CN115563584B - 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
- CN115563584B CN115563584B CN202211509842.0A CN202211509842A CN115563584B CN 115563584 B CN115563584 B CN 115563584B CN 202211509842 A CN202211509842 A CN 202211509842A CN 115563584 B CN115563584 B CN 115563584B
- Authority
- CN
- China
- Prior art keywords
- user
- service
- layer
- service request
- feature extraction
- 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
Images
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The specification discloses a model training method, a device, a storage medium and an electronic device, which can enable a detection model to distinguish the service types to which input user data belongs in the process of training the detection model, and can determine whether the service request of a user belongs to the detection result of an abnormal service request according to the user data belonging to different service types, thereby reducing the cost of performing abnormal detection on the service request of the user.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a model training method and apparatus, a storage medium, and an electronic device.
Background
With the development of internet technology, more and more internet service providers are beginning to provide various business services in Application programs (APPs) for users to use.
In order to improve the security of the user in the process of using the business service and avoid the leakage of private data of the user, the internet service provider often performs anomaly detection on the business request of the user by adopting different neural network models aiming at different business services, so that the response of the business request with risks can be refused. However, the training, deployment and other work of the multiple neural network models are performed at the same time, which increases the cost of risk control of the internet service provider on the service request of the user.
Therefore, how to reduce the cost of performing risk control on the service request sent by the user is an urgent problem to be solved.
Disclosure of Invention
The specification provides a model training method, a model training device, a storage medium and electronic equipment, which are used for solving the problem of risk in a loan request of a user in the prior art.
The technical scheme adopted by the specification is as follows:
the present specification provides a model training method for training a first detection model, the first detection model comprising: a feature extraction layer, a prediction layer and a decision layer; the method comprises the following steps:
acquiring user data corresponding to a historical service request;
inputting the user data into the feature extraction layer to extract user features from the user data through the feature extraction layer;
inputting the user characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics;
inputting the user characteristics to a decision layer corresponding to the service type to obtain a detection result of whether the historical service request is an abnormal service request;
and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimum deviation between the service class and the service class actually corresponding to the historical service request and the deviation between the detection result and the actual result of whether the historical service request is an abnormal service request as optimization targets.
Optionally, the inputting the user characteristic into the prediction layer to obtain a service category corresponding to the historical service request predicted by the prediction layer according to the user characteristic includes:
for each user characteristic, determining a correlation coefficient between the user characteristic and each other user characteristic;
determining the cross feature corresponding to the user feature according to the correlation coefficient;
and inputting the cross features corresponding to the user features into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the cross features.
Optionally, the feature extraction layer includes each group of parameter matrices;
determining a correlation coefficient between the user characteristic and each of the other user characteristics, specifically including:
for each group of parameter matrixes, determining a correlation coefficient between the user characteristic and each other user characteristic through the group of parameter matrixes to serve as a correlation coefficient corresponding to the group of parameter matrixes;
determining the cross feature corresponding to the user feature according to the correlation coefficient, specifically comprising:
determining each sub-cross feature corresponding to the user feature according to the correlation coefficient corresponding to each group of parameter matrixes;
and fusing the sub-cross features to obtain the cross features corresponding to the user features.
Optionally, the inputting the user characteristic into the prediction layer to obtain a service category corresponding to the historical service request predicted by the prediction layer according to the user characteristic includes:
for each user characteristic, fusing the user characteristic with the cross characteristic corresponding to the user characteristic to obtain a fused characteristic corresponding to the user characteristic;
and inputting the fusion characteristics to the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the fusion characteristics.
Optionally, with a goal of minimizing a deviation between the service class and a service class actually corresponding to the historical service request and a deviation between the detection result and an actual result of whether the historical service request is an abnormal service request as optimization goals, training at least a feature extraction layer and a decision layer in the first detection model specifically includes:
determining a first loss of the detection model according to the service category and a deviation between the service categories actually corresponding to the historical service requests;
determining a second loss of the detection model according to a deviation between the detection result and an actual result of whether the historical service request is an abnormal service request;
and training at least a feature extraction layer and a decision layer in the detection model by taking the minimization of the first loss and the second loss as optimization targets.
Optionally, training at least a feature extraction layer and a decision layer in the first detection model with the minimization of the first loss and the second loss as an optimization objective specifically includes:
weighting and fusing the first loss and the second loss to obtain comprehensive loss;
and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimized comprehensive loss as an optimization target.
The present specification provides a business wind control method for risk monitoring by a second detection model, the detection model comprising: a feature extraction layer and a decision layer; the method comprises the following steps:
acquiring user data of a user;
inputting the user data into the feature extraction layer, so as to extract user features from the user data through the feature extraction layer, and inputting the user features into a decision layer corresponding to each service category to obtain a detection result of each service category;
according to the received service type corresponding to the service request sent by the user, determining a detection result of the service type corresponding to the service request from detection results of all the service types, wherein the detection result is used as a target detection result, and the feature extraction layer in the second detection model and the decision layer corresponding to all the service types are obtained by training through the model training method;
and carrying out service wind control on the user according to the target detection result.
The present specification provides a model training apparatus for training a first detection model, the first detection model comprising: the characteristic extraction layer, the prediction layer and the decision layer comprise:
the acquisition module is used for acquiring user data corresponding to the historical service request;
the feature extraction module is used for inputting the user data into the feature extraction layer so as to extract user features from the user data through the feature extraction layer;
the prediction module is used for inputting the user characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics;
the detection module is used for inputting the user characteristics to a decision layer corresponding to the service type to obtain a detection result of whether the historical service request is an abnormal service request;
and the training module is used for training at least a feature extraction layer and a decision layer in the first detection model by taking the minimized deviation between the service class and the service class actually corresponding to the historical service request and the deviation between the detection result and the actual result of whether the historical service request is an abnormal service request as optimization targets.
This specification provides a business wind control device, the device is used for carrying out risk monitoring through the second detection model, the second detection model includes: the characteristic extraction layer and the decision layer comprise:
the data acquisition module is used for acquiring the user data of the user;
the first detection module is used for inputting the user data into the feature extraction layer, so that the user features are extracted from the user data through the feature extraction layer, and the user features are input into the decision layers corresponding to the service categories to obtain detection results of the service categories;
the second detection module is used for determining a detection result of the service class corresponding to the service request from the detection results of all the service classes according to the received service class corresponding to the service request sent by the user, and the detection result is used as a target detection result, and the feature extraction layer in the second detection model and the decision layer corresponding to all the service classes are obtained by training through the model training method;
and the execution module is used for carrying out service wind control on the user according to the target detection result.
The present specification provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the above-described model training, business wind control method.
The present specification provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above model training and business wind control method when executing the program.
The technical scheme adopted by the specification can achieve the following beneficial effects:
the model training method provided by the specification comprises the steps of firstly obtaining user data corresponding to historical service requests, inputting the obtained user data into a feature extraction layer of a detection model, extracting user features from the user data through the feature extraction layer, inputting the user features into a prediction layer, obtaining service classes corresponding to the historical service requests predicted by the prediction layer according to the user features, inputting the user features into a decision layer corresponding to the service classes, obtaining a detection result of whether the historical service requests are abnormal service requests, minimizing the deviation between the service classes and the service classes actually corresponding to the historical service requests, and taking the deviation between the detection result and the actual result of whether the historical service requests are abnormal service requests as an optimization target, and training at least the feature extraction layer and the decision layer in the detection model.
The method can be seen in that in the process of training the detection model, the detection model can distinguish the service types to which the input user data belong, and can determine whether the service request of the user belongs to the detection result of the abnormal service request according to the user data belonging to different service types, so that the problem that the service requests of different service types need to be detected through a plurality of neural network models can be avoided, and the cost for performing abnormal detection on the service request of the user is reduced.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification and are incorporated in and constitute a part of this specification, illustrate embodiments of the specification and together with the description serve to explain the principles of the specification and not to limit the specification in a limiting sense. In the drawings:
FIG. 1 is a schematic flow chart of a model training method provided herein;
FIG. 2 is a schematic diagram of the internal structure of a first detection model provided in the present specification;
fig. 3 is a schematic flowchart of a service wind control method provided in this specification;
FIG. 4 is a schematic diagram of a model training apparatus provided herein;
fig. 5 is a schematic diagram of a service wind control device provided in the present specification;
fig. 6 is a schematic diagram of an electronic device corresponding to fig. 1 provided in the present specification.
Detailed Description
In order to make the objects, technical solutions and advantages of the present disclosure more clear, the technical solutions of the present disclosure will be clearly and completely described below with reference to the specific embodiments of the present disclosure and the accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present disclosure, and not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without any creative effort belong to the protection scope of the present specification.
The technical solutions provided by the embodiments of the present description are described in detail below with reference to the accompanying drawings.
Fig. 1 is a schematic flowchart of a model training method provided in this specification, including the following steps:
s100: and acquiring user data corresponding to the historical service request.
In this specification, a service platform may obtain user data corresponding to each historical service request of a user, and perform dimensionless processing on the obtained user data (i.e., removing influences caused by non-uniform data units, such as normalization, standardization, and the like), so that a preset first detection model may be trained according to the user data corresponding to each historical service request after processing, so that the first detection model obtained by training may be used to perform wind control on service requests of different service categories of the user, where the first detection model includes: a feature extraction layer, a prediction layer and a decision layer.
Wherein, the different service categories may refer to credit loan services with different loan amounts and modes.
In the above, the user data includes: account registration information of the user, account historical transfer information of the user, account historical debit and credit information of the user, historical operation information of the user, historical credit information of the user, account information of the user, service request information of the user and the like.
The account registration information of the user refers to information filled by the user when registering an account, for example: information such as user name and mobile phone number;
the account historical transfer information of the user refers to transfer information recorded when the user transfers money to other users during account use, such as: transfer time, transfer amount, charge-receiving user, etc.;
the historical account borrowing information of the user refers to information that the user borrows credit in each credit loan service provided by the service platform, for example: borrowing time, borrowing amount, repayment time and the like;
the historical operation information of the user refers to operation information executed by the user in an Application (APP) provided by the service platform, for example: click operation information (i.e., which buttons the user has clicked, etc.), browse operation information (i.e., which pages the user has browsed, browse duration, etc.);
the historical credit information of the user refers to the credit information of the user recorded by the service platform, such as: credit scoring, whether the user belongs to a blacklist, whether violation records exist and the like;
the account information of the user refers to account asset information corresponding to an account number of the user, for example: account balance, etc.;
the service request information of the user refers to service information included in the service request sent by the user, for example: loan amount, request time, etc.
In the present specification, the execution subject for implementing the model training method may refer to a designated device such as a server installed on the service platform, or may refer to a designated device such as a desktop computer or a notebook computer.
S102: inputting the user data into the feature extraction layer to extract user features from the user data through the feature extraction layer.
S104: and inputting the user characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics.
The server may input the user data into the feature extraction layer of the first detection model, so as to extract the user features from the user data through the feature extraction layer, and input the user features into the prediction layer of the first detection model, so as to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user features.
Specifically, the server may determine, for each user characteristic, a correlation coefficient between the user characteristic and each of the other user characteristics, determine a cross characteristic corresponding to the user characteristic according to the correlation coefficient, input the cross characteristic corresponding to the determined user characteristic to the prediction layer, and obtain a service category corresponding to a historical service request predicted by the prediction layer according to the cross characteristic corresponding to the determined user characteristic.
The method for determining the cross feature corresponding to the user feature by the server according to the correlation coefficient may be that the user feature is fused with other user features according to the correlation coefficient between the user feature and each of the other user features to obtain the cross feature corresponding to the user feature.
Further, in order to improve the accuracy of the cross features corresponding to the determined user features, the server may further determine, through preset sets of parameter matrices, a correlation coefficient between the user feature and each of the other user features as a correlation coefficient corresponding to the set of parameter matrices, and then determine, according to the correlation coefficient corresponding to each of the parameter matrices, each sub-cross feature corresponding to the user feature, where the correlation coefficient corresponding to one of the sets of parameter matrices may determine one sub-cross feature corresponding to the user feature, and further may fuse the sub-cross features, so as to obtain the cross feature corresponding to the user feature.
Further, overfitting during training of the detection model is prevented, the server can also fuse the user features and the cross features corresponding to the user features (for example, weighting fusion, splicing fusion and the like) aiming at each user feature to obtain fusion features corresponding to the user features, further, the service categories corresponding to the historical service requests can be determined according to the fusion features, and the obtained fusion features are input into a prediction layer of the first detection model to obtain the service categories corresponding to the historical service requests through the prediction layer.
S106: and inputting the user characteristics to a decision layer corresponding to the service type to obtain a detection result of whether the historical service request is an abnormal service request.
Further, after determining the service category corresponding to the historical service request, the server may input at least one of the user characteristics, the cross characteristics corresponding to the user characteristics, and the fusion characteristics corresponding to the user characteristics into the decision layer corresponding to the determined service category, so as to obtain a detection result of whether the historical service request is an abnormal service request through the decision layer corresponding to the determined service category.
S108: and training at least a feature extraction layer and a decision layer in the detection model by taking the minimum deviation between the service class and the service class actually corresponding to the historical service request and the deviation between the detection result and the actual result of whether the historical service request is an abnormal service request as optimization targets.
Further, the server may determine a first loss of the detection model according to a deviation between a service class determined by the first detection model based on the user characteristic and a service class actually corresponding to the historical service request, and determine a second loss of the detection model according to a deviation between a detection result output by the first detection model and an actual result of whether the historical service request is an abnormal service request, so as to train the first detection model with the minimized first loss and the minimized second loss as optimization targets.
Specifically, the server may perform weighted fusion on the first loss and the second loss to obtain a comprehensive loss, and may further perform training on the first detection model with minimized comprehensive loss as an optimization target.
To elaborate on the above, the present specification also provides a schematic diagram of the first detection model, as shown in fig. 2.
Fig. 2 is a schematic diagram of an internal structure of the first detection model provided in this specification.
As can be seen from fig. 2, the first detection model includes an input layer, a feature extraction layer, a feature intersection layer, a prediction layer, a decision layer, and an output layer. The detection model can receive user data input by the server through the input layer and input the user data into the feature extraction layer, so that the feature extraction layer extracts user features from the user data and inputs the user features into the feature interaction layer, the feature interaction layer can determine cross features corresponding to each user feature according to the input user features, further can fuse each user feature and the cross features corresponding to each user feature to obtain fusion features corresponding to each user feature, further can input the fusion features into the prediction layer, and determines the service category corresponding to the historical service request through the prediction layer.
Further, after determining the service type corresponding to the historical service request, the server may input the user characteristics into a decision layer corresponding to the determined service type, determine, by the decision layer, whether the service request of the user corresponding to the input user data is a detection result of an abnormal service request, and output the detection result through an output layer.
As can be seen from the above, the services of different service categories refer to credit loan services with different loan amounts and manners, and thus, although there are some differences between different credit loan services, for example: the method for carrying out wind control on the service requests of different service classes has a great deal of common points, and when the service requests of different service classes are subjected to wind control, the types of the used user data are the same, and only the numerical value distribution of the user data is different, so that the service requests of different service classes can be subjected to wind control through the trained detection model, and in the process of training the detection model by using the user data corresponding to the service services of different service classes, the common part involved in the process that the detection model judges whether the service requests of different service classes are abnormal service requests can be better trained, so that the accuracy of the detection result output by the detection model can be improved.
In addition, the server can fuse the neural network models for carrying out wind control on the service requests of different service classes into one detection model, so that the cost of wind control can be reduced.
For further explanation of the present specification, a method for performing business wind control after the detection model trained by the above method is deployed in a practical application is described in detail below, as shown in fig. 3.
Fig. 3 is a schematic flow chart of a service wind control method provided in this specification, including the following steps:
s300: user data of a user is obtained.
S302: and inputting the user data into the feature extraction layer, so as to extract user features from the user data through the feature extraction layer, and inputting the user features into the decision layer corresponding to each service category to obtain the detection result of each service category.
S304: and determining a detection result of the service class corresponding to the service request from the detection result of each service class according to the received service class corresponding to the service request sent by the user, wherein the detection result is used as a target detection result, and the feature extraction layer in the second detection model and the decision layer corresponding to each service class are obtained by training through the model training method.
S306: and carrying out service wind control on the user according to the target detection result.
At present, in the credit loan service provided by an internet service provider, a user makes a loan under the guidance of others, and transfers a loan to an account of others, which sometimes causes great loss to the user and the internet service provider.
In this specification, the service platform may obtain user data of a user, and perform a wind control on a service request of the user according to the obtained user data.
In the above description, there may be two occasions when the server acquires the user data, the first is that the server acquires the user data of the user when monitoring a click operation instruction of the user for any one credit loan service, and the second is that the server acquires the user data of the user in response to a service request sent by the user, which will be described in detail below with respect to the two occasions when the server acquires the user data.
Firstly, when monitoring a click operation instruction of a user for any one credit loan service, a server can acquire user data of the user in advance, and input the acquired user data into a feature extraction layer of a second detection model trained in advance, so as to extract user features from the user data through the feature extraction layer, and input the user features into each decision layer, thereby obtaining a detection result of a service request in each service category.
Furthermore, after receiving the service request sent by the user, according to the service class corresponding to the received service request sent by the user, the detection result of the service class corresponding to the received service request sent by the user is determined from the detection results of the service request in each service class, and is used as a target detection result, so that the service request of the user can be subjected to wind control according to the target detection result.
Secondly, after receiving a service request sent by a user, the server may obtain user data of the user, and input the obtained user data into a feature extraction layer of a second detection model trained in advance, so as to extract user features from the user data through the feature extraction layer, and input the user features into a decision layer corresponding to a service category of the service request of the user, so as to obtain a target detection result of whether the service request belongs to an abnormal service request, and further may perform a wind control on the service request of the user according to the target detection result.
It should be noted that the prediction layer in the first detection model is used to train the feature extraction layer, and is not used when the detection model is actually used for wind control, so the second detection model is a detection model that retains the feature extraction layer and the decision layer in the first detection model.
It should be noted that all actions of acquiring signals, information or data in the present application are performed under the premise of complying with the data protection regulation policy responded by the country of the location and obtaining the authorization given by the owner of the corresponding device.
It can be seen from the above that, the server can determine whether the service request of the user belongs to the detection result of the abnormal service request according to the user data belonging to different service classes through the detection model, so that the problem that the service requests of different service classes need to be detected through a plurality of neural network models can be avoided, and the cost for performing abnormal detection on the service request of the user is reduced.
Based on the same idea, the model training method provided for one or more embodiments of the present specification further provides a corresponding model training device, and a business wind control device as shown in fig. 4 and 5.
Fig. 4 is a schematic diagram of a model training apparatus provided in the present specification, the apparatus including:
an obtaining module 401, configured to obtain user data corresponding to a historical service request;
a feature extraction module 402, configured to input the user data into the feature extraction layer, so as to extract user features from the user data through the feature extraction layer;
the prediction module 403 is configured to input the user characteristics into the prediction layer, so as to obtain a service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics;
a detection module 404, configured to input the user characteristic to a decision layer corresponding to the service category to obtain a detection result of whether the historical service request is an abnormal service request;
a training module 405, configured to train at least a feature extraction layer and a decision layer in the first detection model with an optimization objective of minimizing a deviation between the service category and a service category actually corresponding to the historical service request, and a deviation between the detection result and an actual result of whether the historical service request is an abnormal service request.
Optionally, the prediction module 403 is specifically configured to, for each user feature, determine a correlation coefficient between the user feature and each other user feature; determining the cross feature corresponding to the user feature according to the correlation coefficient; and inputting the cross features corresponding to the user features into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the cross features.
Optionally, the predicting module 403 is specifically configured to, for each group of parameter matrices, determine, through the group of parameter matrices, a correlation coefficient between the user characteristic and each other user characteristic as a correlation coefficient corresponding to the group of parameter matrices; determining each sub-cross feature corresponding to the user feature according to the correlation coefficient corresponding to each group of parameter matrixes; and fusing the sub-cross features to obtain the cross features corresponding to the user features.
Optionally, the prediction module 403 is specifically configured to, for each user feature, fuse the user feature with a cross feature corresponding to the user feature to obtain a fused feature corresponding to the user feature; and inputting the fusion characteristics to the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the fusion characteristics.
Optionally, the training module 405 is specifically configured to determine a first loss of the detection model according to a deviation between the service category and a service category actually corresponding to the historical service request; determining a second loss of the detection model according to a deviation between the detection result and an actual result of whether the historical service request is an abnormal service request; and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimization of the first loss and the second loss as optimization targets.
Optionally, the training module 405 is specifically configured to perform weighted fusion on the first loss and the second loss to obtain a comprehensive loss; and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimization of the comprehensive loss as an optimization target.
Fig. 5 is a schematic diagram of a model training apparatus provided in the present specification, the apparatus including:
a data obtaining module 501, configured to obtain user data of the user;
a first detection module 502, configured to input the user data into the feature extraction layer, so as to extract a user feature from the user data through the feature extraction layer, and input the user feature into a decision layer corresponding to each service category, so as to obtain a detection result of each service category;
a second detection module 503, configured to determine, according to the received service category corresponding to the service request sent by the user, a detection result of the service category corresponding to the service request from detection results of the service categories, where the detection result is used as a target detection result, and a feature extraction layer in the second detection model and a decision layer corresponding to each service category are obtained by training through the model training method;
and the execution module 504 is configured to perform service wind control on the user according to the target detection result.
The present specification also provides a computer-readable storage medium having stored thereon a computer program, the computer program being operable to execute a model training method as provided in fig. 1 above.
This specification also provides a schematic block diagram of an electronic device corresponding to that of figure 1, shown in figure 6. As shown in fig. 6, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, but may also include hardware required for other services. The processor reads a corresponding computer program from the non-volatile memory into the memory and then runs the computer program to implement the model training method of fig. 1 described above. Of course, besides the software implementation, the present specification does not exclude other implementations, such as logic devices or a combination of software and hardware, and the like, that is, the execution subject of the following processing flow is not limited to each logic unit, and may be hardware or logic devices.
In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain the corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical blocks. For example, a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), is an integrated circuit whose Logic functions are determined by programming the Device by a user. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing an Integrated Circuit chip, such Programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development and writing, but the original code before compiling is also written by a specific Programming Language, which is called Hardware Description Language (HDL), and HDL is not only one but many, such as ABEL (Advanced Boolean Expression Language), AHDL (alternate Hardware Description Language), traffic, CUPL (core universal Programming Language), HDCal, jhddl (Java Hardware Description Language), lava, lola, HDL, PALASM, rhyd (Hardware Description Language), and vhigh-Language (Hardware Description Language), which is currently used in most popular applications. It will also be apparent to those skilled in the art that hardware circuitry for implementing the logical method flows can be readily obtained by a mere need to program the method flows with some of the hardware description languages described above and into an integrated circuit.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in purely computer readable program code means, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be regarded as a hardware component and the means for performing the various functions included therein may also be regarded as structures within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
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 various elements may be implemented in the same one or more software and/or hardware implementations of the present description.
As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The description has been presented with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the description. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.
As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
This description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially 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.
The above description is only an example of the present specification, and is not intended to limit the present specification. Various modifications and alterations to this description will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of the present specification should be included in the scope of the claims of the present specification.
Claims (11)
1. A model training method for training a first detection model, the first detection model comprising: a feature extraction layer, a prediction layer and a decision layer; the method comprises the following steps:
acquiring user data corresponding to historical service requests;
inputting the user data into the feature extraction layer to extract user features from the user data through the feature extraction layer;
inputting the user characteristics into the prediction layer to obtain service categories corresponding to the historical service requests predicted by the prediction layer according to the user characteristics, wherein different service types correspond to service services of different service modes;
inputting the user characteristics to a decision layer corresponding to the service type to obtain a detection result of whether the historical service request is an abnormal service request;
and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimized deviation between the service category and the service category actually corresponding to the historical service request and the deviation between the detection result and the actual result of whether the historical service request is an abnormal service request as optimization targets.
2. The method according to claim 1, wherein the step of inputting the user characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics specifically comprises:
for each user characteristic, determining a correlation coefficient between the user characteristic and each other user characteristic;
determining the cross feature corresponding to the user feature according to the correlation coefficient;
and inputting the cross features corresponding to the user features into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the cross features.
3. The method of claim 2, wherein the feature extraction layer comprises sets of parameter matrices;
determining a correlation coefficient between the user characteristic and each of the other user characteristics, specifically including:
for each group of parameter matrixes, determining a correlation coefficient between the user characteristic and each other user characteristic through the group of parameter matrixes as a correlation coefficient corresponding to the group of parameter matrixes;
determining the cross feature corresponding to the user feature according to the correlation coefficient, specifically comprising:
determining each sub-cross feature corresponding to the user feature according to the correlation coefficient corresponding to each group of parameter matrixes;
and fusing the sub-cross features to obtain the cross features corresponding to the user features.
4. The method according to claim 2, wherein the step of inputting the user characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the user characteristics specifically comprises:
for each user characteristic, fusing the user characteristic with the cross characteristic corresponding to the user characteristic to obtain a fused characteristic corresponding to the user characteristic;
and inputting the fusion characteristics into the prediction layer to obtain the service category corresponding to the historical service request predicted by the prediction layer according to the fusion characteristics.
5. The method according to claim 1, taking minimizing a deviation between the traffic class actually corresponding to the historical traffic request and a deviation between the detection result and an actual result of whether the historical traffic request is an abnormal traffic request as an optimization goal, training at least a feature extraction layer and a decision layer in the first detection model specifically includes:
determining a first loss of the detection model according to the service category and a deviation between the service categories actually corresponding to the historical service requests;
determining a second loss of the detection model according to a deviation between the detection result and an actual result of whether the historical service request is an abnormal service request;
and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimization of the first loss and the second loss as optimization targets.
6. The method of claim 5, wherein training at least a feature extraction layer and a decision layer in the first detection model with the minimization of the first loss and the second loss as an optimization objective comprises:
weighting and fusing the first loss and the second loss to obtain comprehensive loss;
and training at least a feature extraction layer and a decision layer in the first detection model by taking the minimization of the comprehensive loss as an optimization target.
7. A business wind control method for risk monitoring by a second detection model, the detection model comprising: a feature extraction layer and a decision layer; the method comprises the following steps:
acquiring user data of a user;
inputting the user data into the feature extraction layer, so as to extract user features from the user data through the feature extraction layer, and inputting the user features into a decision layer corresponding to each service category to obtain a detection result of each service category;
according to the received service class corresponding to the service request sent by the user, determining a detection result of the service class corresponding to the service request from detection results of all the service classes, wherein the detection result is used as a target detection result, and the feature extraction layer in the second detection model and the decision layer corresponding to all the service classes are obtained by training through the method of any one of claims 1~6;
and carrying out service wind control on the user according to the target detection result.
8. A model training apparatus, the apparatus for training a first detection model, the first detection model comprising: the characteristic extraction layer, the prediction layer and the decision layer comprise:
the acquisition module is used for acquiring user data corresponding to the historical service request;
the feature extraction module is used for inputting the user data into the feature extraction layer so as to extract user features from the user data through the feature extraction layer;
the prediction module is used for inputting the user characteristics into the prediction layer to obtain the service types corresponding to the historical service requests predicted by the prediction layer according to the user characteristics, wherein different service types correspond to different service modes;
the detection module is used for inputting the user characteristics to a decision layer corresponding to the service type to obtain a detection result of whether the historical service request is an abnormal service request;
and the training module is used for training at least a feature extraction layer and a decision layer in the first detection model by taking the minimum deviation between the service class and the service class actually corresponding to the historical service request and the deviation between the detection result and the actual result of whether the historical service request is an abnormal service request as optimization targets.
9. A business wind control apparatus for risk monitoring by a second detection model, the second detection model comprising: the characteristic extraction layer and the decision layer comprise:
the data acquisition module is used for acquiring the user data of the user;
the first detection module is used for inputting the user data into the feature extraction layer, so that the user features are extracted from the user data through the feature extraction layer, and the user features are input into the decision layers corresponding to the service categories to obtain detection results of the service categories;
a second detection module, configured to determine, according to a received service category corresponding to a service request sent by the user, a detection result of the service category corresponding to the service request from detection results of the service categories, where the detection result is used as a target detection result, and a feature extraction layer in the second detection model and a decision layer corresponding to each service category are obtained by training according to the method claimed in any one of claims 1~6;
and the execution module is used for carrying out service wind control on the user according to the target detection result.
10. A computer readable storage medium storing a computer program which when executed by a processor implements the method of any of claims 1~7 above.
11. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor when executing the program implementing the method of any of claims 1~7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211509842.0A CN115563584B (en) | 2022-11-29 | 2022-11-29 | Model training method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211509842.0A CN115563584B (en) | 2022-11-29 | 2022-11-29 | Model training method and device, storage medium and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115563584A CN115563584A (en) | 2023-01-03 |
| CN115563584B true CN115563584B (en) | 2023-04-07 |
Family
ID=84770018
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211509842.0A Active CN115563584B (en) | 2022-11-29 | 2022-11-29 | Model training method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115563584B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116502679B (en) * | 2023-05-15 | 2023-09-05 | 之江实验室 | Model construction method and device, storage medium and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021169115A1 (en) * | 2020-02-29 | 2021-09-02 | 平安科技（深圳）有限公司 | Risk control method, apparatus, electronic device, and computer-readable storage medium |
| CN114943307A (en) * | 2022-06-28 | 2022-08-26 | 支付宝(杭州)信息技术有限公司 | Model training method and device, storage medium and electronic equipment |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002054303A1 (en) * | 2000-12-29 | 2002-07-11 | Navamani Dhanasingh | Intelleigent interactive e-commerce technology |
| CN108053314A (en) * | 2018-01-31 | 2018-05-18 | 国信优易数据有限公司 | A kind of Loan Demand Forecasting Methodology |
| CN114997472A (en) * | 2022-05-13 | 2022-09-02 | 北京三快在线科技有限公司 | Model training method, business wind control method and business wind control device |
-
2022
- 2022-11-29 CN CN202211509842.0A patent/CN115563584B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021169115A1 (en) * | 2020-02-29 | 2021-09-02 | 平安科技（深圳）有限公司 | Risk control method, apparatus, electronic device, and computer-readable storage medium |
| CN114943307A (en) * | 2022-06-28 | 2022-08-26 | 支付宝(杭州)信息技术有限公司 | Model training method and device, storage medium and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115563584A (en) | 2023-01-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110503435B (en) | Transaction early warning method, device and equipment based on blockchain | |
| CN110674188A (en) | Feature extraction method, device and equipment | |
| CN115238826B (en) | Model training method and device, storage medium and electronic equipment | |
| CN115563584B (en) | Model training method and device, storage medium and electronic equipment | |
| CN113079224A (en) | Account binding method and device, storage medium and electronic equipment | |
| CN115759122A (en) | Intention identification method, device, equipment and readable storage medium | |
| CN111311267B (en) | Multi-account risk prevention and control method, system and equipment | |
| CN113408254A (en) | Page form information filling method, device, equipment and readable medium | |
| CN115545353B (en) | Business wind control method, device, storage medium and electronic equipment | |
| CN115545720B (en) | Model training method, business wind control method and business wind control device | |
| CN116822606A (en) | Training method, device, equipment and storage medium of anomaly detection model | |
| CN112286968A (en) | Service identification method, equipment, medium and electronic equipment | |
| CN117093862A (en) | Model training method and device, electronic equipment and storage medium | |
| CN116188023A (en) | Method, device, storage medium and equipment for training risk identification model | |
| CN115660105A (en) | Model training method, business wind control method and business wind control device | |
| CN109598511B (en) | Account risk identification method, device and equipment | |
| CN115470501A (en) | Data protection method and device, storage medium and electronic equipment | |
| CN110728516A (en) | Method, device and equipment for updating wind control model | |
| CN115730233B (en) | Data processing method and device, readable storage medium and electronic equipment | |
| CN116340852B (en) | Model training and business wind control method and device | |
| CN114006737B (en) | Account safety detection method and detection device | |
| CN117592102A (en) | Service execution method, device, equipment and storage medium | |
| CN116188021A (en) | Data processing method, device and equipment | |
| CN117455493A (en) | Model training method and device, storage medium and electronic equipment | |
| CN115048578A (en) | Recommendation method and device |
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 |