邦学习系统的控制程序,并执行以下操作:
获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,并执行以下步骤:
根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离;
检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离,其中,所述预设条件为连线距离的距离值最小;
控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,并执行以下步骤:
在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接;
或者,当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,并执行以下步骤:
在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,在执行根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离的步骤之后,执行以下步骤:
若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离;
选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服务器之间的各连线距离中,符合预设条件的第二目标连线距离;
控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,在执行控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤之后,执行以下步骤:
在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接。
进一步地,处理器1001还可以用于调用存储器1005中存储的联邦学习系统的控制程序,并执行以下步骤:
检测各所述客户端与恢复之后的所述目标子中心节点服务端之间的连线距离;
提取各所述连线距离中的第一目标连线距离,并控制所述第一目标连线距离所对应的所述客户端重新与对应的所述子中心节点服务端建立连接。
基于上述的结构,提出本申请联邦学习系统的控制方法的各个实施例。
请参照图2,图2为本申请联邦学习系统的控制方法第一实施例的流程示意图。
本申请实施例提供了联邦学习系统的控制方法的实施例,需要说明的是,虽然在流程图中示出了逻辑顺序,但是在某些情况下,可以以不同于此处的顺序执行所示出或描述的步骤。
本申请实施例联邦学习系统的控制方法应用于上述终端设备,本申请实施例终端设备可以是PC,便携计算机等终端设备,在此不做具体限制,进一步地,本申请联邦学习系统的控制方法应用于联邦学习系统,如图3所示为本申请联邦学习系统的控制方法应用于该联邦学习系统的一个应用场景,该场景所示的联邦学习系统包括至少一个主中心节点服务端、多个子中心节点服务端以及多个客户端,其中,客户端与所述子中心节点服务端相连接,子中心节点服务端与主中心节点服务端相连接。
本实施例联邦学习系统的控制方法包括:
步骤S100,获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息。
在当前联邦学习系统中,同时获取主中心节点服务端、各子中心节点服务端和各客户端的地理位置信息,并时刻检测是否存在新接入当前联邦学习系统中的客户端,若检测到有新的客户端接入当前联邦学习系统,则获取该新的客户端的地理位置信息。
进一步地,步骤S100,包括:
获取所述主中心节点服务端、各所述子中心节点服务端的地理位置信息;
判断所述客户端是否接入所述联邦学习系统;
若所述客户端接入所述联邦学习系统,则获取所述客户端的地理位置信息。
在分层的横向联邦学习系统中,基于LBS(Location Based Service:地理位置信息服务技术)获取当前联邦学习系统中,主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息。
具体地,例如,基于LBS获取主中心节点服务端的地理位置信息(如,获取到主中心节点服务端的地理位置坐标为:M主(x,y)),然后逐一获取各子中心节点服务端的地理位置信息(如,获取到各子中心节点服务端的地理位置坐标为:Mi(x i,y i),其中i为大于等于零的整数),并逐一获取与各子中心节点服务端相连接的各客户端的地理位置信息(获取到各客户端的地理位置坐标为:
Figure PCTCN2020107327-appb-000001
其中j为大于等于零的整数)。
进一步地,在另一个实施例中,本申请联邦学习系统的控制方法还可以先获取主中心节点服务端以及各子中心节点服务端的地理位置信息,并逐一检测当前联邦学习系统中的各客户端是否已经接入联邦学习系统,并在检测到当前客户端接入联邦学习系统时,获取该当前客户端的地理位置信息。
步骤S200,基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
根据获取到的当前分层的横向联邦学习系统中,主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息,逐一计算三者相互之间的连线距离,并依据该连线距离的距离数值大小,控制当前分层的横向联邦学习系统中的各客户端与各子中心节点服务端的其中之一建立连接、或者直接与主中心节点服务端建立连接。
具体地,例如,当基于LBS获取到主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息(即地理位置坐标)之后,逐一计算各客户端与各子中心节点服务端之间的连线距离,并计算各客户端与主中心节点服务端之间的连线距离,并检测各客户端与各子中节点服务端之间连线距离中距离值最小连线,从而按照该连线控制各客户端与各子中心节点服务端相连接,或者,在检测到客户端主中心节点服务端之间连线距离的距离值,比当前客户端与某一子中节点服务端之间连线距离的距离值还要小时,则控制当前客户端直接与主中心节点服务端相连接。
进一步地,步骤S200,包括:
步骤S201,根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离。
具体地,例如,当基于LBS获取到主中心节点服务端、各子中心节点服务端以及各客户端的地理位置坐标之后,依据平面内两点坐标距离的计算公式,逐一计算各客户端分别与各子中心节点服务端之间的连线距离以得出距离值:
Figure PCTCN2020107327-appb-000002
以及计算出各客户端与主中心节点服务端之间的连线距离并得出距离值D j
步骤S202,检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离。
逐一检测依据地理位置信息计算出的各客户端与各子中心节点服务端之间,各客户端与主中心节点服务端之间的各连线距离的距离值中,符合预设条件的距离值所标识第一目标连线距离。
本实例中,预设条件为:当前客户端与各子中心节点服务端之间,或者与主中心节点 服务之间的连线距离的距离值最小。
具体地,例如,在依据平面内两点坐标距离的计算公式,逐一计算各客户端分别与各子中心节点服务端之间的连线距离以得出距离值:
Figure PCTCN2020107327-appb-000003
以及计算出各客户端与主中心节点服务端之间的连线距离并得出距离值:D j之后,逐一提取各客户端中的其中一个客户端与各子中心节点服务端之间连线距离的各距离值:
Figure PCTCN2020107327-appb-000004
并检测提取出的各距离值中的最小距离值:
Figure PCTCN2020107327-appb-000005
然后,获取当前客户端与主中心节点服务端之间连线距离的距离值:D j,并进一步比较当前最小距离值:
Figure PCTCN2020107327-appb-000006
是否小于距离值:D j
步骤S203,控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
按照检测出的各客户端与各子中心节点服务端之间,各客户端与主中心节点服务端之间的各连线距离的距离值中,符合预设条件的距离值所标识第一目标连线距离对应的连线,控制各客户端与各子中心节点服务端或者直接与主中心节点服务端建立连接。
进一步地,步骤S203,包括:
步骤S2031,在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接。
具体地,例如,在提取出当前客户端与各子中心节点服务端之间连线距离的各距离值
Figure PCTCN2020107327-appb-000007
并检测提取出的各距离值中的最小距离值
Figure PCTCN2020107327-appb-000008
并进一步检测到当最小距离值
Figure PCTCN2020107327-appb-000009
小于当前客户端与主中心节点服务端之间连线距离的距离值D j时,获取当前距离值
Figure PCTCN2020107327-appb-000010
所对应的连线,并按照该连线,控制当前客户端与连线上的子中心节点服务端建立连接。
步骤S2032,当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
具体地,例如,在提取出当前客户端与各子中心节点服务端之间连线距离的各距离值
Figure PCTCN2020107327-appb-000011
并检测提取出的各距离值中的最小距离值
Figure PCTCN2020107327-appb-000012
并进一步检测到当前客户端与主中心节点服务端之间连线距离的距离值D j小于当最小距离值
Figure PCTCN2020107327-appb-000013
时,获取当前距离值D j所对应的连线,并按照该连线控制当前客户端与主中心节点服务端建立连接。
进一步地,本申请联邦学习系统的控制方法还包括:
步骤A,在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤。
在分层的横向联邦学习系统中,基于现有的位置信息服务技术,时刻对各客户端分别与主中心节点服务端、以及分别与各子中心节点服务端之间连线的连线距离进行动态监测,并按照检测到的各连线距离中,符合预设条件的第一目标连线距离,控制该第一目标连线距离所对应的客户端与对应的子中心节点服务端建立连接或者直接与主中心节点服务端建立连接。从而,在客户端地理位置变化过程中,实现始终采用客户端与子中心节点服务端或者与主中心节点服务端之间连线距离最短的连接线路将客户端接入当前联学习系统,提升了联邦学习系统的整体性能。
具体地,例如,在基于LBS获取到主中心节点服务端、各子中心节点服务端以及各客户端的地理位置坐标,并依据平面内两点坐标距离的计算公式,计算各客户端分别与各子中心节点服务端之间的连线距离以得出距离值:
Figure PCTCN2020107327-appb-000014
以及计算出各客户端与主中心节点服务端之间的连线距离并得出距离值D j之后,持续对各客户端与各子中心节点服务端之间连线距离的各距离值
Figure PCTCN2020107327-appb-000015
各客户端与主中心节点服务端之间的连线距离并得出距离值D j中的最小距离值
Figure PCTCN2020107327-appb-000016
minD j进行监测,并始终按照监测到的最小距离值
Figure PCTCN2020107327-appb-000017
或者最小距离值D j,控制各客户端与最小距离值
Figure PCTCN2020107327-appb-000018
所对应的连线的子中心节点服务端建立连接,或者与最小距离值D j所对应的连线的主中心节点服务端建立连接。
进一步地,在步骤S201之后,本申请联邦学习系统的控制方法,还包括:
步骤S204若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离。
在分层的横向联邦学习系统中,时刻对主中心节点服务端、各子中心节点服务端以及各客户端的工作状态进行监测,当检测到各子中心节点服务端中的任意一个或者多个宕机(即出现故障)时,随即获取与当前横向联邦学习系统中,与该出现故障的目标子中心节点服务端相连接的各目标客户端与全部子中心节点服务端中的其他各子中心节点服务端之间,以及与主中心节点服务端之间的连线距离。
具体地,例如,在如图3所示的分为三层的横向联邦学习系统中,当检测到子中心节点服务端1至5中的子中心节点服务端4宕机(即出现故障)时,随即逐一获取与当前目标子中心节点服务端4建立连接的客户端5至7,与当前横向联邦学习系统中其他各子中心节点服务端(即子中心节点服务端1至3和子中心节点服务端4)之间,以及与当前横向联邦学习系统中主中心节点服务端之间的连线距离。
步骤S205,选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服 务器之间的各连线距离中,符合预设条件的第二目标连线距离。
检测获取到的各目标客户端与各子中心节点服务端之间,各目标客户端与主中心节点服务端之间的各连线距离的距离值中,符合预设条件(即连线距离的距离值最小)的距离值所标识第二目标连线距离。
具体地,例如,在如图3所示的分为三层的横向联邦学习系统中,当获取到目标客户端5与当前横向联邦学习系统中的子中心节点服务端1至3和子中心节点服务端4之间各连线距离的各距离值
Figure PCTCN2020107327-appb-000019
以及与当前横向联邦学习系统中主中心节点服务端之间连线距离的距离值D j之后,检测各距离值
Figure PCTCN2020107327-appb-000020
中的最小距离值
Figure PCTCN2020107327-appb-000021
为当前目标客户端5与子中心节点服务端3之间连线距离的距离值,并且当前最小距离值
Figure PCTCN2020107327-appb-000022
小于距离值D j,从而确定当前目标客户端5与子中心节点服务端3之间的连线距离为第二目标连线距离。
步骤S206,控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
按照检符合预设条件的距离值所标识第二目标连线距离对应的连线,控制各目标客户端与各目标子中心节点服务端,或者直接与主中心节点服务端建立连接。
具体地,例如,在如图3所示的分为三层的横向联邦学习系统中,在确定目标客户端5与子中心节点服务端3之间的连线距离为第二目标连线距离之后,则按照当前第二目标连线距离对应的连线,控制目标客户端5与子中心节点服务端3建立连接,并在基于上述同样的操作确定目标客户端6与主中心节点服务端之间的连线距离为第二目标连线距离、目标客户端7与子中心节点服务端5之间的连线距离为第二目标连线距离之后,分别按照第二目标连线距离对应的连线,控制目标客户端6直接与主中心节点服务端建立连接,控制目标客户端7与子中心节点服务端5建立连接。
本申请通过在分层的横向联邦学习系统中,基于LBS(Location Based Service:地理位置信息服务技术)获取当前联邦学习系统中,主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息,根据获取到的当前分层的横向联邦学习系统中,主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息计算三者相互之间的各连线距离,逐一检测各连线距离的距离值中,符合预设条件的距离值所标识第一目标连线距离,按照第一目标连线距离对应的连线,控制各客户端与各子中心节点服务端或者直接与主中心节点服务端建立连接,并且,在分层的横向联邦学习系统中,时刻对主中心节点服务端、各子中心节点服务端以及各客户端的工作状态进行监测,当检测到各子中心节点服务端中的任意一个或者多个宕机(即出现故障)时,随即获取与当前横向联邦学习系统中,与该 出现故障的目标子中心节点服务端相连接的各目标客户端与全部子中心节点服务端中的其他各子中心节点服务端之间,以及与主中心节点服务端之间的各连线距离,检测各连线距离的距离值中,符合预设条件(即连线距离的距离值最小)的距离值所标识第二目标连线距离,按照检符合预设条件的距离值所标识第二目标连线距离对应的连线,控制各目标客户端与各目标子中心节点服务端,或者直接与主中心节点服务端建立连接。
实现了,在分层的横向联邦学习系统中,通过采集主中心节点服务端、各子中心节点服务端以及各客户端的地理位置信息,从而根据采集到的地理位置信息计算出各客户端各子中心节点服务端、以及主中心节点服务端相互之间的连线距离,并基于该连线距离控制当前联邦学习系统中的各客户端与各子中心节点服务端或者直接与主中心节点建立连接,并在任意节点(例如多个子中心节点服务端中的任意一个或者主中心节点服务端)发生故障时,将发生故障的当前节点所连接的参与者,连接至其他相近的节点上,从而避免了在分层的横向联邦学习中,若任意节点发生故障,系统将当前节点上的参与者所拥有的数据进行剔除造成的数据浪费,并且,还避免了整个联邦学习系统无法利用该部分参与者所拥有的数据进行联邦学习模型的训练的技术问题,提升了联邦学习进行模型训练从而为用户提供服务的性能。
进一步地,提出本申请联邦学习系统的控制方法的第二实施例。
基于上述联邦学习系统的控制方法第一实施例,本实施例中,在上述步骤S206,控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤之后,本申请联邦学习系统的控制方法还包括:
步骤S207,在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接。
在分层的横向联邦学习系统中,时刻对主中心节点服务端、各子中心节点服务端以及各客户端的工作状态进行监测,当检测到各子中心节点服务端中宕机(即出现故障)的目标子中心节点服务端恢复运行时,控制当前联邦学习系统中的各客户端,与当前目标子中心节点服务端重新建立连接,从而减小当前与各目标客户端建立连接的其他各子中心节点服务端的运行压力,进一步保证了横向联邦学习系统的整体性能。
进一步地,步骤S207,包括:
检测各所述客户端与恢复之后的所述目标子中心节点服务端之间的连线距离;
提取各所述连线距离中的第一目标连线距离,并控制所述第一目标连线距离所对应的 所述客户端重新与对应的所述子中心节点服务端建立连接。
具体地,例如,在如图3所示的分为三层的横向联邦学习系统中,当检测到出现故障的目标子中心节点服务端4恢复正常运行(例如工作人员完成修改并测试运行正常)时,重新检测当前联邦学习系统中的客户端1至客户端9分别与恢复的目标自中心节点服务端4之间的连线距离,并提取出客户端1至客户端9分别与恢复的目标自中心节点服务端4之间的连线距离中,距离值最小的第一目标连线距离,并随即控制距离值最小的第一目标连线距离所对应的客户端5至7,重新与当前目标子中心节点服务端4建立连接。
进一步地,在另一实施例中,当检测到当前横向联邦学习系统中的主中心节点服务端宕机(即出现故障)时,随即控制当前横向联邦学习系统中与主中心节点服务端所连接的子中心节点服务端以及客户端,与距离主中心节点服务端最近的子中心节点服务端建立连接,从而避免了当前整个联邦学习系统因主中心节点服务端出现故障而整个瘫痪,提升了横向联邦学习系统的整体性能,同样的,在检测到主中心节点服务端重新恢复正常运行之后,控制主中心节点服务端在出现故障之前所连接的子中心节点服务端或者客户端,与主中心节点服务端重新建立连接,从而减小与距离主中心节点服务端最近的子中心节点服务端的运行压力,进一步保证了横向联邦学习系统的整体性能。
具体地,例如,在如图3所示的分为三层的横向联邦学习系统中,当检测到主中心节点服务端出现故障时,随即控制子中心节点服务端2至5与距离主中心节点服务端最近的子中心节点服务端1建立连接,由子中心节点服务端1充电临时的主中心节点服务端,并在检测到主中心节点服务端重新恢复正常运行之后,控制子中心节点服务端2至5重新与主中心节点服务端建立连接。
本申请通过当检测到各子中心节点服务端中宕机(即出现故障)的目标子中心节点服务端恢复运行时,控制当前目标子中心节点服务端在出现故障之前所连接的各目标客户端,与当前目标子中心节点服务端重新建立连接,从而减小当前与各目标客户端建立连接的其他各子中心节点服务端的运行压力,进一步保证了横向联邦学习系统的整体性能;在另一方面,当检测到当前横向联邦学习系统中的主中心节点服务端宕机时,随即控制当前横向联邦学习系统中与主中心节点服务端所连接的子中心节点服务端以及客户端,与距离主中心节点服务端最近的子中心节点服务端建立连接,从而避免了当前整个联邦学习系统因主中心节点服务端出现故障而整个瘫痪,提升了横向联邦学习系统的整体性能,同样的,在检测到主中心节点服务端重新恢复正常运行之后,控制主中心节点服务端在出现故障之前所连接的子中心节点服务端或者客户端,与主中心节点服务端重新建立连接,从而减小 与距离主中心节点服务端最近的子中心节点服务端的运行压力,进一步保证了横向联邦学习系统的整体性能。
此外,请参照图4,本申请实施例还提出一种联邦学习系统的控制装置,所述联邦学习的控制装置应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,其中,所述客户端与所述子中心节点服务端相连接,所述子中心节点服务端与所述主中心节点服务端相连接,所述子中心节点服务端和所述客户端为多个,
所述联邦学习系统的控制装置包括:
获取模块,用于获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
控制模块,用于基于所述地理位置信息,控制各所述客户端通过各所述子中心节点服务端与所述主中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
在一实施例中,控制模块,包括:
第一检测单元,用于根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离;
第二检测单元,用于检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离,其中,所述预设条件为连线距离的距离值最小;
第一控制单元,用于控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
在一实施例中,第一控制单元,包括:
第一控制子单元,用于在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接;
第二控制子单元,用于当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
在一实施例中,控制模块还用于:在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤。在一实施例中,控制 模块,还包括:
第一获取单元,用于若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离;
选取单元,用于选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服务器之间的各连线距离中,符合预设条件的第二目标连线距离;
第二控制单元,用于控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
在一实施例中,控制模块,还包括:
第三控制单元,用于在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接。
在一实施例中,第三控制单元,包括:
第三检测单元,用于检测各所述客户端与恢复之后的所述目标子中心节点服务端之间的连线距离;
第三控制子单元,用于提取各所述连线距离中的第一目标连线距离,并控制所述第一目标连线距离所对应的所述客户端重新与对应的所述子中心节点服务端建立连接。
本实施例提出的联邦学习系统的控制装置各个功能模块运行时实现如上所述的参数相似性的评估方法的步骤,在此不再赘述。
此外,本申请实施例还提出一种存储介质,应用于计算机,即所述存储介质为计算机可读存储介质,所述介质上存储有联邦学习系统的控制程序,所述联邦学习系统的控制程序被处理器执行时实现如上所述的联邦学习系统的控制方法的步骤。
其中,在所述处理器上运行的联邦学习系统的控制程序被执行时所实现的方法可参照本申请基于联邦学习系统的控制方法各个实施例,此处不再赘述。
需要说明的是,在本文中,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。
上述本申请实施例序号仅仅为了描述,不代表实施例的优劣。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可 借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种联邦学习系统的控制方法,其中,所述联邦学习的控制方法应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,其中,所述客户端与所述子中心节点服务端相连接,所述子中心节点服务端与所述主中心节点服务端相连接,所述子中心节点服务端和所述客户端为多个,
    所述联邦学习系统的控制方法包括以下步骤:
    获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
    基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
  2. 如权利要求1所述的联邦学习系统的控制方法,其中,所述基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接的步骤,包括:
    根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离;
    检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离,其中,所述预设条件为连线距离的距离值最小;
    控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
  3. 如权利要求2所述的联邦学习系统的控制方法,其中,所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤,包括:
    在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接;
    或者,当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
  4. 如权利要求2所述的联邦学习系统的控制方法,其中,所述联邦学习系统的控制方法,还包括:
    在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤。
  5. 如权利要求2所述的联邦学习系统的控制方法,其中,在所述根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离的步骤之后,所述方法还包括:
    若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离;
    选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服务器之间的各连线距离中,符合预设条件的第二目标连线距离;
    控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
  6. 如权利要求5所述的联邦学习系统的控制方法,其中,在所述基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接的步骤之后,所述方法还包括:
    在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接。
  7. 如权利要求1-6所述的联邦学习系统的控制方法,其中,所述在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接的步骤,包括:
    检测各所述客户端与恢复之后的所述目标子中心节点服务端之间的连线距离;
    提取各所述连线距离中的第一目标连线距离,并控制所述第一目标连线距离所对应的所述客户端重新与对应的所述子中心节点服务端建立连接。
  8. 一种联邦学习系统的控制装置,其中,所述联邦学习的控制装置应用于联邦学习系统,所述联邦学习系统包括:主中心节点服务端、子中心节点服务端以及客户端,其中,所述客户端与所述子中心节点服务端相连接,所述子中心节点服务端与所述主中心节点服务端相连接,所述子中心节点服务端和所述客户端为多个,
    所述联邦学习系统的控制装置包括:
    获取模块,用于获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
    控制模块,用于基于所述地理位置信息,控制各所述客户端通过各所述子中心节点服务端与所述主中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
  9. 一种终端设备,其中,所述终端设备包括:存储器、处理器及存储在所述存储器上并可在所述处理器上运行的联邦学习系统的控制程序,所述联邦学习系统的控制程序被所述处理器执行时实现如下步骤:
    获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
    基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
  10. 如权利要求9所述的终端设备,其中,所述基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接的步骤,包括:
    根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离;
    检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离,其中,所述预设条件为连线距离的距离值最小;
    控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
  11. 如权利要求10所述的终端设备,其中,所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤,包括:
    在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接;
    或者,当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
  12. 如权利要求10所述的终端设备,其中,所述联邦学习系统的控制方法,还包括:
    在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述 主中心节点服务端建立连接的步骤。
  13. 如权利要求10所述的终端设备,其中,在所述根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离的步骤之后,所述方法还包括:
    若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离;
    选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服务器之间的各连线距离中,符合预设条件的第二目标连线距离;
    控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
  14. 如权利要求13所述的终端设备,其中,在所述基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接的步骤之后,所述方法还包括:
    在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接。
  15. 如权利要求9-14所述的终端设备,其中,所述在所述目标子中心节点服务端恢复之后,控制各所述客户端重新与所述目标子中心节点服务端建立连接的步骤,包括:
    检测各所述客户端与恢复之后的所述目标子中心节点服务端之间的连线距离;
    提取各所述连线距离中的第一目标连线距离,并控制所述第一目标连线距离所对应的所述客户端重新与对应的所述子中心节点服务端建立连接。
  16. 一种存储介质,其中,应用于计算机,所述存储介质上存储有联邦学习系统的控制程序,所述联邦学习系统的控制程序被处理器执行时实现如下步骤:
    获取所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端的地理位置信息;
    基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接。
  17. 如权利要求16所述的存储介质,其中,所述基于所述地理位置信息,控制各所述客户端与各所述子中心节点服务端建立连接,或者,控制各所述客户端直接与所述主中心节点服务端建立连接的步骤,包括:
    根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各 所述客户端相互之间的连线距离;
    检测各所述客户端分别与各所述子中心节点以及与所述主中心节点服务端之间的各连线距离中,符合预设条件的第一目标连线距离,其中,所述预设条件为连线距离的距离值最小;
    控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
  18. 如权利要求17所述的存储介质,其中,所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤,包括:
    在检测到当前所述第一目标连线距离为所述客户端与所述子中心节点服务端之间的连线距离时,控制所述客户端与所述子中心节点服务端建立连接;
    或者,当检测到当前所述第一目标连线距离为所述客户端与所述主中心节点服务端之间的连线距离时,控制所述客户端直接与所述主中心节点服务端建立连接。
  19. 如权利要求17所述的存储介质,其中,所述联邦学习系统的控制方法,还包括:
    在各所述连线距离中动态检测所述第一目标连线距离,并执行所述控制所述第一目标连线距离所对应的所述客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接的步骤。
  20. 如权利要求17所述的存储介质,其中,在所述根据所述地理位置信息检测所述主中心节点服务端、各所述子中心节点服务端以及各所述客户端相互之间的连线距离的步骤之后,所述方法还包括:
    若所述子中心节点服务端宕机,获取当前宕机的目标子中心节点服务端上连接的各目标客户端与其它各所述子中心节点服务端以及所述主中心节点服务端之间的连线距离;
    选取各所述目标客户端分别与各所述子中心节点以及与所述主中心节点服务器之间的各连线距离中,符合预设条件的第二目标连线距离;
    控制所述第二目标连线距离所对应的所述目标客户端与对应的所述子中心节点服务端建立连接或者直接与所述主中心节点服务端建立连接。
PCT/CN2020/107327 2019-08-26 2020-08-06 联邦学习系统的控制方法、装置、终端设备及存储介质 Ceased WO2020259717A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910800297.2A CN110380917B (zh) 2019-08-26 2019-08-26 联邦学习系统的控制方法、装置、终端设备及存储介质
CN201910800297.2 2019-08-26

Publications (1)

Publication Number Publication Date
WO2020259717A1 true WO2020259717A1 (zh) 2020-12-30

Family

ID=68260969

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/107327 Ceased WO2020259717A1 (zh) 2019-08-26 2020-08-06 联邦学习系统的控制方法、装置、终端设备及存储介质

Country Status (2)

Country Link
CN (1) CN110380917B (zh)
WO (1) WO2020259717A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113011599A (zh) * 2021-03-23 2021-06-22 上海嗨普智能信息科技股份有限公司 基于异构数据的联邦学习系统
CN113159329A (zh) * 2021-04-27 2021-07-23 Oppo广东移动通信有限公司 模型训练方法、装置、设备及存储介质
CN114281231A (zh) * 2021-10-12 2022-04-05 腾讯科技(深圳)有限公司 信息呈现方法、装置、电子设备和存储介质
CN115086399A (zh) * 2022-07-28 2022-09-20 深圳前海环融联易信息科技服务有限公司 基于超网络的联邦学习方法、装置及计算机设备
CN116522379A (zh) * 2023-03-30 2023-08-01 昆山杜克大学 基于分层联邦系统的频繁数据挖掘方法及相关设备

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110380917B (zh) * 2019-08-26 2022-01-07 深圳前海微众银行股份有限公司 联邦学习系统的控制方法、装置、终端设备及存储介质
WO2021092758A1 (zh) * 2019-11-12 2021-05-20 深圳前海微众银行股份有限公司 一种联邦学习方法、装置及计算机可读存储介质
CN110995488B (zh) * 2019-12-03 2020-11-03 电子科技大学 一种基于分层参数服务器的多机构协同学习系统及方法
CN111538608B (zh) * 2020-04-30 2024-10-29 深圳前海微众银行股份有限公司 预防终端设备宕机的方法、终端设备及存储介质
CN111537945B (zh) * 2020-06-28 2021-05-11 南方电网科学研究院有限责任公司 基于联邦学习的智能电表故障诊断方法及设备
CN111722043B (zh) * 2020-06-29 2021-09-14 南方电网科学研究院有限责任公司 一种电力设备故障检测方法、装置及系统
US11909482B2 (en) * 2020-08-18 2024-02-20 Qualcomm Incorporated Federated learning for client-specific neural network parameter generation for wireless communication
CN112070240B (zh) * 2020-09-07 2022-10-14 清华大学 一种高效通信的分层联邦学习框架及其优化方法和系统
CN113919511B (zh) * 2020-09-24 2025-03-18 京东科技控股股份有限公司 联邦学习方法及装置
CN112671613B (zh) * 2020-12-28 2022-08-23 深圳市彬讯科技有限公司 联邦学习集群监控方法、装置、设备及介质
CN113887740B (zh) * 2021-10-21 2024-07-16 支付宝(杭州)信息技术有限公司 联合更新模型的方法、装置及系统
CN117014449A (zh) * 2022-04-29 2023-11-07 索尼集团公司 用于无线通信系统的电子设备、方法和存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107871160A (zh) * 2016-09-26 2018-04-03 谷歌公司 通信高效联合学习
CN109189825A (zh) * 2018-08-10 2019-01-11 深圳前海微众银行股份有限公司 横向数据切分联邦学习建模方法、服务器及介质
CN109586969A (zh) * 2018-12-13 2019-04-05 平安科技(深圳)有限公司 内容分发网络容灾方法、装置、计算机设备及存储介质
US10270599B2 (en) * 2017-04-27 2019-04-23 Factom, Inc. Data reproducibility using blockchains
CN109711556A (zh) * 2018-12-24 2019-05-03 中国南方电网有限责任公司 机巡数据处理方法、装置、网级服务器和省级服务器
CN110380917A (zh) * 2019-08-26 2019-10-25 深圳前海微众银行股份有限公司 联邦学习系统的控制方法、装置、终端设备及存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109167695B (zh) * 2018-10-26 2021-12-28 深圳前海微众银行股份有限公司 基于联邦学习的联盟网络构建方法、设备及可读存储介质
CN109699043B (zh) * 2018-12-25 2022-10-21 北京云中融信网络科技有限公司 链路建立方法和装置
CN109871702B (zh) * 2019-02-18 2024-06-28 深圳前海微众银行股份有限公司 联邦模型训练方法、系统、设备及计算机可读存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107871160A (zh) * 2016-09-26 2018-04-03 谷歌公司 通信高效联合学习
US10270599B2 (en) * 2017-04-27 2019-04-23 Factom, Inc. Data reproducibility using blockchains
CN109189825A (zh) * 2018-08-10 2019-01-11 深圳前海微众银行股份有限公司 横向数据切分联邦学习建模方法、服务器及介质
CN109586969A (zh) * 2018-12-13 2019-04-05 平安科技(深圳)有限公司 内容分发网络容灾方法、装置、计算机设备及存储介质
CN109711556A (zh) * 2018-12-24 2019-05-03 中国南方电网有限责任公司 机巡数据处理方法、装置、网级服务器和省级服务器
CN110380917A (zh) * 2019-08-26 2019-10-25 深圳前海微众银行股份有限公司 联邦学习系统的控制方法、装置、终端设备及存储介质

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HU BINXUAN ET AL.: "Federated Region-Learning: An Edge Computing Based Framework for Urban Environment Sensing", IEEE, 21 February 2019 (2019-02-21), XP033519811, DOI: 20201016104341X *
HU BINXUAN ET AL.: "Federated Region-Learning: An Edge Computing Based Framework for Urban Environment Sensing", IEEE, 21 February 2019 (2019-02-21), XP033519811, DOI: 20201016104409Y *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113011599A (zh) * 2021-03-23 2021-06-22 上海嗨普智能信息科技股份有限公司 基于异构数据的联邦学习系统
CN113159329A (zh) * 2021-04-27 2021-07-23 Oppo广东移动通信有限公司 模型训练方法、装置、设备及存储介质
CN113159329B (zh) * 2021-04-27 2023-10-31 Oppo广东移动通信有限公司 模型训练方法、装置、设备及存储介质
CN114281231A (zh) * 2021-10-12 2022-04-05 腾讯科技(深圳)有限公司 信息呈现方法、装置、电子设备和存储介质
CN114281231B (zh) * 2021-10-12 2023-10-20 腾讯科技(深圳)有限公司 信息呈现方法、装置、电子设备和存储介质
CN115086399A (zh) * 2022-07-28 2022-09-20 深圳前海环融联易信息科技服务有限公司 基于超网络的联邦学习方法、装置及计算机设备
CN116522379A (zh) * 2023-03-30 2023-08-01 昆山杜克大学 基于分层联邦系统的频繁数据挖掘方法及相关设备

Also Published As

Publication number Publication date
CN110380917A (zh) 2019-10-25
CN110380917B (zh) 2022-01-07

Similar Documents

Publication Publication Date Title
WO2020259717A1 (zh) 联邦学习系统的控制方法、装置、终端设备及存储介质
US20240235925A1 (en) Intelligent network
JP5745993B2 (ja) 地域基盤の検索語を推薦するシステム及びその方法
CN108664957B (zh) 车牌号码匹配方法及装置、字符信息匹配方法及装置
CN110428058A (zh) 联邦学习模型训练方法、装置、终端设备及存储介质
CN106921504B (zh) 一种确定不同用户的关联路径的方法和设备
WO2023093638A1 (zh) 异常数据识别方法、装置、设备和存储介质
CN103177083A (zh) 一种工单处理系统
CN110611667A (zh) 边缘计算环境下动态的位置隐私保护方法及装置
CN112769837A (zh) 基于WebSocket的通信传输方法、装置、设备、系统及存储介质
WO2017054705A1 (zh) 一种局域网内基于物理位置的终端监控方法及系统
CN104125281A (zh) 云数据中心申请资源审批状态记录方法及装置
CN118199900A (zh) 一种算力调度方法、装置和算网管理设备
CN104244168A (zh) 基于android系统与微信平台的盲人帮扶系统
KR20180089479A (ko) 사용자 데이터 공유 방법 및 디바이스
CN112131903A (zh) 设备数据的分析方法、装置、服务平台、系统及介质
CN103249020A (zh) 基于云服务的应急呼叫处理方法及系统以及移动终端
CN109672721B (zh) 媒体文件推送方法、装置、服务端及计算机可读存储介质
CN115984643A (zh) 模型训练方法、相关设备及存储介质
CN110366097A (zh) 一种基于位置信息识别风险的方法、装置和电子设备
CN110149371A (zh) 设备连接方法、装置及终端设备
CN108810803A (zh) 寻找失联用户的方法及系统
CN116708430A (zh) 注册中心集群调度方法、装置、计算机设备及存储介质
CN110363638A (zh) 金融业务中非目标用户的业务处理方法、装置和电子设备
CN106878312B (zh) 一种基于边簇图的语义位置隐私保护方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20832140

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20832140

Country of ref document: EP

Kind code of ref document: A1