CN112183584A - Method and device for model training and business execution - Google Patents
Method and device for model training and business execution Download PDFInfo
- Publication number
- CN112183584A CN112183584A CN202010932642.0A CN202010932642A CN112183584A CN 112183584 A CN112183584 A CN 112183584A CN 202010932642 A CN202010932642 A CN 202010932642A CN 112183584 A CN112183584 A CN 112183584A
- Authority
- CN
- China
- Prior art keywords
- user
- loss
- application platform
- model
- trained
- 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
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000012549 training Methods 0.000 title claims abstract description 46
- 230000000875 corresponding effect Effects 0.000 claims description 104
- 238000004590 computer program Methods 0.000 claims description 18
- 230000002596 correlated effect Effects 0.000 claims description 18
- 238000003860 storage Methods 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 16
- 230000008569 process Effects 0.000 description 11
- 230000006870 function Effects 0.000 description 10
- 230000006872 improvement Effects 0.000 description 9
- 238000012545 processing Methods 0.000 description 8
- 230000006399 behavior Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012954 risk control Methods 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Abstract
The specification discloses a method and a device for model training and service execution, which are characterized in that user information of a user on each application platform is acquired, a user comprehensive characteristic and a first loss of the user are determined through a first model to be trained corresponding to each application platform, a target event and a label are determined, the user comprehensive characteristic is input into a second model to be trained corresponding to the target event, the execution probability of the user for executing the target event is obtained, a second loss is determined according to the execution probability and the label, and each first model to be trained and each second model to be trained are trained according to the first loss and the second loss. The first model and the second model are obtained through the training of the method, so that the user comprehensive characteristics are obtained through the first models, the result obtained based on the user comprehensive characteristics and the second model is more reasonable, the problem of user information redundancy is solved, and the problem of user information loss is solved to a certain extent.
Description
Technical Field
The present disclosure relates to the field of internet technologies, and in particular, to a method and an apparatus for model training and service execution.
Background
At present, the influence of internet finance on daily life is gradually increased. The internet financial platform can determine the credit of the user, and can recommend payment modes, internet financial products and the like to the user according to the credit of the user.
Specifically, because the user has user information on all application platforms such as take-out platforms, when the user generates services related to finance on the application platforms, the application platforms jump to the internet finance platform to process the related services, so that the internet finance platform can respectively determine user credit scores of the user on each application platform according to the user information on each application platform, process the user credit scores of each application platform to obtain a comprehensive credit score of the user, and evaluate the user credit based on the comprehensive credit score.
However, in practical situations, the user information of the same user on different application platforms may be the same, and there may be a case where the determined user credit is unreasonable due to the redundancy of the user information. In addition, the application platform may have a problem of partial loss of user information, and the above-mentioned content also has a situation that the determined user credit is unreasonable due to the loss of user information.
Disclosure of Invention
The embodiments of the present specification provide a method and an apparatus for model training and service execution, so as to partially solve the above problems in the prior art.
The embodiment of the specification adopts the following technical scheme:
the present specification provides a method of model training, the method comprising:
acquiring user information of a user on each application platform;
aiming at each application platform, inputting user information on the application platform into a first model to be trained corresponding to the application platform to obtain user characteristics, corresponding to the application platform, of the user output by the first model to be trained;
determining user comprehensive characteristics and first loss of the user according to user characteristics of the user corresponding to each application platform;
determining a target event, judging whether the user executes the target event historically or not, taking a judgment result as a label, and inputting the user comprehensive characteristics into a second model to be trained corresponding to the target event to obtain the execution probability of the user executing the target event, which is output by the second model to be trained;
determining a second loss of the second model to be trained according to the execution probability and the label;
and training each first model to be trained and each second model to be trained according to the first loss and the second loss.
Optionally, the user characteristics include common characteristics and characteristic characteristics;
determining a user comprehensive characteristic and a first loss of the user according to the user characteristics of each application platform, specifically comprising:
determining average common characteristics according to the common characteristics of all the application platforms;
and determining the user comprehensive characteristics and the first loss according to the average common characteristics and the characteristic characteristics of each application platform.
Optionally, determining the first loss according to the average commonality characteristic and the characteristic of each application platform, specifically including:
respectively determining first common loss of the first model to be trained corresponding to each application platform according to the common characteristics of each application platform and the average common characteristics;
respectively determining first characteristic losses of the first model to be trained corresponding to each application platform according to the characteristic features of each application platform and the average commonality feature;
and determining the first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform.
Optionally, respectively determining a first characteristic loss of the first model to be trained corresponding to each application platform according to the characteristic feature of each application platform and the average commonality feature, specifically including:
aiming at each application platform, determining a first orthogonal loss of the application platform according to the characteristic feature of the application platform and the average commonality feature;
determining a second orthogonality loss of the application platform according to the characteristic features of the application platform and the characteristic features of any other application platform;
and determining a first characteristic loss of the first model to be trained corresponding to the application platform according to the first orthogonal loss and each second orthogonal loss.
Optionally, determining the first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform, specifically including:
determining a sum value of first common losses of a first model to be trained corresponding to each application platform as a first common loss sum value and a sum value of first characteristic losses as a first characteristic loss sum value;
and determining the first loss according to a preset common loss weight, a preset characteristic loss weight, the first common loss and value and the first characteristic loss and value.
Optionally, training each first model to be trained and each second model to be trained according to the first loss and the second loss specifically includes:
determining a final loss according to a preset first loss weight and a preset second loss weight, and the first loss and the second loss, wherein the final loss is positively correlated with the first loss, and the final loss is positively correlated with the second loss;
and training each first model to be trained and each second model to be trained by taking the final loss minimization as an optimization target.
A method for service execution provided by this specification, the method including:
acquiring user information of a target user on each application platform;
respectively determining the user characteristics of the target user on each application platform according to the user information of the target user on each application platform and the pre-trained first model corresponding to each application platform;
determining the user comprehensive characteristics of the target user according to the user characteristics on each application platform;
inputting the user comprehensive characteristics into a pre-trained second model to obtain a prediction result output by the second model;
and executing service to the target user according to the prediction result.
The present specification provides an apparatus for model training, the apparatus comprising:
the first acquisition module is used for acquiring user information of a user on each application platform;
the first output module is used for inputting the user information on the application platform into a first model to be trained corresponding to the application platform aiming at each application platform to obtain the user characteristics, corresponding to the application platform, of the user output by the first model to be trained;
a first loss determining module, configured to determine, according to a user characteristic of each application platform corresponding to the user, a user comprehensive characteristic and a first loss of the user;
the event determining module is used for determining a target event, judging whether the target event is executed by the user in history or not, taking a judgment result as a label, and inputting the comprehensive characteristics of the user into a second model to be trained corresponding to the target event to obtain the execution probability of the target event executed by the user and output by the second model to be trained;
a second loss determining module, configured to determine a second loss of the second model to be trained according to the execution probability and the label;
and the training module is used for training each first model to be trained and each second model to be trained according to the first loss and the second loss.
The present specification provides an apparatus for service execution, the apparatus comprising:
the second acquisition module is used for acquiring user information of the target user on each application platform;
the user characteristic determining module is used for respectively determining the user characteristics of the target user on each application platform according to the user information of the target user on each application platform and the pre-trained first model corresponding to each application platform;
a comprehensive characteristic determining module for determining the comprehensive characteristics of the users of the target users according to the characteristics of the users on each application platform;
the second output module is used for inputting the user comprehensive characteristics into a pre-trained second model to obtain a prediction result output by the second model;
and the service execution module is used for executing service to the target user according to the prediction result.
The present specification provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the above-described method of model training and business execution.
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, and when the processor executes the computer program, the method for model training and business execution is implemented.
The embodiment of the specification adopts at least one technical scheme which can achieve the following beneficial effects:
the specification can obtain the user information of the user on each application platform, and for each application platform, the user information on the application platform is input into the first model to be trained corresponding to the application platform, so as to obtain the user characteristics corresponding to the application platform of the user output by the first model to be trained, determining a user comprehensive characteristic and a first loss of the user according to the user characteristic of the user corresponding to each application platform, and determines the target event, judges whether the user has executed the target event historically, takes the judgment result as a label, inputting the comprehensive characteristics of the user into a second model to be trained corresponding to the target event to obtain the execution probability of the user executing the target event output by the second model to be trained, and determining second loss of the second model to be trained according to the execution probability and the label, and training each first model to be trained and each second model to be trained according to the first loss and the second loss. The first models corresponding to the application platforms and the second models corresponding to the target events are obtained through the training of the method, so that the comprehensive characteristics of the user are obtained through the first models, the result is obtained based on the comprehensive characteristics of the user and the second models, the result is more reasonable compared with the result obtained in the prior art, the problem of user information redundancy is solved, and the problem of user information loss is solved to a certain extent.
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 specification and not to limit the specification in a non-limiting sense. In the drawings:
FIG. 1 is a flow chart of a method for model training provided by an embodiment of the present disclosure;
fig. 2 is a flowchart of a method for executing a service according to an embodiment of the present disclosure;
FIG. 3 is a schematic structural diagram of a model training apparatus provided in an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of a service execution apparatus provided in an embodiment of the present specification;
fig. 5 is a schematic diagram of an electronic device corresponding to fig. 1 provided in an embodiment of the present disclosure.
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 internet financial platform may determine a user credit, which may be represented by a user credit score. When each application platform generates certain services, each application platform can jump to the internet financial platform so that the internet financial platform processes the services. For example, when the takeout platform generates services such as payment, the takeout platform can jump to the internet financial platform, and the internet financial platform executes the services such as payment. Therefore, when determining the user credit, the existing internet financial platform may first determine the user credit of each application platform, and then synthesize the user credit of each application platform to obtain the comprehensive credit of the user.
Specifically, for each application platform, data processing may be performed on the user information of the user on the application platform, for example, the user information of the user on the application platform may be input into a user credit model corresponding to the application platform, so as to obtain the user credit of the user on the application platform. And processing the user credit of the user on each application platform, for example, setting credit weight for each application platform, and determining the weighted sum value of the user credit of the user on each application platform to obtain the comprehensive credit of the user.
The user information of the same user on each application platform may be partially the same or may have a deficiency. For example, the same user includes information such as an identification in the information on the takeout platform, but lacks information such as a contact address, and the information such as the identification in the information on the car rental platform is the same as the identification on the takeout platform.
Therefore, for the same user information existing on each application platform, when the user score is determined, the user credit is obtained for multiple times by taking the user information as the basis, so that the comprehensive credit of the user is unreasonable, and for the user information which is missing on some application platforms and exists on other application platforms, when the user score is determined, for the application platforms which lack the user information, the obtained user credit of the user on the application platforms is unreasonable because the missing user information cannot be complemented, so that the comprehensive credit of the user is unreasonable.
Therefore, the present specification provides a method for model training and service execution, in which a first model corresponding to each application platform and a second model corresponding to a target event are obtained through training, and when an internet financial platform executes a service, a user integrated credit can be determined through each first model and each second model, and the service is executed based on the user integrated credit.
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 flowchart of a method for training a model provided in an embodiment of the present disclosure, which may specifically include the following steps:
s100: and acquiring user information of the user on each application platform.
In this specification, the internet financial platform may obtain user information of a user on each application platform, where the application platform is a platform that provides a specific application for the user, for example, a takeout platform, a car rental platform, and the like, and the user may take out a spot on the takeout platform. The internet financial platform is a platform for executing financial-related services of each application platform, and risk control, payment modes or financial products recommendation and the like can be carried out to users through the internet financial platform. In addition, the internet financial platform can be simply regarded as a virtual cash register, for example, a user can take a sale on the takeout platform, and when a payment service is generated, the takeout platform can jump to the internet financial platform, so that the internet financial platform can execute the payment service.
The user information may include user registration information as well as user behavior information.
The user registration information may include information that a user account, a contact address, and the like may be used as a user identity, and may also include information that the user account jumps between different application platforms, and the like. The user registration information of each application platform includes the same kind of information, for example, each application platform includes a contact address. However, the user registration information of the same user on each application platform is not necessarily identical. For example, the same user may have multiple contact addresses, and the contact addresses on the takeaway platform may be different from the contact addresses on the rental platform. Moreover, the user registration information of the same user on each application platform does not necessarily exist, for example, the user stores the contact address and the bank card number on the takeout platform, and only stores the contact address on the car rental platform, but does not store the bank card number.
The user behavior information may include user registration time, the number of times the user logs in the application platform, browsing information of the user on the application platform, and the like, and may further include operations of the user on the application platform, such as operations of generating an order, and the like. The user preference can be obtained through the user behavior information, and the activity degree of the user on each application platform can be determined through the user behavior information of the user on the application platform.
Therefore, for any user, the internet financial platform can acquire the user registration information and the user behavior information of the user on each application platform.
S102: and aiming at each application platform, inputting the user information on the application platform into a first model to be trained corresponding to the application platform to obtain the user characteristics, corresponding to the application platform, of the user output by the first model to be trained.
After obtaining the user information of the user on each application platform, for each application platform, first, the internet financial platform may determine, according to characteristics of the application platform, a first model corresponding to the application platform, where the characteristics of the application platform may include a user registration number, data distribution, and the like, and the first model may be a machine learning model, for example, a Multilayer Perceptron (MLP), a neural network model, and the like.
Secondly, the internet financial platform can train the first model corresponding to the application platform. The user information on the application platform can be input into the first model to be trained corresponding to the application platform, and the user characteristics of the user corresponding to the application platform and output by the first model to be trained are obtained.
The core ideas for constructing the first models corresponding to the application platforms are the same, so that the feature dimensions of the input information and the feature dimensions of the output information of the first models corresponding to the application platforms are the same. The input information of the first model corresponding to each application platform is the user information of the user on each application platform, and as can be seen from the above, the user information on each application platform of the same user is different, so that the user characteristics output by the first model corresponding to each application platform are different, but the feature dimensions of the user characteristics are the same.
Based on the above, if the user has user information missing on a certain application platform, the missing user information may be set to zero and input into the first model corresponding to the application platform.
S104: and determining the user comprehensive characteristics and the first loss of the user according to the user characteristics of the user corresponding to each application platform.
After obtaining the user characteristics of the user corresponding to each application platform, the internet financial platform can determine the user comprehensive characteristics and the first loss of the user. The user comprehensive characteristics can be represented as information obtained by integrating user information of the user on each application platform. The first loss can be represented as a loss obtained by integrating the losses of the first model corresponding to each application platform.
Specifically, the user characteristics may include common characteristics and characteristic characteristics, where the common characteristics may represent the common characteristics of the user information of the user on each application platform, and the characteristic characteristics output by the first model corresponding to each application platform may represent the characteristics of the user information of the user on the application platform. If the common characteristics of the first model outputs corresponding to the application platforms are the same, the consistency of the user information of the user on the application platforms is higher, and if the characteristic characteristics of the first model outputs corresponding to the application platforms are orthogonal, the diversity of the user information of the user on the application platforms is better.
In one embodiment provided in this specification, the user features are represented as 256-dimensional feature vectors, where the first 128 dimensions represent common features and the second 128 dimensions represent characteristic features, and since the dimensions of the common features, the characteristic features, and the user features are the same, the first 128 dimensions of the user features may be used as the first 128 dimensions of data of the common features, the second 128 dimensions of data of the common features may be set to zero, and at the same time, the first 128 dimensions of data of the characteristic features may be set to zero, and the second 128 dimensions of data of the user features may be used as the second 128 dimensions of data of the characteristic features. The user feature output by the first model can be denoted as V, the commonality feature can be denoted as V1, and the feature features can be denoted as V2, V, V1, and V2, if the feature dimensions are the same, then V can be denoted as V1+ V2.
The internet financial platform can determine average common characteristics according to the common characteristics of the application platforms, and determine user comprehensive characteristics and first loss according to the average common characteristics and the characteristic characteristics of the application platforms.
First, average commonality characteristics. And determining the average value of the common characteristics of all the application platforms as the average common characteristics. Therefore, the average commonality characteristic can be shown in equation (1).
Wherein the content of the first and second substances,
for average commonality characteristics, n is the number of application platforms, Vi,1Is a common feature of the ith application platform.
And then the user comprehensive characteristics. And determining the sum of the average common characteristic and the characteristic of each application platform as the user comprehensive characteristic. Therefore, the user comprehensive characteristics can be shown as formula (2).
Wherein the content of the first and second substances,
for user-integrated features, Vi,2Is the characteristic feature of the ith application platform.
The foregoing is a manner for determining a user comprehensive characteristic provided in this specification, and in this specification, the user comprehensive characteristic may also be determined according to the average commonality characteristic and the characteristic of each application platform in other manners. For example, a difference between the average commonality feature and each commonality feature may be determined, and the difference, the average commonality feature, and a sum of each characteristic feature may be used as the user integrated feature. That is, as described above, the user comprehensive characteristic represents information obtained by synthesizing the user information of the user on each application platform, and thus, the user comprehensive characteristic can be obtained by synthesizing the user characteristics corresponding to each application platform. That is, as long as the user characteristics corresponding to each application platform are known, each user characteristic can be integrated to obtain the user integrated characteristic.
Meanwhile, the internet financial platform can determine the first loss according to the average commonality characteristic and the characteristic of each application platform.
Specifically, according to the commonality characteristic and the average commonality characteristic of each application platform, the internet financial platform can respectively determine the first commonality loss of the first model to be trained corresponding to each application platform.
And aiming at the first model to be trained corresponding to each application platform, the internet financial platform can determine the first common loss of the first model to be trained according to the common characteristics of the application platform and the average common characteristics. And determining the difference value of the common characteristic of the application platform and the average common characteristic as the difference value corresponding to the first model to be trained, wherein the difference value corresponding to the first model to be trained is positively correlated with the first common loss of the first model to be trained.
In this specification, the power of the difference corresponding to the first model to be trained may be used as the first common loss of the first model to be trained. Of course, the first common loss of the first model to be trained may also be determined in other manners in this specification, and as long as the common characteristic and the average common characteristic of the application platform are determined, the first common loss of the first model to be trained may be determined according to a positive correlation between the difference corresponding to the first model to be trained and the first common loss of the first model to be trained.
And according to the characteristic features and the average commonality features of the application platforms, the internet financial platform can respectively determine the first characteristic loss of the first model to be trained corresponding to each application platform.
The method comprises the steps of determining a first orthogonal loss of each application platform according to the characteristic feature and the average common feature of the application platform, determining a second orthogonal loss of the application platform according to the characteristic feature of the application platform and the characteristic feature of any other application platform, and determining a first characteristic loss of a first model to be trained corresponding to the application platform according to the first orthogonal loss and each second orthogonal loss, wherein the first characteristic loss is positively correlated with the first orthogonal loss, and the first characteristic loss is positively correlated with the second orthogonal loss.
In this specification, a square of a product of the characteristic feature of the application platform and the average commonality feature is determined as a first square value, a product of a square of a modulus of the characteristic feature of the application platform and a square of a modulus of the average commonality feature is determined as a second square value, and a ratio of the first square value and the second square value is determined as the first orthogonality loss.
For each other application platform, determining the square of the product of the characteristic feature of the application platform and the characteristic feature of the other application platform as a third square value, determining the product of the square of the module of the characteristic feature of the application platform and the square of the module of the characteristic feature of the other application platform as a fourth square value, and determining the ratio of the third square value to the fourth square value as the orthogonal loss corresponding to the application platform and the other application platform; and determining the sum of the orthogonal losses of the application platform and other application platforms as the second orthogonal loss.
Therefore, after the first orthogonal loss and the second orthogonal loss are obtained, the first characteristic loss can be determined from the relationship in which the first characteristic loss is positively correlated with the first orthogonal loss and the first characteristic loss is positively correlated with the second orthogonal loss. In this specification, the sum of the first and second orthogonal losses may be directly used as the first characteristic loss, or the first and second orthogonal loss weights may be set and the weighted sum of the first and second orthogonal losses may be determined as the first characteristic loss.
As can be seen from the above, the first characteristic loss can be expressed as equation (3).
Wherein li,tA first characteristic loss, V, of the first model to be trained corresponding to the ith application platformj,2The characteristic features corresponding to the jth application platform.
And then, determining a first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform.
The internet financial platform can determine a sum value of first common losses of a first model to be trained corresponding to each application platform as a first common loss sum value and a sum value of first characteristic losses as a first characteristic loss sum value, and determine the first losses according to a preset common loss weight, a preset characteristic loss weight, a preset first common loss sum value and a preset first characteristic loss sum value, wherein the first losses are positively correlated with the first common losses sum value, and the first losses are positively correlated with the first characteristic losses sum value.
In this specification, the first loss and the value of the first commonality and the weighted sum of the first characteristic losses and values may be directly regarded as the first loss. Of course, as long as the first loss of commonality and the value of the first loss of commonality and the first loss of characteristic and the value of the first loss are determined, the first loss can be determined based on the relationship that the first loss is positively correlated with the first loss of commonality and the first loss is positively correlated with the first loss of characteristic and the value.
When the commonality loss weight and the characteristic loss weight are set, the greater the commonality loss weight is, the more attention is paid to the consistency of the user information on each application platform when the first model is trained, and the greater the characteristic loss weight is, the more attention is paid to the diversity of the user information on each application platform when the first model is trained.
Therefore, as can be seen from the above, the first loss can be expressed as equation (4).
Wherein L is1Is the first loss, a is the weight of the common loss, b is the weight of the characteristic loss, li,gA first common loss, l, of the first model to be trained corresponding to the ith application platformi,tA first characteristic of a first model to be trained corresponding to the ith application platformAnd (4) sexual loss.
S106: determining a target event, judging whether the user executes the target event historically or not, taking a judgment result as a label, and inputting the user comprehensive characteristics into a second model to be trained corresponding to the target event to obtain the execution probability of the user executing the target event output by the second model to be trained.
S108: and determining a second loss of the second model to be trained according to the execution probability and the label.
After determining the user integrated characteristic of the user and the first loss, the internet financial platform may determine a target event. The target events can include events such as overdue payment and the like, exist on each application platform, and can be executed on each application platform by a user. That is, the above-mentioned contents are to stand in the application platform dimension, determine the first model corresponding to each application platform, and obtain the user comprehensive characteristics and the first loss, and then stand in the event dimension, determine the second model corresponding to the event and the second loss.
Specifically, the internet financial platform may determine whether the user has executed the target event in history, and use the determination result as a label, that is, if it is determined that the user has executed the target event in history, the execution is used as a label result, and if it is determined that the user has not executed the target event in history, the non-execution is used as a label result. Of course, the labeling result can also be represented by using special marks such as 0 and 1, 0 representing no execution, 1 representing execution, etc.
And, the internet financial platform may determine a second model corresponding to the target event, wherein the second model is a machine learning model, such as a logistic regression model.
The internet financial platform can input the comprehensive characteristics of the user into the second model to be trained corresponding to the target event to obtain the execution probability of the user executing the target event output by the second model to be trained. Then, a second loss of the second model to be trained is determined according to the execution probability and the label.
Specifically, the execution probability and the annotated cross-entropy Loss may be determined as a second Loss of the second model to be trained, or the execution probability and the annotated focus Loss (Focal Loss) may also be determined as a second Loss of the second model to be trained.
Of course, in this specification, the determination result that the target event is executed by the user historically may be recorded as 1, the determination result that the target event is not executed by the user historically may be recorded as 0, and a difference between the execution probability and the determination result may be determined as an execution difference, and the execution difference is positively correlated with the second loss.
In addition, in this specification, the internet financial platform may further determine a plurality of target events and a second model corresponding to each target event, so that, according to the above, a second loss of each second model may be obtained. The input information of each second model is the same, namely, the input information is the comprehensive characteristics of the user.
S110: and training each first model to be trained and each second model to be trained according to the first loss and the second loss.
After the first loss and the second loss of the second model corresponding to the target event are obtained, the internet financial platform can determine the final loss according to the preset first loss weight, the preset second loss weight, the preset first loss and the preset second loss, wherein the final loss is positively correlated with the first loss, the final loss is positively correlated with the second loss, the final loss is minimized to be an optimization target, and each first model to be trained and each second model to be trained are trained.
Therefore, the final loss can be shown in equation (5).
L＝αL1+βL2 (5)
Where L is the final loss, α is the first loss weight, and β is the second loss weight.
In this specification, if the number of the target events is at least two, the sum of the second losses of the second model to be trained corresponding to each target event may be determined, or the weight of each target event may be determined, and the weighted sum of the second losses of the second model to be trained corresponding to each target event may be determined. The final loss is based on equation (5) and can be shown as equation (6).
Where m is the number of target events, wmIs the weight of the ith target event, Li,2And the second loss of the second model to be trained corresponding to the ith target event.
In addition, in this specification, n first models to be trained and m second models to be trained are trained simultaneously, when a training condition is satisfied, one second model to be trained may be selected from the m second models to be trained as an appointed second model to be trained, the n first models to be trained and the appointed second model to be trained are trained, and after the training is completed, the second models to be trained and the n trained first models are sequentially selected from the remaining second models to be trained until all the second models to be trained are trained. The training condition may include that the number of times of training is greater than a preset number of times of training.
In this specification, after the training of each of the first model and the second model is completed, each of the first model and the second model may be applied to a scenario where the internet financial platform performs a service.
Fig. 2 is a flowchart of a method for executing a service according to an embodiment of the present disclosure, which may specifically include the following steps:
s200: and acquiring user information of the target user on each application platform.
S202: and respectively determining the user characteristics of the target user on each application platform according to the user information of the target user on each application platform and the pre-trained first model corresponding to each application platform.
S204: and determining the user comprehensive characteristics of the target user according to the user characteristics on each application platform.
S206: and inputting the user comprehensive characteristics into a pre-trained second model to obtain a prediction result output by the second model.
S208: and executing service to the target user according to the prediction result.
Specifically, regarding the pre-training process of each first model and each second model, reference may be made to the above steps S100 to S110, and similarly, specific contents of the user characteristics and the user comprehensive characteristics of the target user on each application platform may be determined, and may be referred to above, which is not described herein again.
And inputting the user comprehensive characteristics of the target user into a second model corresponding to the target event to obtain a prediction result output by the second model, determining a user credit representation value according to the prediction result, and executing a service for the target user according to the user credit representation value.
Specifically, the internet financial platform may preset a user credit weight corresponding to the target event, and determine the user credit representation value according to the prediction result and the user credit weight. For example, the product of the prediction result and the user credit weight may be determined as the user credit characterizing value.
When the number of the target events is multiple, the weight of each target event can be determined, and the sum of the products of the weight of each target event and the prediction result is used as the user credit representation value.
The internet financial platform can set a plurality of user credit characteristic value intervals, and determines the user credit characteristic value interval in which the user characteristic value of the target user is located according to the user credit characteristic value of the target user. And executing the service to the target user according to the corresponding relation between the credit representation value interval of each user and the service.
Based on the method for model training shown in fig. 1, an embodiment of the present specification further provides a schematic structural diagram of an apparatus for model training, as shown in fig. 3.
Fig. 3 is a schematic structural diagram of an apparatus for model training provided in an embodiment of the present disclosure, where the apparatus includes:
a first obtaining module 301, configured to obtain user information of a user on each application platform;
a first output module 302, configured to, for each application platform, input user information on the application platform into a first model to be trained corresponding to the application platform, so as to obtain a user characteristic, output by the first model to be trained, of the user corresponding to the application platform;
a first loss determining module 303, configured to determine a user comprehensive characteristic and a first loss of the user according to a user characteristic of each application platform corresponding to the user;
an event determining module 304, configured to determine a target event, determine whether the target event is executed by the user in history, use a determination result as a label, and input the user comprehensive characteristic into a second model to be trained corresponding to the target event, so as to obtain an execution probability, output by the second model to be trained, of the target event executed by the user;
a second loss determining module 305, configured to determine a second loss of the second model to be trained according to the execution probability and the label;
a training module 306, configured to train each first model to be trained and each second model to be trained according to the first loss and the second loss.
Optionally, the user characteristics include common characteristics and characteristic characteristics;
the first loss determining module 303 is specifically configured to determine an average commonality characteristic according to a commonality characteristic of each application platform; and determining the user comprehensive characteristics and the first loss according to the average common characteristics and the characteristic characteristics of each application platform.
Optionally, the first loss determining module 303 is specifically configured to determine, according to the commonality characteristic and the average commonality characteristic of each application platform, a first commonality loss of the first model to be trained corresponding to each application platform respectively; respectively determining first characteristic losses of the first model to be trained corresponding to each application platform according to the characteristic features of each application platform and the average commonality feature; and determining the first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform.
Optionally, the first loss determining module 303 is specifically configured to, for each application platform, determine a first orthogonal loss of the application platform according to the characteristic feature of the application platform and the average commonality feature; determining a second orthogonality loss of the application platform according to the characteristic features of the application platform and the characteristic features of any other application platform; and determining a first characteristic loss of the first model to be trained corresponding to the application platform according to the first orthogonal loss and each second orthogonal loss.
Optionally, the first loss determining module 303 is specifically configured to determine a sum of first common losses of the first model to be trained corresponding to each application platform as a first common loss sum value, and determine a sum of first characteristic losses as a first characteristic loss sum value; and determining the first loss according to a preset common loss weight, a preset characteristic loss weight, the first common loss and value and the first characteristic loss and value.
Optionally, the training module 306 is specifically configured to determine a final loss according to a preset first loss weight, a preset second loss weight, and the first loss and the second loss, where the final loss is positively correlated with the first loss, and the final loss is positively correlated with the second loss; and training each first model to be trained and each second model to be trained by taking the final loss minimization as an optimization target.
Based on the method for executing the service shown in fig. 2, the embodiment of this specification further provides a schematic structural diagram of a device for executing the service, as shown in fig. 4.
Fig. 4 is a schematic structural diagram of an apparatus for service execution according to an embodiment of the present disclosure, where the apparatus includes:
a second obtaining module 401, configured to obtain user information of a target user on each application platform;
a user feature determining module 402, configured to determine, according to the user information of the target user on each application platform and a pre-trained first model corresponding to each application platform, user features of the target user on each application platform respectively;
a comprehensive characteristic determining module 403, configured to determine a user comprehensive characteristic of the target user according to user characteristics on each application platform;
a second output module 404, configured to input the user comprehensive characteristics into a pre-trained second model, so as to obtain a prediction result output by the second model;
and the service execution module 405 is configured to execute a service for the target user according to the prediction result.
The present specification also provides a computer readable storage medium, which stores a computer program, and the computer program can be used to execute the method for model training and business execution provided by the above content.
Based on the method for model training and service execution shown in the above, the embodiment of this specification further provides a schematic structure diagram of the electronic device shown in fig. 5. As shown in fig. 5, 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 nonvolatile memory into the memory and then runs the computer program to implement the method for model training and service execution 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 modules. 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 making 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, pl (core universal Programming Language), HDCal (jhdware Description Language), lang, Lola, HDL, laspam, hardward Description Language (vhr Description Language), vhal (Hardware Description Language), and vhigh-Language, which are currently used in most common. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.
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 as pure computer readable program code, 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 considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure 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 flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams 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 an … …" does not exclude the presence of other like elements 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 the like) 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, etc. 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 method of model training, the method comprising:
acquiring user information of a user on each application platform;
aiming at each application platform, inputting user information on the application platform into a first model to be trained corresponding to the application platform to obtain user characteristics, corresponding to the application platform, of the user output by the first model to be trained;
determining user comprehensive characteristics and first loss of the user according to user characteristics of the user corresponding to each application platform;
determining a target event, judging whether the user executes the target event historically or not, taking a judgment result as a label, and inputting the user comprehensive characteristics into a second model to be trained corresponding to the target event to obtain the execution probability of the user executing the target event, which is output by the second model to be trained;
determining a second loss of the second model to be trained according to the execution probability and the label;
and training each first model to be trained and each second model to be trained according to the first loss and the second loss.
2. The method of claim 1, wherein user characteristics include commonality characteristics and characteristic characteristics;
determining a user comprehensive characteristic and a first loss of the user according to the user characteristics of each application platform, specifically comprising:
determining average common characteristics according to the common characteristics of all the application platforms;
and determining the user comprehensive characteristics and the first loss according to the average common characteristics and the characteristic characteristics of each application platform.
3. The method according to claim 2, wherein determining the first loss according to the average commonality characteristic and the characteristic of each application platform specifically comprises:
respectively determining first common loss of the first model to be trained corresponding to each application platform according to the common characteristics of each application platform and the average common characteristics;
respectively determining first characteristic losses of the first model to be trained corresponding to each application platform according to the characteristic features of each application platform and the average commonality feature;
and determining the first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform.
4. The method according to claim 3, wherein the determining the first characteristic loss of the first model to be trained corresponding to each application platform according to the characteristic feature of each application platform and the average commonality feature comprises:
aiming at each application platform, determining a first orthogonal loss of the application platform according to the characteristic feature of the application platform and the average commonality feature;
determining a second orthogonality loss of the application platform according to the characteristic features of the application platform and the characteristic features of any other application platform;
and determining a first characteristic loss of the first model to be trained corresponding to the application platform according to the first orthogonal loss and each second orthogonal loss.
5. The method of claim 3, wherein determining the first loss according to the first common loss and the first characteristic loss of the first model to be trained corresponding to each application platform specifically comprises:
determining a sum value of first common losses of a first model to be trained corresponding to each application platform as a first common loss sum value and a sum value of first characteristic losses as a first characteristic loss sum value;
and determining the first loss according to a preset common loss weight, a preset characteristic loss weight, the first common loss and value and the first characteristic loss and value.
6. The method of claim 1, wherein training each of the first model to be trained and the second model to be trained according to the first loss and the second loss comprises:
determining a final loss according to a preset first loss weight and a preset second loss weight, and the first loss and the second loss, wherein the final loss is positively correlated with the first loss, and the final loss is positively correlated with the second loss;
and training each first model to be trained and each second model to be trained by taking the final loss minimization as an optimization target.
7. A method of service execution, the method comprising:
acquiring user information of a target user on each application platform;
respectively determining the user characteristics of the target user on each application platform according to the user information of the target user on each application platform and the pre-trained first model corresponding to each application platform;
determining the user comprehensive characteristics of the target user according to the user characteristics on each application platform;
inputting the user comprehensive characteristics into a pre-trained second model to obtain a prediction result output by the second model;
and executing service to the target user according to the prediction result.
8. An apparatus for model training, the apparatus comprising:
the first acquisition module is used for acquiring user information of a user on each application platform;
the first output module is used for inputting the user information on the application platform into a first model to be trained corresponding to the application platform aiming at each application platform to obtain the user characteristics, corresponding to the application platform, of the user output by the first model to be trained;
a first loss determining module, configured to determine, according to a user characteristic of each application platform corresponding to the user, a user comprehensive characteristic and a first loss of the user;
the event determining module is used for determining a target event, judging whether the target event is executed by the user in history or not, taking a judgment result as a label, and inputting the comprehensive characteristics of the user into a second model to be trained corresponding to the target event to obtain the execution probability of the target event executed by the user and output by the second model to be trained;
a second loss determining module, configured to determine a second loss of the second model to be trained according to the execution probability and the label;
and the training module is used for training each first model to be trained and each second model to be trained according to the first loss and the second loss.
9. An apparatus for service execution, the apparatus comprising:
the second acquisition module is used for acquiring user information of the target user on each application platform;
the user characteristic determining module is used for respectively determining the user characteristics of the target user on each application platform according to the user information of the target user on each application platform and the pre-trained first model corresponding to each application platform;
a comprehensive characteristic determining module for determining the comprehensive characteristics of the users of the target users according to the characteristics of the users on each application platform;
the second output module is used for inputting the user comprehensive characteristics into a pre-trained second model to obtain a prediction result output by the second model;
and the service execution module is used for executing service to the target user according to the prediction result.
10. A computer-readable storage medium, characterized in that the storage medium stores a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-6 or 7.
11. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any of claims 1-6 or 7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010932642.0A CN112183584A (en) | 2020-09-08 | 2020-09-08 | Method and device for model training and business execution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010932642.0A CN112183584A (en) | 2020-09-08 | 2020-09-08 | Method and device for model training and business execution |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112183584A true CN112183584A (en) | 2021-01-05 |
Family
ID=73924931
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010932642.0A Pending CN112183584A (en) | 2020-09-08 | 2020-09-08 | Method and device for model training and business execution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112183584A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113159834A (en) * | 2021-03-31 | 2021-07-23 | 支付宝(杭州)信息技术有限公司 | Commodity information sorting method, device and equipment |
| CN113610175A (en) * | 2021-08-16 | 2021-11-05 | 上海冰鉴信息科技有限公司 | Service strategy generation method and device and computer readable storage medium |
-
2020
- 2020-09-08 CN CN202010932642.0A patent/CN112183584A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113159834A (en) * | 2021-03-31 | 2021-07-23 | 支付宝(杭州)信息技术有限公司 | Commodity information sorting method, device and equipment |
| CN113159834B (en) * | 2021-03-31 | 2022-06-07 | 支付宝(杭州)信息技术有限公司 | Commodity information sorting method, device and equipment |
| CN113610175A (en) * | 2021-08-16 | 2021-11-05 | 上海冰鉴信息科技有限公司 | Service strategy generation method and device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110413877B (en) | Resource recommendation method and device and electronic equipment | |
| CN111563551B (en) | Multi-mode information fusion method and device and electronic equipment | |
| CN108460523B (en) | Wind control rule generation method and device | |
| CN114202370A (en) | Information recommendation method and device | |
| CN113688313A (en) | Training method of prediction model, information pushing method and device | |
| CN110008991B (en) | Risk event identification method, risk identification model generation method, risk event identification device, risk identification equipment and risk identification medium | |
| CN112214652B (en) | Message generation method, device and equipment | |
| CN113643119A (en) | Model training method, business wind control method and business wind control device | |
| CN113641896A (en) | Model training and recommendation probability prediction method and device | |
| CN116306868B (en) | Model processing method, device and equipment | |
| CN112183584A (en) | Method and device for model training and business execution | |
| CN115146731A (en) | Model training method, business wind control method and business wind control device | |
| CN114997472A (en) | Model training method, business wind control method and business wind control device | |
| CN110033092B (en) | Data label generation method, data label training device, event recognition method and event recognition device | |
| CN111582872A (en) | Abnormal account detection model training method, abnormal account detection device and abnormal account detection equipment | |
| CN113222649A (en) | Method and device for recommending service execution mode | |
| CN116308738B (en) | Model training method, business wind control method and device | |
| CN115130621A (en) | Model training method and device, storage medium and electronic equipment | |
| CN116230146A (en) | Data processing method, training method of ICD (ICD coding) model and related equipment | |
| CN111401641B (en) | Service data processing method and device and electronic equipment | |
| CN111461892B (en) | Method and device for selecting derived variables of risk identification model | |
| CN110458393B (en) | Method and device for determining risk identification scheme and electronic equipment | |
| CN109903165B (en) | Model merging method and device | |
| CN114119139A (en) | Information recommendation method and device, storage medium and electronic equipment | |
| CN112184074A (en) | Prompt message output method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210105 |