CN115375458A - Model training method and device, storage medium and electronic equipment - Google Patents
Model training method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN115375458A CN115375458A CN202211042961.XA CN202211042961A CN115375458A CN 115375458 A CN115375458 A CN 115375458A CN 202211042961 A CN202211042961 A CN 202211042961A CN 115375458 A CN115375458 A CN 115375458A
- Authority
- CN
- China
- Prior art keywords
- model
- sample data
- intermediate result
- party
- gradient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a model training method and device, a storage medium and electronic equipment, which can be applied to the financial field or other fields and can acquire a first model and first sample data; the first sample data includes financial behavior information of the user; generating a first intermediate result from the first model and the first sample data; generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by the second party; generating a second intermediate result by a second participant according to the second model and second sample data; the second sample data includes communication behavior information of the user; sending the first encryption gradient to an intermediate party of the federal learning system to obtain a first gradient; model parameters of the first model are updated according to the first gradient. The evaluation accuracy of the credit risk can be improved.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a model training method and apparatus, a storage medium, and an electronic device.
Background
Currently, with the development of economic society, financial products of banks are rapidly developed. While the bank provides credit service, the bank needs to evaluate the credit risk of the information of the credit customer so as to reduce the default risk of the customer.
The existing credit risk mode is evaluated through characteristics in bank data, however, the data characteristics in the bank cannot comprehensively reflect the risk condition of a user, and the credit risk evaluation accuracy is low.
Disclosure of Invention
The invention aims to provide a model training method which can improve the evaluation accuracy of credit risk.
The invention also provides a model training device for ensuring the realization and the application of the method in practice.
A model training method, comprising:
a first party for use in a federated learning system, the method comprising:
acquiring a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
generating a first intermediate result from the first model and the first sample data;
generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition;
sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
and updating the model parameters of the first model according to the first gradient obtained by decryption of the middle party so as to realize training of the first model.
The method, optionally, includes a process of acquiring the first sample data, including:
obtaining original sample data;
and preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
The method described above, optionally, after generating the first intermediate result according to the first model and the first sample data, further includes:
encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result;
sending the first encrypted intermediate result to the second party.
Optionally, the method further includes, after the updating the model parameter of the first model according to the first gradient obtained by the decryption by the intermediary, that:
judging whether the updated first model meets a preset training completion condition or not;
under the condition that the updated first model meets a preset training completion condition, performing risk monitoring on a target user by using the trained first model;
and under the condition that the updated first model does not meet the preset training completion condition, acquiring new first sample data, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
A model training apparatus for use with a first party to a federal learning system, the apparatus comprising:
an acquisition unit configured to acquire a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
a first generating unit, configured to generate a first intermediate result according to the first model and the first sample data;
a second generating unit, configured to generate a first encryption gradient of the first model according to the first intermediate result and the received second encrypted intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition;
the transmission unit is used for sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
and the updating unit is used for updating the model parameters of the first model according to the first gradient obtained by the decryption of the middle party so as to realize the training of the first model.
The above apparatus, optionally, the obtaining unit includes:
the acquisition subunit is used for acquiring original sample data;
and the preprocessing unit is used for preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
The above apparatus, optionally, further comprises:
the encryption unit is used for encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result;
a sending unit, configured to send the first encrypted intermediate result to the second party.
The above apparatus, optionally, further comprises:
the judging unit is used for judging whether the updated first model meets a preset training completion condition or not;
the first execution unit is used for monitoring risks of a target user by using the trained first model under the condition that the updated first model meets a preset training completion condition;
and the second execution unit is used for acquiring new first sample data under the condition that the updated first model does not meet the preset training completion condition, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
A storage medium comprising stored instructions, wherein the instructions, when executed, control a device on which the storage medium resides to perform a model training method as described above.
An electronic device comprising a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by one or more processors to perform the model training method as described above.
Compared with the prior art, the invention has the following advantages:
the invention provides a model training method and device, a storage medium and electronic equipment, wherein the method comprises the following steps: acquiring a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records; generating a first intermediate result from the first model and the first sample data; generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition; sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient; and updating the model parameters of the first model according to the first gradient obtained by decryption of the intermediate party so as to realize the training of the first model. By applying the method provided by the embodiment of the invention, the model can be trained through the financial behavior information and the communication behavior information of the user, and the evaluation accuracy of the credit risk can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a method of model training according to the present invention;
FIG. 2 is a flow chart of a process for obtaining first sample data according to the present invention;
FIG. 3 is a flow chart of a method of another model training method provided by the present invention; (ii) a
FIG. 4 is a schematic structural diagram of a model training apparatus according to the present invention;
fig. 5 is a schematic structural diagram of an electronic device provided in the present invention;
fig. 6 is a schematic structural diagram of a federated learning system provided in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
The federated learning is a distributed machine learning framework, the core of the federated learning is to solve the conflict between data islands and data privacy protection, and participating parties can benefit from the federated learning by establishing a data federation under the condition of not sharing data, so that the overall continuous progress of the technology is promoted. Longitudinal federated learning refers to that under the condition that users participating in data sets of all parties of the joint modeling are overlapped more and user features are overlapped less, the data sets are segmented according to feature dimensions, and data with the same users and different user features are extracted for training. The longitudinal federal learning is often used for solving the problem that one party has too few data dimensions, and the modeling target cannot be well realized only by using one party data, or one party only has a Y label and needs to use the characteristics of other participants to construct a scene of a joint model, and the longitudinal federal learning is mostly used for joint modeling among different industries.
For example, in the joint modeling between banks and operators, the intersection of users is large, but the bank records are financial behavior information such as the balance information, credit behavior, repayment records and the like of the users, and the operators have communication behavior information such as user short messages, internet surfing conditions and the like, and the information content of the model is improved through characteristic complementation between the participants so as to enhance the identification and prediction capabilities of the joint model.
The embodiment of the invention provides a model training method, which can be applied to electronic equipment, wherein the electronic equipment can be a first participant of a federal learning system, and a flow chart of the method is shown in fig. 1, and specifically comprises the following steps:
s101: acquiring a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information includes at least one of receipt and payment information, credit information, and a payment record.
In this embodiment, the first model may be a linear regression model LR, an X-gboost model, or the like.
S102: a first intermediate result is generated from the first model and the first sample data.
Alternatively, the first sample data may be input into the first model to obtain a first intermediate result, which may be a characteristic weighting value of the first model.
S103: generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of short message information of the user and internet surfing condition.
In this embodiment, the second intermediate result may be a characteristic weighted value calculated by the second model based on the second sample data.
In some alternative embodiments, the noise in the second intermediate result may be removed, and then the first encryption gradient may be calculated based on the first intermediate result and the second intermediate result after the noise is removed.
S104: and sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain the first gradient.
In this embodiment, the intermediate party may decrypt the first encryption gradient according to a preset decryption key to obtain the first gradient, and then send the first gradient to the first participant.
S105: and updating the model parameters of the first model according to the first gradient obtained by decryption of the middle party so as to realize training of the first model.
The invention provides a bank stock client wind control method based on longitudinal federal learning, which improves the information quantity of a model through characteristic complementation among participants so as to enhance the identification and prediction capabilities of a combined model, thereby protecting the privacy of users and improving the risk early warning precision of an algorithm.
In an embodiment provided by the present invention, based on the implementation process, specifically, the process of acquiring the first sample data, as shown in fig. 2, includes:
s201: and acquiring original sample data.
S202: and preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
In this embodiment, the proximity algorithm may be a K proximity algorithm, and the original sample data may be repaired by preprocessing the original sample data through the proximity algorithm.
In an embodiment provided by the present invention, based on the foregoing implementation process, specifically, as shown in fig. 3, after the generating a first intermediate result according to the first model and the first sample data, the method further includes the following steps:
s106: and encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result.
S107: sending the first encrypted intermediate result to the second party.
In this embodiment, the first encrypted intermediate result is sent to the second party, so that the second party generates a second encryption gradient of the second model according to the second intermediate result and the first encrypted intermediate result; and sending the second encryption gradient to the intermediate party, decrypting the second encryption gradient by the intermediate party to obtain a second gradient, and updating the model parameter of the second model by the second participant according to the second gradient.
In an embodiment provided by the present invention, based on the foregoing implementation process, specifically, after the updating the model parameters of the first model according to the first gradient obtained by the decryption by the intermediary, the method further includes:
judging whether the updated first model meets a preset training completion condition or not;
under the condition that the updated first model meets a preset training completion condition, performing risk monitoring on a target user by using the trained first model;
and under the condition that the updated first model does not meet the preset training completion condition, acquiring new first sample data, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
In this embodiment, the training completion condition may be that the loss function of the first model converges, or that the training frequency of the first model is greater than a preset training frequency threshold, or that the precision of the first model meets a preset precision requirement.
Alternatively, the target user may be a user to be monitored.
In some embodiments, the risk monitoring may be performed on the target user by the trained first model and the trained second model together.
Corresponding to the method illustrated in fig. 1, an embodiment of the present invention further provides a model training apparatus, which is used for implementing the method illustrated in fig. 1 specifically, and the model training apparatus provided in the embodiment of the present invention may be applied to an electronic device, and a schematic structural diagram of the model training apparatus is illustrated in fig. 4, and specifically includes:
an obtaining unit 401 configured to obtain a first model and first sample data; the first sample data includes financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
a first generating unit 402, configured to generate a first intermediate result according to the first model and the first sample data;
a second generating unit 403, configured to generate a first encryption gradient of the first model according to the first intermediate result and the received second encrypted intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of short message information of a user and internet surfing condition;
a transmission unit 404, configured to send the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
an updating unit 405, configured to update the model parameter of the first model according to the first gradient obtained by the decryption by the intermediary, so as to implement training of the first model.
In an embodiment provided by the present invention, based on the foregoing implementation process, optionally, the obtaining unit 401 includes:
an obtaining subunit, configured to obtain original sample data;
and the preprocessing unit is used for preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
In an embodiment provided by the present invention, based on the implementation process, optionally, the model training apparatus further includes:
the encryption unit is used for encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result;
a sending unit, configured to send the first encrypted intermediate result to the second party.
In an embodiment provided by the present invention, based on the implementation process, optionally, the model training apparatus further includes:
the judging unit is used for judging whether the updated first model meets a preset training completion condition or not;
the first execution unit is used for monitoring risks of a target user by using the trained first model under the condition that the updated first model meets a preset training completion condition;
and the second execution unit is used for acquiring new first sample data under the condition that the updated first model does not meet a preset training completion condition, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
The specific principle and the implementation process of each unit and each module in the model training device disclosed in the embodiment of the present invention are the same as those of the model training method disclosed in the embodiment of the present invention, and reference may be made to corresponding parts in the model training method provided in the embodiment of the present invention, which are not described herein again.
The embodiment of the invention also provides a storage medium, which comprises a stored instruction, wherein when the instruction runs, the equipment where the storage medium is located is controlled to execute the model training method.
An electronic device is provided in an embodiment of the present invention, and the structural diagram of the electronic device is shown in fig. 5, which specifically includes a memory 501 and one or more instructions 502, where the one or more instructions 502 are stored in the memory 501, and are configured to be executed by one or more processors 503 to perform the following operations according to the one or more instructions 502:
acquiring a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
generating a first intermediate result from the first model and the first sample data;
generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition;
sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
and updating the model parameters of the first model according to the first gradient obtained by decryption of the middle party so as to realize training of the first model.
As shown in fig. 6, assuming that the LR model is selected in the model learning module, and the encryption transmission algorithm is performed by using homomorphic encryption, the overall process is divided into 3 stages: (1) sharing model distribution; (2) local model training; and (3) collecting, aggregating and updating model information. The specific process is described as follows: (1) The central server (i.e., the intermediary) distributes the public keys to model a and model B for encrypting the data that needs to be exchanged during the training process. (2) The interaction between alignment data a and alignment data B is in encrypted form for computing intermediate results of the gradient. (3) The alignment data A and the alignment data B are calculated respectively based on the learning rate of encryption, and the alignment data B calculates loss according to the label data thereof and summarizes the results to a central server. The central server calculates the total learning rate by summarizing the results and decrypts it. (4) The central server respectively transmits the decrypted learning rates back to the first model A and the second model B; the first model a and the second model B update the parameters of the respective models according to the gradient. And iterating the steps until the loss function converges, so that the whole training process is completed.
In the local model training process, the LR model training process is realized in three stages:
(1) Data preprocessing: and for some data samples, value missing possibly occurs, which affects the training accuracy of the model to a certain extent, and in the stage, data filling is performed by adopting an algorithm based on a gray K neighbor of a characteristic weight.
(2) A model training stage: this phase may give the probability of y <0 or y >0 for an input x, and then infer whether the sample is a positive or negative sample. LR introduces sigmoid function to infer the probability of a sample being a positive sample, and the probability of an input sample x being a positive sample can be expressed as: p (y | x) = g (y), the probability of y =1 for a known model θ and sample x may be expressed as:
when g (y) >0.5, judging the sample as a positive sample, and corresponding to y >0; conversely, when g (y) <0.5, it is judged as a negative sample, corresponding to y <0.LR employs a logarithmic loss function, which can be expressed as:
a and B are data partners, wherein B possesses a tag. Presence data set
Separately initializing model parameters θ A 、θ B The objective function is:
order:
[[·]]if homomorphic encryption is represented, the original objective function can be represented as follows after homomorphic encryption:
in the same way, the order of the method,
the gradient can be expressed as follows:
(3) And a parameter encryption uploading stage.
It should be noted that the model training method and apparatus, the storage medium, and the electronic device provided by the present invention can be used in the fields of artificial intelligence, block chaining, distributed, cloud computing, big data, internet of things, mobile internet, network security, chip, virtual reality, augmented reality, holography, quantum computing, quantum communication, quantum measurement, digital twinning, and finance. The above description is only an example, and does not limit the application fields of the model training method and apparatus, the storage medium, and the electronic device provided by the present invention.
It should be noted that, in this specification, each embodiment is described in a progressive manner, and each embodiment focuses on differences from other embodiments, and portions that are the same as and similar to each other in each embodiment may be referred to. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Finally, it should also be noted that, in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functions of the units may be implemented in the same software and/or hardware or in a plurality of software and/or hardware when implementing the invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
The model training method provided by the invention is described in detail above, and the principle and the implementation mode of the invention are explained in the text by applying specific examples, and the description of the above examples is only used for helping understanding the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A method of model training for application to a first party of a federated learning system, the method comprising:
acquiring a first model and first sample data; the first sample data includes financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
generating a first intermediate result from the first model and the first sample data;
generating a first encryption gradient of the first model according to the first intermediate result and the received second encryption intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition;
sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
and updating the model parameters of the first model according to the first gradient obtained by decryption of the intermediate party so as to realize the training of the first model.
2. The method of claim 1, wherein obtaining the first sample data comprises:
obtaining original sample data;
and preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
3. The method of claim 1, after generating a first intermediate result from the first model and the first sample data, further comprising:
encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result;
sending the first encrypted intermediate result to the second party.
4. The method of claim 1, wherein after updating the model parameters of the first model according to the first gradient decrypted by the intermediary party, the method further comprises:
judging whether the updated first model meets a preset training completion condition or not;
under the condition that the updated first model meets a preset training completion condition, performing risk monitoring on a target user by using the trained first model;
and under the condition that the updated first model does not meet the preset training completion condition, acquiring new first sample data, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
5. A model training apparatus for use with a first party of a federal learning system, the apparatus comprising:
an acquisition unit configured to acquire a first model and first sample data; the first sample data comprises financial behavior information of a user; the financial behavior information comprises at least one of receipt and payment information, credit information and repayment records;
a first generating unit, configured to generate a first intermediate result according to the first model and the first sample data;
a second generating unit, configured to generate a first encryption gradient of the first model according to the first intermediate result and the received second encrypted intermediate result; the second encrypted intermediate result is obtained by encrypting the second intermediate result by a second party of the federal learning system; the second intermediate result is generated by the second participant according to a second model and second sample data; the second sample data comprises communication behavior information of a user; the communication behavior information comprises at least one of user short message information and internet surfing condition;
the transmission unit is used for sending the first encryption gradient to an intermediate party of the federal learning system, so that the intermediate party of the federal learning system decrypts the first encryption gradient to obtain a first gradient;
and the updating unit is used for updating the model parameters of the first model according to the first gradient obtained by the decryption of the middle party so as to realize the training of the first model.
6. The apparatus of claim 5, wherein the obtaining unit comprises:
the acquisition subunit is used for acquiring original sample data;
and the preprocessing unit is used for preprocessing the original sample data by applying a preset proximity algorithm to obtain first sample data.
7. The apparatus of claim 5, further comprising:
the encryption unit is used for encrypting the first intermediate result according to the key issued by the intermediate party to obtain a first encrypted intermediate result;
a sending unit, configured to send the first encrypted intermediate result to the second party.
8. The apparatus of claim 5, further comprising:
the judging unit is used for judging whether the updated first model meets a preset training completion condition or not;
the first execution unit is used for monitoring risks of a target user by using the trained first model under the condition that the updated first model meets a preset training completion condition;
and the second execution unit is used for acquiring new first sample data under the condition that the updated first model does not meet the preset training completion condition, and returning to execute the step of generating a first intermediate result according to the first model and the first sample data.
9. A storage medium comprising stored instructions, wherein the instructions, when executed, control a device on which the storage medium resides to perform the model training method of any one of claims 1 to 4.
10. An electronic device comprising a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by the one or more processors to perform the model training method of any one of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211042961.XA CN115375458A (en) | 2022-08-29 | 2022-08-29 | Model training method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211042961.XA CN115375458A (en) | 2022-08-29 | 2022-08-29 | Model training method and device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115375458A true CN115375458A (en) | 2022-11-22 |
Family
ID=84070406
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211042961.XA Pending CN115375458A (en) | 2022-08-29 | 2022-08-29 | Model training method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115375458A (en) |
-
2022
- 2022-08-29 CN CN202211042961.XA patent/CN115375458A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110399742B (en) | Method and device for training and predicting federated migration learning model | |
| CN110189192B (en) | Information recommendation model generation method and device | |
| CN110245510B (en) | Method and apparatus for predicting information | |
| US20230078061A1 (en) | Model training method and apparatus for federated learning, device, and storage medium | |
| CN111428881B (en) | Recognition model training method, device, equipment and readable storage medium | |
| WO2020220823A1 (en) | Method and device for constructing decision trees | |
| CN111695674B (en) | Federal learning method, federal learning device, federal learning computer device, and federal learning computer readable storage medium | |
| CN112989399B (en) | Data processing system and method | |
| CN113542228B (en) | Data transmission method and device based on federal learning and readable storage medium | |
| CA3095309A1 (en) | Application of trained artificial intelligence processes to encrypted data within a distributed computing environment | |
| CN112039702A (en) | Model parameter training method and device based on federal learning and mutual learning | |
| CN111612167A (en) | Joint training method, device, equipment and storage medium of machine learning model | |
| CN114611008B (en) | User service strategy determination method and device based on federal learning and electronic equipment | |
| CN109767330A (en) | For managing system, the method and apparatus of works | |
| CN114742239A (en) | Financial insurance claim risk model training method and device based on federal learning | |
| CN113722753A (en) | Privacy data processing method, device and system based on block chain | |
| CN117422553A (en) | Transaction processing method, device, equipment, medium and product of blockchain network | |
| CN112598311A (en) | Risk operation identification model construction method and risk operation identification method | |
| CN115375458A (en) | Model training method and device, storage medium and electronic equipment | |
| CN114463063A (en) | Data processing method and related device | |
| CN114493850A (en) | Artificial intelligence-based online notarization method, system and storage medium | |
| CN110879893A (en) | Logistics supply chain big data processing system based on block chain and cloud computing | |
| US12008009B2 (en) | Pre-computation and memoization of simulations | |
| CN115358826A (en) | Product recommendation method and device, storage medium and electronic equipment | |
| US20230137497A1 (en) | Pre-computation and memoization of simulations |
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 |