CN114611008B - User service strategy determination method and device based on federal learning and electronic equipment - Google Patents
User service strategy determination method and device based on federal learning and electronic equipment Download PDFInfo
- Publication number
- CN114611008B CN114611008B CN202210499538.6A CN202210499538A CN114611008B CN 114611008 B CN114611008 B CN 114611008B CN 202210499538 A CN202210499538 A CN 202210499538A CN 114611008 B CN114611008 B CN 114611008B
- Authority
- CN
- China
- Prior art keywords
- party
- user
- feature
- characteristic
- information
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 73
- 238000012549 training Methods 0.000 claims abstract description 103
- 238000012545 processing Methods 0.000 claims description 21
- 238000003066 decision tree Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 12
- 238000010801 machine learning Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010921 in-depth analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5041—Network service management, e.g. ensuring proper service fulfilment according to agreements characterised by the time relationship between creation and deployment of a service
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Bioethics (AREA)
- Medical Informatics (AREA)
- Computational Linguistics (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a method and a device for determining a user service strategy based on federal learning, electronic equipment and a computer readable medium. The method comprises the following steps: a label party for model training encrypts user labels of a plurality of users to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encrypted information; the at least one characteristic party performs model training according to characteristic information quantity and the label party on the basis of federal learning in sequence to generate a user scoring model; generating user characteristics according to user data of a current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scoring; and comparing the user score with a preset interval to determine the service strategy of the user and push the service strategy. The method can rapidly and accurately provide the most appropriate service strategy for the user under the condition of guaranteeing the safety of the user data.
Description
Technical Field
The application relates to the field of computer information processing, in particular to a user service strategy determination method and device based on federal learning, electronic equipment and a computer readable medium.
Background
For an organization providing user service, the comprehensive and deep analysis of the user can be helpful for providing better service for the user. However, in many cases, the user information submitted by the user is not sufficient, and particularly for the user who has just landed for registration, the user service organization can only obtain simple user information. In this case, it is an important trend to integrate user data distributed in various places and in various organizations.
However, there are difficult barriers between data sources of the respective organizations, and the data required for artificial intelligence generally involves a plurality of fields. In most industries, data exists in an isolated island form, and due to the problems of industry competition, privacy and safety, complicated administrative procedures and the like, even if data integration is realized among different departments of the same company, important resistance is faced, and in reality, the data which is scattered in various places and various mechanisms is almost impossible to integrate, or the required cost is huge.
In the prior art, the training of a machine learning model can be carried out jointly among mechanisms through federal learning, plaintext data needs to be protected when modeling is carried out under a federal learning scene, and data interacted among the mechanisms cannot have plaintext or cannot be reversely deduced. The most common scenario is federal learning by two-party agencies, where one party provides the tags and the other provides the features. However, in practical situations, a plurality of organizations are often required to conduct federal study together, and currently, no good solution exists for federal study in a multi-party scenario.
The above information disclosed in this background section is only for enhancement of understanding of the background of the application and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
In view of this, the present application provides a method, an apparatus, an electronic device, and a computer-readable medium for determining a user service policy based on federal learning, which can perform comprehensive and accurate evaluation on a user on the premise of guaranteeing user data security, system data security, and transaction security, thereby quickly and accurately providing a most appropriate service policy for the user.
Other features and advantages of the present application will be apparent from the following detailed description, or may be learned by practice of the application.
According to an aspect of the present application, a method for determining a user service policy based on federal learning is provided, where the method includes: a label party for model training encrypts user labels of a plurality of users to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encryption information; the at least one characteristic party performs model training according to characteristic information quantity and the label party on the basis of federal learning in sequence to generate a user scoring model; generating user characteristics according to user data of a current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scores; and comparing the user score with a preset interval to determine the service strategy of the user and push the service strategy.
Optionally, the method further comprises: a label party for model training encrypts user labels of a plurality of users to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encryption information; sorting the at least one feature party according to the feature information amount corresponding to the feature party; and the at least one characteristic party sequentially performs model training with the label party based on federal learning according to the sequence until all the characteristic parties finish training to generate the user scoring model.
Optionally, the method for generating encrypted information by encrypting the user tags of multiple users by the tag party of model training includes: a label party for model training generates an encryption key through an addition homomorphic encryption mode; encrypting the user tags of the plurality of users by the encryption key to generate encrypted tags; and the label party trained by the model sends the encryption key and the encryption label as encryption information to the at least one characteristic party.
Optionally, the at least one feature party performs model training with the tag party based on federal learning in sequence according to the feature information quantity to generate a user rating model, including: sorting the at least one feature party according to the feature information amount corresponding to the feature party; and the at least one characteristic party performs model training according to the sequence and the label party on the basis of federal learning, and a user scoring model is generated until all characteristic parties are trained.
Optionally, at least one feature party of the model training generates a feature information quantity of a user feature corresponding to the feature information quantity according to the encrypted information, including: each characteristic party of the model training carries out box separation processing on the corresponding user characteristics to generate a plurality of box separation characteristic sets; generating a plurality of box information quantities of a plurality of box characteristic sets according to the encrypted information; and each characteristic party of the model training generates characteristic information quantity according to the corresponding plurality of sub-box information quantities.
Optionally, each feature party of the model training performs binning processing on the user features corresponding to the feature party to generate a plurality of binning feature sets, including: each feature side of the model training divides the corresponding user feature according to a decision tree box dividing mode to generate a plurality of dividing points; and performing binning processing on the user features according to the plurality of segmentation points to generate a plurality of binning feature sets.
Optionally, generating a plurality of binning information volumes of a plurality of binning feature sets according to the encryption information includes: encrypting the characteristic values in the plurality of sub-box characteristic sets according to an encryption key in the encryption information to generate an encrypted value; generating the plurality of binned information quantities by encrypting values in the plurality of binned feature sets after the cryptographic computation.
Optionally, the performing, by the at least one feature party, model training with the label party based on federal learning in sequence according to the ranking until all feature parties finish training to generate the user rating model includes: extracting an initial feature party by the at least one feature party according to the ranking; the initial characteristic party performs federal learning through the corresponding user characteristics, the encryption information and the label party to generate a plurality of first scores; extracting a next feature party of the initial feature party as a current feature party according to the sequence; the current characteristic party performs federal learning according to the corresponding user characteristic, the encrypted label, the plurality of first scores and the label party to generate a plurality of second scores; and sequentially extracting the next feature party and the next label party according to the sequence for federal learning until all the feature parties finish training to generate the user scoring model.
Optionally, the initial feature party performs federal learning by using its corresponding user feature, encryption information, and tag party, and generates a plurality of first scores, including: the initial characteristic party encrypts the user characteristics according to the encryption key in the encryption information to generate encryption characteristics; the initial characteristic party and the label party carry out federal learning through the encryption characteristics and the encryption labels in the encryption information; when training is finished, generating an initial model; a plurality of first scores for a plurality of users of the tag party are calculated from the initial model.
Optionally, the current feature party performs federal learning according to the user feature, the encrypted tag, the plurality of first scores and the tag party corresponding thereto, and generates a plurality of second scores, including: the current characteristic party encrypts the user characteristics according to the encryption key in the encryption information to generate encryption characteristics; the current characteristic party and the label party carry out federal learning through encrypted characteristics, encrypted labels in the encrypted information and the plurality of first scores; when training is finished, generating a current model; a plurality of second scores for a plurality of users of the tag party are calculated according to the current model.
According to an aspect of the present application, a user service policy determination apparatus based on federal learning is provided, the apparatus including: the encryption module is used for controlling a label party trained by the model to encrypt user labels of a plurality of users to generate encrypted information; the characteristic module is used for controlling at least one characteristic party of the model training to generate the characteristic information quantity of the corresponding user characteristic according to the encrypted information; the training module is used for performing model training on the at least one characteristic party and the label party in sequence based on federal learning according to the characteristic information quantity to generate a user grading model; the characteristic module is used for generating user characteristics according to the user data of the current user; the scoring module is used for inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scores; and the strategy module is used for comparing the user score with a preset interval so as to determine the service strategy of the current user and push the service strategy.
According to an aspect of the present application, an electronic device is provided, the electronic device including: one or more processors; storage means for storing one or more programs; when executed by one or more processors, cause the one or more processors to implement a method as above.
According to an aspect of the application, a computer-readable medium is proposed, on which a computer program is stored, which program, when being executed by a processor, carries out the method as above.
According to the method, the device, the electronic equipment and the computer readable medium for determining the user service strategy based on the federal learning, the user characteristics are generated according to the user data of the user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scoring; the user scores are compared with the preset interval to determine the service strategy of the user and push the service strategy, so that the user can be comprehensively and accurately evaluated on the premise of guaranteeing user data safety, system data safety and transaction safety, and the most appropriate service strategy can be quickly and accurately provided for the user.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings. The drawings described below are only some embodiments of the present application and other drawings may be derived by those skilled in the art without inventive effort.
Fig. 1 is a system diagram illustrating a method and apparatus for user service policy determination based on federal learning in accordance with an exemplary embodiment.
Fig. 2 is a flow diagram illustrating a method for user service policy determination based on federal learning in accordance with an exemplary embodiment.
Fig. 3 is a flow chart illustrating a method for user service policy determination based on federal learning in accordance with another exemplary embodiment.
Fig. 4 is a schematic diagram illustrating a user service policy determination method based on federal learning in accordance with another exemplary embodiment.
Fig. 5 is a flow chart illustrating a method for user service policy determination based on federal learning in accordance with another exemplary embodiment.
Fig. 6 is a block diagram illustrating a federated learning-based user service policy determination mechanism in accordance with another exemplary embodiment.
FIG. 7 is a block diagram illustrating an electronic device in accordance with an example embodiment.
FIG. 8 is a block diagram illustrating a computer-readable medium in accordance with an example embodiment.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and a repetitive description thereof will be omitted.
Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the application. One skilled in the relevant art will recognize, however, that the subject matter of the present application can be practiced without one or more of the specific details, or with other methods, components, devices, steps, and so forth. In other instances, well-known methods, devices, implementations, or operations have not been shown or described in detail to avoid obscuring aspects of the application.
The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. I.e. these functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor means and/or microcontroller means.
The flow charts shown in the drawings are merely illustrative and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.
It will be understood that, although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one element from another. Thus, a first component discussed below may be termed a second component without departing from the teachings of the present concepts. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It will be appreciated by those skilled in the art that the drawings are merely schematic representations of exemplary embodiments, and that the blocks or processes shown in the drawings are not necessarily required to practice the present application and are, therefore, not intended to limit the scope of the present application.
The technical terms involved in the present application are explained as follows:
and longitudinal federated learning, namely a distributed machine learning training mode, wherein the training of the model can be completed by combining two clients under the condition of not transmitting original data. Longitudinal federal learning is generally that the feature side provides feature data and the tag side provides tag data and additional feature data.
Multi-party federal learning: beyond federal learning by agencies.
Homomorphic encryption, which is a special encryption method and can perform algebraic operation on ciphertext data, wherein the result obtained by the operation after ciphertext decryption is the same as the result obtained by directly calculating on plaintext data.
Multi-party secure computation: the method is a method for combining a plurality of mutually untrusted label parties to safely complete the calculation of a function together.
Information amount (information value, IV): and is used for expressing the prediction capability of the characteristic on the target prediction, and the higher the IV value is, the stronger the prediction capability of the characteristic is.
After the applicant of the present application has conducted an in-depth analysis on the prior art, it is considered that in multi-party federal learning, it is generally necessary to coordinate a plurality of participants to perform calculation safely. The calculation can be realized by designing a safe multiparty calculation protocol, the arithmetic operators of algebraic operation are abstracted by the multiparty interactive calculation protocol, and then the arithmetic operators are utilized to design an algorithm model. And the data security in the model iteration process can be ensured by the improvement of the model algorithm level. The first method is generally higher in complexity, and is easily affected by bandwidth and computing resources in a multi-party environment, and in an industrial scenario, different companies have different complexities on the flow of data auditing, and often it is difficult to directly coordinate. The second method is generally based on security technologies such as semi-homomorphic encryption, and two parties with higher maturity have federal learning in practical application, and this method is generally limited to two parties, for example, one party provides a label, and the other party provides a scenario of features.
The applicant of the present application thinks that in the prior art, the federal learning technology performed by two parties is mature and uncertain, and expanding on the basis of federal learning of two parties is a mode with high reliability and feasibility. In the application, the two-party federal learning in the prior art is expanded, the scheme in the application can be applied to the scene of two-party to multi-party combined learning, the technical method in the application has low requirements on resources such as calculation, bandwidth and the like, and has stronger compatibility in actual deployment and stronger robustness to data loss and other conditions.
The content of the application is explained below with the aid of specific examples.
Fig. 1 is a system block diagram illustrating a method and apparatus for user service policy determination based on federal learning according to an exemplary embodiment.
As shown in fig. 1, the system architecture 10 may include terminal devices 101, 102, 103, a network 104 and a tagger server 105, feature side servers 106, 107, 108. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the tagger server 105 and the feature server 106, 107, 108. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The user may use the terminal devices 101, 102, 103 to interact with the tagger server 105, the feature side servers 106, 107, 108 over the network 104 to receive or send messages, etc. Various communication client applications, such as an internet service application, a shopping application, a web browser application, an instant messaging tool, a mailbox client, social platform software, and the like, may be installed on the terminal devices 101, 102, and 103.
The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The tager server 105 and the feature server 106, 107, 108 may be servers that provide various services, such as a backend management server that supports internet service-like websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze the received user data, and feed back the processing result to the administrator of the internet service website and/or the terminal device 101, 102, 103.
Any one or more of tagger server 105, characterizer server 106, characterizer server 107, characterizer server 108 may be referred to as a participant server.
The tagger server 105 may, for example, encrypt user tags for a plurality of users to generate encrypted information; the feature side servers 106, 107, 108 may generate feature information amounts of their corresponding user features, for example, from the encryption information; the feature information quantities of the feature side servers 106, 107 and 108 are sequentially subjected to federal learning with the label side server 105, and a user scoring model is generated after training is completed.
The participant server (tagger server 105 and/or characterizer server 106 and/or characterizer server 107 and/or characterizer server 108) may generate user characteristics by the terminal device 101, 102, 103 obtaining user data of the current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scores; and comparing the user score with a preset interval to determine the service strategy of the current user and push the service strategy.
The participant server may be a server of one entity, or may be, for example, a plurality of servers, a portion of which may be used to provide a service policy for a user in response to a user request; a portion of the participant servers may also be federated machine learning trained, e.g., in pairwise associations.
It should be noted that the method for determining the user service policy based on the federal learning provided in the embodiment of the present application may be executed by the tag side server 105 and/or the feature side servers 106, 107, 108, and accordingly, the device for determining the user service policy based on the federal learning may be disposed in the tag side server 105 and/or the feature side servers 106, 107, 108. And the web page end for providing internet service platform browsing for the user is generally located in the terminal equipment 101, 102, 103.
Fig. 2 is a flow chart illustrating a method for user service policy determination based on federal learning in accordance with an exemplary embodiment. The federally learned user service policy determination method 20 describes a process of providing a user policy for a user in a practical application scenario, and includes at least steps S202 to S206.
As shown in fig. 2, in S202, a user profile is generated from user data of the current user.
In the embodiment of the application, the current user and the user can be individual users, small and micro enterprise users, enterprise users and the like, wherein the user data can include basic information authorized by the user, such as service account information, user terminal equipment identification information, user region information and the like; the user information may also include behavior information, which may be, for example, page operation data of the user, service access duration of the user, service access frequency of the user, and the like, and specific content of the user information may be determined according to an actual application scenario, which is not limited herein.
The user characteristics may be generated from the user data, and more particularly, the user data may be converted into a numerical form according to different attribute categories as the user characteristics.
A plurality of feature information may be generated based on the user information and a feature policy. The data cleaning and data fusion can be carried out on the user information so as to convert the user information into a plurality of characteristic data, and more particularly, the variable loss rate analysis and processing and abnormal value processing can be carried out on the user information; and the user information discretized by the continuous variable can be subjected to WOE conversion, discrete variable WOE conversion, text variable processing, text variable word2vec processing and the like.
The method can comprehensively consider in many aspects such as variable coverage, single value coverage, correlation and significance with the target variable, the distinguishing degree (KS) and Information Value (IV) of the target variable, the characteristic importance of tree models (such as XGboost, RF and the like), and the like, and screen the characteristics with high coverage and obvious distinguishing effect on the target variable as the user characteristics.
In S204, the user characteristics are input into a user scoring model generated through multi-party federal learning, so as to obtain a user score. It is worth mentioning that the user scoring model used in the present application may be a label party for model training in multi-party federal learning, or a feature party, which may be referred to as a participant of model training hereinafter.
In one embodiment, the user scoring model is generated through multi-party federal learning training conducted by multiple data agencies. As described above, it is almost impossible to integrate data scattered in various regions and organizations in reality, machine learning models can be trained among the organizations by federal learning, user data can be protected by modeling in a federal learning scenario, and user data interacted among the organizations is encrypted.
In the application, through multi-party longitudinal federal learning, a plurality of participants are coordinated to participate in model training, and model training in a supervision mode is realized under the condition that labels are not leaked, so that a user scoring model is generated.
In one embodiment, the machine learning model trained in multi-party federated learning may be a decision tree model, a gradient boosting decision tree model, a neural network model, a convolutional neural network model, and the like, which are not limited in this application.
In S206, the user score is compared with a preset interval to determine the service policy of the current user and push the service policy. The user service policy may be targeted first, and may for example be resource allocation for the user or provide the user with an ad hoc resource. According to the target of the user service strategy, extracting a statistical analysis value aiming at the target from user data of a large number of historical users so as to generate a plurality of preset intervals. And making different user service strategies for users in different user intervals.
According to the user service strategy determining method based on the federal learning, user labels of a plurality of users are encrypted through a label party trained by a model to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encryption information; the at least one characteristic party performs model training according to characteristic information quantity and the label party on the basis of federal learning in sequence to generate a user scoring model; generating user characteristics according to user data of a current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scoring; the user scores are compared with the preset interval to determine the service strategy of the user and push the service strategy, so that the user can be comprehensively and accurately evaluated on the premise of guaranteeing user data safety, system data safety and transaction safety, and the most appropriate service strategy can be quickly and accurately provided for the user.
It should be clearly understood that this application describes how to make and use particular examples, but the principles of this application are not limited to any details of these examples. Rather, these principles can be applied to many other embodiments based on the teachings of the present disclosure.
Fig. 3 is a flow chart illustrating a method for user service policy determination based on federal learning in accordance with another exemplary embodiment. The federal-learning-based user service policy determination method 30 describes a process of generating a user scoring model through federal learning, and may include steps S302 to S306.
As shown in fig. 3, in S302, the labeler for model training encrypts user labels of a plurality of users to generate encrypted information.
In one embodiment, the label party of the model training may generate an encryption key, for example, by an additive homomorphic encryption manner; encrypting the user tags of the plurality of users by the encryption key to generate encrypted tags; and the label party trained by the model sends the encryption key and the encryption label as encryption information to the at least one characteristic party. In the application, the addition homomorphic encryption mode can realize model training in a supervision mode under the condition that the label is not leaked.
In S304, at least one feature side of the model training generates feature information amount of the user feature corresponding to the feature information amount according to the encryption information.
In one embodiment, each feature party of model training performs binning processing on the corresponding user features thereof to generate a plurality of binning feature sets; generating a plurality of box information quantities of a plurality of box characteristic sets according to the encrypted information; and each feature party of the model training generates feature information quantity according to the plurality of corresponding box information quantities.
More specifically, each feature side of the model training divides the corresponding user features according to a decision tree binning mode to generate a plurality of dividing points; and performing binning processing on the user features according to the plurality of segmentation points to generate a plurality of binning feature sets.
More specifically, feature values in the plurality of box-dividing feature sets are encrypted and calculated according to encryption keys in the encryption information to generate encrypted values; generating the plurality of binned information amounts by encrypting the encrypted values in the plurality of binned feature sets after the calculation. The characteristic values can be encrypted based on a homomorphic encryption key sent by the tag party, and each characteristic value generates a corresponding encryption value.
Each feature side generates a feature information amount according to an average value of information amounts of a plurality of corresponding sub-boxes.
In S306, the at least one feature party performs model training with the label party based on federal learning in turn according to the feature information amount to generate a user rating model. The at least one feature party can be sorted according to the feature information amount corresponding to the feature party, for example; and the at least one characteristic party performs model training with the label party based on federal learning in sequence according to the sequence until all the characteristic parties finish training, and generates a user scoring model.
In one embodiment, an initial feature party may be extracted by the at least one feature party according to the ranking; the initial characteristic party performs federal learning through the corresponding user characteristics, the encryption information and the label party to generate a plurality of first scores; extracting a next feature party of the initial feature party as a current feature party according to the sorting; the current characteristic party performs federal learning according to the corresponding user characteristic, the encrypted label, the plurality of first scores and the label party to generate a plurality of second scores; and sequentially extracting the next feature party and the next label party according to the sequence, and performing model training based on federal learning until all feature parties finish training to generate the user scoring model.
The specific content of the "at least one feature party performs model training in sequence according to the ranking and the label party based on federal learning until all feature parties finish training to generate the user scoring model" is described in detail in the embodiments corresponding to fig. 4 and 5.
According to the user service strategy determining method based on the federal learning, aggregation of a plurality of participants can be decomposed into aggregation of a plurality of parties (federal learning of two parties) under a multi-party federal learning scene, respective data elements are fused, a characteristic party with high IV and a label party are selected in sequence to train a stable machine learning model through calculating the average IV value of the characteristic party, and a final model training result is formed through stacking training in sequence based on a single machine learning model.
The user service strategy determining method based on the federal learning is expanded on the basis of the existing mature federal learning of two parties, the threshold of hardware resources and the coupling degree of data of each party are effectively reduced, and the condition that in practical application, the training of the whole model is invalid due to the fact that one tag party acquires data overtime is avoided.
Fig. 4 is a schematic diagram illustrating a user service policy determination method based on federal learning in accordance with another exemplary embodiment. Fig. 4 is a schematic diagram 40 illustrating a detailed description of "the at least one feature party performs model training according to the sequence and the label party performs model training based on federal learning in turn until all feature parties have completed training and the user scoring model" in S306 in the flow illustrated in fig. 3.
As shown in FIG. 4, for example, tagger M may be a provider of user tags, and featurers N1, N2, N3 may be providers of user feature data.
The labeler M encrypts the user label and sends the encrypted user label and the encrypted key to the feature party N1, the feature party N2 and the feature party N3; the feature party N1, the feature party N2 and the feature party N3 are respectively split according to existing user features and a splitting mode of decision tree types to generate a plurality of sub-box feature sets, then the encryption value of each sub-box feature set is calculated through an encryption key, and finally, the feature party N1, the feature party N2 and the feature party N3 respectively obtain feature information quantity according to the encryption value corresponding to each sub-box feature set corresponding to the feature party N1, the feature party N2 and the feature party N3.
The feature side N1, the feature side N2, and the feature side N3 share feature information quantities for ranking, first extract a feature side (as shown in the figure, N2) with the highest feature information quantity, perform model training on the feature side N2 and the tagger M, and generate first scores of multiple users in the tagger M based on an initial model obtained by training.
And taking the next feature party of the initial feature party as a current feature party (as shown in the figure, the feature party can be N1), performing model training on the feature party N1 and the label party M, adjusting the model training process based on the first score, obtaining the current model after the training is finished, and generating second scores of a plurality of users in the label party M based on the current model.
And finally, performing model training on the feature party N3 and the label party M, adjusting in the training process based on the second score, obtaining a user score model after the training is finished, and generating third scores of a plurality of users in the label party M based on the user scores.
The trained user scoring model can be distributed to the tagger M, the feature side N1, the feature side N2 and the feature side N3 so as to score real-time users in practical application.
According to the user service strategy determining method based on the federal learning, the two-party federal learning method in the prior art is expanded, an addition homomorphic encryption mode is utilized, under the form of encrypting the label, the characteristic IV value of each characteristic party is firstly calculated, the characteristic parties are sequentially selected through the characteristic IV values so as to be convenient for pairwise training with the label parties, and a model training framework with a hierarchical structure is formed, so that the multi-party federal learning is realized in a simple and convenient mode.
Fig. 5 is a flow chart illustrating a method for federally learned user service policy determination in accordance with another exemplary embodiment. The flow 50 shown in fig. 5 is a detailed description of "the initial feature party generates a plurality of first scores by federately learning the corresponding user features, encrypted information, and tag parties in fig. 4".
As shown in fig. 5, in S502, the initial feature party encrypts the user feature according to the encryption key in the encryption information, so as to generate an encrypted feature.
In S504, the initial feature side and the label side perform federal learning by using the encryption feature and the encryption label in the encryption information.
In S506, an initial model is generated after training is completed.
Specifically, the machine learning model learning process for the label side and the feature side in the federal learning can be as follows: respectively constructing adjustment models, inputting encryption features and corresponding encryption labels into the adjustment models to obtain prediction labels, comparing the prediction labels with corresponding real labels, judging whether the prediction labels are consistent with the real labels, counting the number of the prediction labels consistent with the real labels, calculating the proportion of the number of the prediction labels consistent with the real labels in the number of all the prediction labels, converging the adjustment models if the proportion is larger than or equal to a preset proportion value to obtain the initial trained value, adjusting parameters in the adjustment models if the proportion is smaller than the preset proportion value, and predicting the prediction labels of the encryption features through the adjusted adjustment models until the proportion is larger than or equal to the preset proportion value. The method for adjusting the parameters in the adjustment model may be performed by using a random gradient descent algorithm, a gradient descent algorithm, or a normal equation.
If the times of adjusting the parameters of the adjusting model exceed the preset times, the model used for building the adjusting model can be replaced, so that the model training efficiency is improved.
In S508, a plurality of first scores of a plurality of users of the tag side are calculated according to the initial model. The model tagger may input user characteristics of a plurality of users into the initial model to generate a plurality of first scores.
In one embodiment, the current feature party performs federal learning according to the corresponding user feature, the encrypted tag, the plurality of first scores and the tag party, and generates a plurality of second scores, including: the current characteristic party encrypts the user characteristics according to the encryption key in the encryption information to generate encryption characteristics; the current characteristic party and the label party carry out federal learning through encrypted characteristics, encrypted labels in the encrypted information and the plurality of first scores; when training is finished, generating a current model; a plurality of second scores for a plurality of users of the tagger are calculated from the current model.
Specifically, the machine learning model learning process local to the label side and the second or later other feature side in federal learning can be as follows: and respectively constructing an adjustment model, and inputting the encryption characteristics and the corresponding encryption labels into the adjustment model to obtain the prediction labels. And weighting the user tags in the first score according to preset weight and the predicted tags to obtain the final output tags. And comparing the output label with the corresponding real label, judging whether the output label is consistent with the real label, and adjusting the parameters in the adjustment model according to the judgment until the training requirement is met to obtain the trained current model.
Those skilled in the art will appreciate that all or part of the steps implementing the above embodiments are implemented as computer programs executed by a CPU. When executed by the CPU, performs the functions defined by the methods provided herein. The program may be stored in a computer readable storage medium, which may be a read-only memory, a magnetic or optical disk, or the like.
Furthermore, it should be noted that the above-mentioned figures are only schematic illustrations of the processes involved in the method according to exemplary embodiments of the present application and are not intended to be limiting. It will be readily understood that the processes shown in the above figures are not intended to indicate or limit the chronological order of the processes. In addition, it is also readily understood that these processes may be performed, for example, synchronously or asynchronously in multiple modules.
The following are embodiments of the apparatus of the present application that may be used to perform embodiments of the method of the present application. For details which are not disclosed in the embodiments of the apparatus of the present application, reference is made to the embodiments of the method of the present application.
Fig. 6 is a block diagram illustrating a federated learning-based user service policy determination apparatus in accordance with another exemplary embodiment. As shown in fig. 6, the federally-learned user service policy determination apparatus 60 includes: a cryptographic module 602, a features module 604, a training module 606, a features module 608, a scoring module 610, and a policy module 612.
The encryption module 602 is configured to encrypt user tags of multiple users to generate encrypted information by a control module trained by a tagging party; the encryption module 602 is further configured to generate an encryption key by the label party for model training through an addition homomorphic encryption manner; encrypting the user tags of the plurality of users by the encryption key to generate encrypted tags; and the label party trained by the model sends the encryption key and the encryption label as encryption information to the at least one characteristic party.
The feature module 604 is configured to control at least one feature party for model training to generate a feature information amount of a user feature corresponding to the feature party according to the encrypted information; the feature module 604 is further configured to perform binning processing on the user features corresponding to each feature party of the model training to generate a plurality of binning feature sets; generating a plurality of box information quantities of a plurality of box characteristic sets according to the encrypted information; and each characteristic party of the model training generates characteristic information quantity according to the corresponding plurality of sub-box information quantities.
The training module 606 is used for the at least one feature party to perform model training with the label party based on federal learning in sequence according to the feature information quantity so as to generate a user scoring model; the training module 606 is further configured to rank the at least one feature party according to a feature information amount corresponding to the feature party; and the at least one characteristic party performs model training according to the sequence and the label party on the basis of federal learning, and a user scoring model is generated until all characteristic parties are trained.
The feature module 608 is configured to generate a user feature according to user data of a current user;
the scoring module 610 is configured to input the user characteristics into a user scoring model generated through multi-party federal learning, so as to obtain a user score;
the policy module 612 is configured to compare the user score with a preset interval, so as to determine a service policy of the current user and perform pushing.
According to the user service strategy determining device based on federal learning, user labels of a plurality of users are encrypted through a label party trained by a model to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encrypted information; the at least one characteristic party performs model training sequentially with the label party based on federal learning according to the characteristic information quantity to generate a user grading model; generating user characteristics according to user data of a current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scoring; the user score is compared with a preset interval to determine the service strategy of the user and push the service strategy, so that the user can be comprehensively and accurately evaluated on the premise of guaranteeing user data safety, system data safety and transaction safety, and the most appropriate service strategy can be quickly and accurately provided for the user.
FIG. 7 is a block diagram of an electronic device shown in accordance with an example embodiment.
An electronic device 700 according to this embodiment of the present application is described below with reference to fig. 7. The electronic device 700 shown in fig. 7 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present application.
As shown in fig. 7, electronic device 700 is embodied in the form of a general purpose computing device. The components of the electronic device 700 may include, but are not limited to: at least one processing unit 710, at least one memory unit 720, a bus 730 that connects the various system components (including the memory unit 720 and the processing unit 710), a display unit 740, and the like.
Wherein the storage unit stores program code, which can be executed by the processing unit 710, to cause the processing unit 710 to execute the steps according to various exemplary embodiments of the present application in the present specification. For example, the processing unit 710 may perform the steps as shown in fig. 2, 3, 5.
The memory unit 720 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM) 7201 and/or a cache memory unit 7202, and may further include a read only memory unit (ROM) 7203.
The memory unit 720 may also include a program/utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination thereof may comprise an implementation of a network environment.
The electronic device 700 may also communicate with one or more external devices 700' (e.g., keyboard, pointing device, bluetooth device, etc.), such that a user can communicate with devices with which the electronic device 700 interacts, and/or any devices (e.g., router, modem, etc.) with which the electronic device 700 can communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 750. Also, the electronic device 700 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the internet) via the network adapter 760. The network adapter 760 may communicate with other modules of the electronic device 700 via the bus 730. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, as shown in fig. 8, the technical solution according to the embodiment of the present application may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present application.
The software product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In situations involving remote computing devices, the remote computing devices may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to external computing devices (e.g., through the internet using an internet service provider).
The computer readable medium carries one or more programs which, when executed by a device, cause the computer readable medium to perform the functions of: a label party for model training encrypts user labels of a plurality of users to generate encrypted information; at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encryption information; the at least one characteristic party performs model training sequentially with the label party based on federal learning according to the characteristic information quantity to generate a user grading model; generating user characteristics according to user data of a current user; inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scoring; and comparing the user score with a preset interval to determine the service strategy of the user and push the service strategy.
Those skilled in the art will appreciate that the modules described above may be distributed in the apparatus according to the description of the embodiments, or may be modified accordingly in one or more apparatuses unique from the embodiments. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiment of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present application.
Exemplary embodiments of the present application are specifically illustrated and described above. It is to be understood that the application is not limited to the details of construction, arrangement or method of operation set forth herein; on the contrary, the application is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.
Claims (10)
1. A method for determining user service based on federal learning is characterized by comprising the following steps:
a label party for model training encrypts user labels of a plurality of users to generate encrypted information;
at least one characteristic party of the model training generates characteristic information quantity of the corresponding user characteristic according to the encryption information;
sorting the at least one feature party according to the feature information amount corresponding to the feature party;
extracting an initial feature party by the at least one feature party according to the ranking;
the initial characteristic party performs federal learning through the corresponding user characteristics, the encryption information and the label party to generate a plurality of first scores;
extracting a next feature party of the initial feature party as a current feature party according to the sequence;
the current characteristic party performs federal learning according to the corresponding user characteristic, the encrypted label, the plurality of first scores and the label party to generate a plurality of second scores;
sequentially extracting the next feature party and the label party of the current feature party according to the sequence, and carrying out model training based on federal learning until all feature parties finish training to generate a user scoring model;
generating user characteristics according to user data of a current user;
inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scores;
and comparing the user score with a preset interval to determine the service strategy of the current user and push the service strategy.
2. The method of claim 1, wherein the model-trained tagger encrypts user tags for a plurality of users to generate encrypted information, comprising:
generating an encryption key by a label party of the model training in an addition homomorphic encryption mode;
encrypting the user tags of the plurality of users by the encryption key to generate encrypted tags;
and the label party trained by the model sends the encryption key and the encryption label as encryption information to the at least one characteristic party.
3. The method of claim 1, wherein at least one characterizer of model training generates the characteristic information quantity of the corresponding user characteristic thereof according to the encryption information, and the method comprises the following steps:
each feature side of model training carries out binning processing on the corresponding user features to generate a plurality of binning feature sets;
generating a plurality of box information quantities of a plurality of box characteristic sets according to the encrypted information;
and each characteristic party of the model training generates characteristic information quantity according to the corresponding plurality of sub-box information quantities.
4. The method of claim 3, wherein each eigen party of the model training performs binning processing on its corresponding user features to generate a plurality of binned feature sets, including:
each feature side of the model training divides the corresponding user features according to a decision tree box dividing mode to generate a plurality of dividing points;
and performing binning processing on the user characteristics according to the plurality of segmentation points to generate a plurality of binning characteristic sets.
5. The method of claim 3, wherein generating a plurality of binned quantities of information for a plurality of binned feature sets from the encrypted information comprises:
encrypting the characteristic values in the plurality of sub-box characteristic sets according to an encryption key in the encryption information to generate an encrypted value;
generating the plurality of binned information quantities by encrypting values in the plurality of binned feature sets after the cryptographic computation.
6. The method of claim 1, wherein the initial feature party performs federal learning by its corresponding user features, encryption information and tag parties to generate a plurality of first scores, comprising:
the initial characteristic party encrypts the user characteristics according to the encryption key in the encryption information to generate encryption characteristics;
the initial characteristic party and the label party carry out federal learning through the encryption characteristics and the encryption labels in the encryption information;
when training is finished, generating an initial model;
a plurality of first scores for a plurality of users of a tagger are calculated from the initial model.
7. The method of claim 1, wherein the current feature party performs federal learning based on the corresponding user feature, encrypted tag, the plurality of first scores, and tag party to generate a plurality of second scores, comprising:
the current characteristic party encrypts the user characteristics according to the encryption key in the encryption information to generate encryption characteristics;
the current characteristic party and the label party carry out federal learning through encrypted characteristics, encrypted labels in the encrypted information and the plurality of first scores;
when training is finished, generating a current model;
a plurality of second scores for a plurality of users of the tag party are calculated according to the current model.
8. A federally-learned user service policy determination apparatus, comprising:
the encryption module is used for encrypting the user labels of a plurality of users to generate encrypted information by controlling the label party for model training;
the characteristic module is used for controlling at least one characteristic party of the model training to generate the characteristic information quantity of the corresponding user characteristic according to the encrypted information;
the training module is used for sequencing the at least one characteristic party according to the characteristic information quantity corresponding to the characteristic party; extracting an initial feature party by the at least one feature party according to the ranking; the initial characteristic party performs federal learning through the corresponding user characteristics, the encryption information and the label party to generate a plurality of first scores; extracting a next feature party of the initial feature party as a current feature party according to the sorting; the current characteristic party performs federal learning according to the corresponding user characteristic, the encrypted label, the plurality of first scores and the label party to generate a plurality of second scores; sequentially extracting the next feature party and the label party of the current feature party according to the sequence, and carrying out model training based on federal learning until all feature parties finish training to generate a user scoring model;
the characteristic module is used for generating user characteristics according to the user data of the current user;
the scoring module is used for inputting the user characteristics into a user scoring model generated through multi-party federal learning to obtain user scores;
and the strategy module is used for comparing the user score with a preset interval so as to determine the service strategy of the current user and push the service strategy.
9. An electronic device, comprising:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method recited in any of claims 1-7.
10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210499538.6A CN114611008B (en) | 2022-05-09 | 2022-05-09 | User service strategy determination method and device based on federal learning and electronic equipment |
| PCT/CN2022/121544 WO2023216494A1 (en) | 2022-05-09 | 2022-09-27 | Federated learning-based user service strategy determination method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210499538.6A CN114611008B (en) | 2022-05-09 | 2022-05-09 | User service strategy determination method and device based on federal learning and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114611008A CN114611008A (en) | 2022-06-10 |
| CN114611008B true CN114611008B (en) | 2022-07-22 |
Family
ID=81867895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210499538.6A Active CN114611008B (en) | 2022-05-09 | 2022-05-09 | User service strategy determination method and device based on federal learning and electronic equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN114611008B (en) |
| WO (1) | WO2023216494A1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114611008B (en) * | 2022-05-09 | 2022-07-22 | 北京淇瑀信息科技有限公司 | User service strategy determination method and device based on federal learning and electronic equipment |
| CN116823346B (en) * | 2023-07-17 | 2024-04-05 | 广州百奕信息科技有限公司 | Civil aviation digital asset public comprehensive service method based on blockchain technology |
| CN117560229B (en) * | 2024-01-11 | 2024-04-05 | 吉林大学 | Federal non-intrusive load monitoring user verification method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111563267A (en) * | 2020-05-08 | 2020-08-21 | 京东数字科技控股有限公司 | Method and device for processing federal characteristic engineering data |
| CN112396189A (en) * | 2020-11-27 | 2021-02-23 | 中国银联股份有限公司 | Method and device for multi-party construction of federal learning model |
| WO2021114927A1 (en) * | 2019-12-11 | 2021-06-17 | 支付宝(杭州)信息技术有限公司 | Method and apparatus for multiple parties jointly performing feature assessment to protect privacy security |
| WO2022016964A1 (en) * | 2020-07-23 | 2022-01-27 | 深圳前海微众银行股份有限公司 | Vertical federated modeling optimization method and device, and readable storage medium |
| CN114139450A (en) * | 2021-11-29 | 2022-03-04 | 百保(上海)科技有限公司 | Scoring card modeling method and device based on privacy protection |
| CN114372871A (en) * | 2022-01-07 | 2022-04-19 | 中国工商银行股份有限公司 | Method and device for determining credit score value, electronic device and storage medium |
| CN114401079A (en) * | 2022-03-25 | 2022-04-26 | 腾讯科技（深圳）有限公司 | Multi-party joint information value calculation method, related equipment and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019175032A1 (en) * | 2018-03-14 | 2019-09-19 | Koninklijke Philips N.V. | Scoring intravascular lesions and stent deployment in medical intraluminal ultrasound imaging |
| KR102510513B1 (en) * | 2020-07-27 | 2023-03-16 | 한국전자통신연구원 | Deep learning based beamforming method and apparatus for the same |
| CN114418380A (en) * | 2022-01-17 | 2022-04-29 | 建信金融科技有限责任公司 | User scoring method based on federal learning, server and storage medium |
| CN114330759B (en) * | 2022-03-08 | 2022-08-02 | 富算科技(上海)有限公司 | Training method and system for longitudinal federated learning model |
| CN114611008B (en) * | 2022-05-09 | 2022-07-22 | 北京淇瑀信息科技有限公司 | User service strategy determination method and device based on federal learning and electronic equipment |
-
2022
- 2022-05-09 CN CN202210499538.6A patent/CN114611008B/en active Active
- 2022-09-27 WO PCT/CN2022/121544 patent/WO2023216494A1/en unknown
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021114927A1 (en) * | 2019-12-11 | 2021-06-17 | 支付宝(杭州)信息技术有限公司 | Method and apparatus for multiple parties jointly performing feature assessment to protect privacy security |
| CN111563267A (en) * | 2020-05-08 | 2020-08-21 | 京东数字科技控股有限公司 | Method and device for processing federal characteristic engineering data |
| WO2022016964A1 (en) * | 2020-07-23 | 2022-01-27 | 深圳前海微众银行股份有限公司 | Vertical federated modeling optimization method and device, and readable storage medium |
| CN112396189A (en) * | 2020-11-27 | 2021-02-23 | 中国银联股份有限公司 | Method and device for multi-party construction of federal learning model |
| CN114139450A (en) * | 2021-11-29 | 2022-03-04 | 百保(上海)科技有限公司 | Scoring card modeling method and device based on privacy protection |
| CN114372871A (en) * | 2022-01-07 | 2022-04-19 | 中国工商银行股份有限公司 | Method and device for determining credit score value, electronic device and storage medium |
| CN114401079A (en) * | 2022-03-25 | 2022-04-26 | 腾讯科技（深圳）有限公司 | Multi-party joint information value calculation method, related equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114611008A (en) | 2022-06-10 |
| WO2023216494A1 (en) | 2023-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114611008B (en) | User service strategy determination method and device based on federal learning and electronic equipment | |
| Grover et al. | Perceived usefulness, ease of use and user acceptance of blockchain technology for digital transactions–insights from user-generated content on Twitter | |
| CN109284313B (en) | Federal modeling method, device and readable storage medium based on semi-supervised learning | |
| US11196541B2 (en) | Secure machine learning analytics using homomorphic encryption | |
| WO2021179720A1 (en) | Federated-learning-based user data classification method and apparatus, and device and medium | |
| Fang et al. | Privacy wizards for social networking sites | |
| Javed et al. | Bidirectional absolute GRA/GIA model for uncertain systems: application in project management | |
| CN111967615B (en) | Multi-model training method and device based on feature extraction, electronic equipment and medium | |
| CN111563267B (en) | Method and apparatus for federal feature engineering data processing | |
| CN111553744A (en) | Federal product recommendation method, device, equipment and computer storage medium | |
| WO2024082514A1 (en) | Service index prediction method and apparatus, and device and storage medium | |
| Zhang | Corporate accounting information disclosure based on FPGA and neural network | |
| Igbekele et al. | Research trends on CAPTCHA: A systematic literature | |
| Althati et al. | Scalable Machine Learning Solutions for Heterogeneous Data in Distributed Data Platform | |
| CN116502732B (en) | Federal learning method and system based on trusted execution environment | |
| CN112734050A (en) | Text model training method, text model recognition device, text model equipment and storage medium | |
| CN113032838B (en) | Label prediction model generation method, prediction method, model generation device, system and medium based on privacy calculation | |
| CN111553743A (en) | Federal product recommendation method, device, equipment and computer storage medium | |
| CN114091815A (en) | Resource request processing method, device and system and electronic equipment | |
| Mitrovic et al. | Cybersecurity Culture as a critical component of Digital Transformation and Business Model Innovation in SMEs | |
| CN113902545A (en) | Resource limit distribution method and device and electronic equipment | |
| Sumathi et al. | Scale-based secured sensitive data storage for banking services in cloud | |
| CN112598127A (en) | Federal learning model training method and device, electronic equipment, medium and product | |
| CN112508075B (en) | DBSCAN clustering method based on transverse federation and related equipment thereof | |
| Dyussenov et al. | E-Government Policy Implementation in Thailand: Success or Failure? |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |