CN115618962B - Model training method, business wind control method and device - Google Patents
Model training method, business wind control method and device Download PDFInfo
- Publication number
- CN115618962B CN115618962B CN202211276984.7A CN202211276984A CN115618962B CN 115618962 B CN115618962 B CN 115618962B CN 202211276984 A CN202211276984 A CN 202211276984A CN 115618962 B CN115618962 B CN 115618962B
- Authority
- CN
- China
- Prior art keywords
- data
- service
- prediction
- mixed
- sample data
- 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
- G06N20/00—Machine learning
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Medical Informatics (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The specification discloses a model training method, a business wind control method and a business wind control device. First, service data of a service performed by each user in history is acquired to construct each sample data. And then, mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data. And then, inputting at least two sample data into a prediction model to be trained so as to respectively extract service features from the at least two sample data through a feature extraction layer, inputting the service features into a feature mixing layer, mixing the respectively extracted service features in a mixing mode to obtain mixed features, and then inputting the mixed features into the prediction layer so as to obtain a first prediction result according to the mixed features. And finally, training the feature extraction layer and the prediction layer by taking the deviation between the minimized first prediction result and the mixed label data as an optimization target. The method can improve the accuracy of the prediction result determined by the prediction model.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method for model training, a method for business wind control, and an apparatus thereof.
Background
With the rapid development of internet technology, more and more services can be realized through the internet, and accordingly, a plurality of service risks are also accompanied. For example, some users may download or purchase cheating software or scripts on the internet, running in the background of the mobile phone continuously and regularly to simulate the behavior of the users to participate in marketing activities, thereby obtaining illegal benefits. Thus, risk control and privacy data protection for a business are often an integral part of the business process.
At present, in the process of business wind control, a common means is to collect a large amount of user data to train a wind control model. However, when the acquired user data is small, the accuracy of the predicted result predicted by the trained prediction model is low.
Therefore, how to improve the accuracy of the predicted result predicted by the prediction model under the condition that the acquired user data is less is a problem to be solved.
Disclosure of Invention
The specification provides a model training method, device, storage medium and electronic equipment, so as to improve the accuracy of a predicted result predicted by a prediction model under the condition that acquired user data is less.
The technical scheme adopted in the specification is as follows:
the present specification provides a method of model training, comprising:
acquiring service data of each user executing service in history, wherein the service data comprises operation data of the user, basic data of the user, application data of the user and equipment data of terminal equipment of the user;
constructing each sample data according to each acquired service data;
mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data;
inputting the at least two sample data into a prediction model to be trained, so as to respectively extract service features from the at least two sample data through a feature extraction layer in the prediction model, and inputting the respectively extracted service features into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the respectively extracted service features according to the mixing mode to obtain mixed features;
inputting the mixed characteristics to a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed characteristics;
and training at least the feature extraction layer and the prediction layer in the prediction model by taking the deviation between the minimized first risk prediction result and the mixed label data as an optimization target, wherein the prediction model comprises a wind control model for carrying out business wind control.
Optionally, before inputting the at least two sample data into the predictive model to be trained, the method further comprises:
mixing at least two sample data according to the mixing mode to obtain mixed sample data;
with the objective of optimizing minimizing the deviation between the first risk prediction result and the mixed label data, at least before training the feature extraction layer and the prediction layer in the prediction model, the method includes:
inputting the mixed sample data into a prediction model to be trained, extracting service features from the mixed sample data through a feature extraction layer in the prediction model, and inputting the service features extracted from the mixed sample data into the prediction layer to obtain a second risk prediction result;
training at least the feature extraction layer and the prediction layer in the prediction model with a minimum deviation between the first risk prediction result and the mixed label data as an optimization target, specifically including:
and training at least the feature extraction layer and the prediction layer in the prediction model with the aim of minimizing the deviation between the first risk prediction result and the mixed label data and the aim of minimizing the deviation between the second risk prediction result and the mixed label data.
Optionally, mixing at least two sample data according to a preset mixing mode to obtain mixed sample data, which specifically includes:
for the service data corresponding to each service dimension contained in the at least two sample data, mixing the service data corresponding to the service dimension contained in each sample data according to the weight corresponding to the service dimension in the mixing mode to obtain mixed service data corresponding to the service dimension;
and determining mixed sample data according to the mixed service data corresponding to each service dimension.
Optionally, the service features extracted respectively are input to a feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively according to the mixing mode to obtain mixed features, and specifically includes:
inputting the respectively extracted service features to a feature mixing layer in the prediction model, and mixing the feature data corresponding to the feature dimension in each service feature according to the weight corresponding to the feature dimension in the mixing mode aiming at the feature data corresponding to each feature dimension contained in the respectively extracted service features to obtain mixed feature data corresponding to the feature dimension;
And determining the mixed characteristic according to the mixed characteristic data corresponding to each characteristic dimension.
Optionally, before inputting the at least two sample data into the predictive model to be trained, the method further comprises:
determining the number of the obtained mixed sample data as a first number;
and determining the number of the sample data input into the prediction model to be trained as a second number according to the first number and the preset sample proportion.
Inputting the at least two sample data into a prediction model to be trained so as to extract service features from the at least two sample data through a feature extraction layer in the prediction model, wherein the method specifically comprises the following steps of:
the first quantity of mixed sample data and the second quantity of sample data are input into a predictive model to be trained, so that the first quantity of mixed service features are extracted from the first quantity of mixed sample data through a feature extraction layer in the predictive model, and the second quantity of service features are respectively extracted from the second quantity of sample data.
The specification provides a method for business wind control, which comprises the following steps:
acquiring service data when a target user executes a service;
Inputting the business data into a pre-trained prediction model to extract business features from the business data through a feature extraction layer in the prediction model, and inputting the business features into a prediction layer in the prediction model to enable the prediction layer to obtain a prediction result when a target user executes business according to the business features, wherein the prediction model is trained through the model training method;
and carrying out service wind control 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 method of model training and the method of business wind control described above.
The present specification provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of model training and the method of business wind control described above when executing the program.
The above-mentioned at least one technical scheme that this specification adopted can reach following beneficial effect:
in the model training method provided in the present specification, first, service data of a service executed by each user historically is acquired. And secondly, constructing each sample data according to each acquired service data. And then, mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data. And then, inputting at least two sample data into a prediction model to be trained, respectively extracting service features from the at least two sample data through a feature extraction layer in the prediction model, inputting the extracted service features into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the respectively extracted service features according to a mixing mode to obtain mixed features, and then inputting the mixed features into a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed features. And finally, taking the minimized deviation between the first risk prediction result and the mixed label data as an optimization target, and training at least a feature extraction layer and a prediction layer in the prediction model.
According to the method, the service features extracted from at least two sample data can be input into the feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively in a mixing mode to obtain mixed features. And then, inputting the mixed features into a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed features. And finally, taking the minimized deviation between the first risk prediction result and the mixed label data as an optimization target, and training at least a feature extraction layer and a prediction layer in the prediction model. According to the method, the mixed service characteristics are obtained through the characteristic mixing layer, and the mixed service characteristics are combined with the mixed label data so as to increase training samples for training the prediction model. Therefore, the accuracy of the risk prediction result determined by the prediction model is improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification, illustrate and explain the exemplary embodiments of the present specification and their description, are not intended to limit the specification unduly. Attached at
In the figure:
FIG. 1 is a flow chart of a method for model training according to an embodiment of the present disclosure;
fig. 2 is a flow chart of a method for business wind control according to an embodiment of the present disclosure;
FIG. 3 is a schematic structural diagram of a prediction model according to an embodiment of the present disclosure;
FIG. 4 is a schematic structural diagram of a model training device according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of a device for service wind control according to an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the present specification more apparent, the technical solutions of the present specification will be clearly and completely described below with reference to specific embodiments of the present specification and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.
The following describes in detail the technical solutions provided by the embodiments of the present specification with reference to the accompanying drawings.
For the purposes of making the objects, technical solutions and advantages of the present specification more apparent, the technical solutions of the present specification will be clearly and completely described below with reference to specific embodiments of the present specification and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.
The following describes in detail the technical solutions provided by the embodiments of the present specification with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a model training method in the present specification, which specifically includes the following steps:
s100: service data of a service executed by each user in history is obtained.
In the embodiment of the present specification, the execution subject of the model training method may refer to an electronic device such as a server or a desktop computer. For convenience of description, the method of model training provided in the present specification will be described below with only the server as the execution subject.
In the embodiment of the present specification, the server may acquire service data of a service that each user has historically performed.
The service data may include operation data of the user, basic data of the user, application data of the user, and device data of the terminal device of the user. The operation data referred to herein may refer to an operation of a client by a user when executing a service. Such as the user's click frequency, the user's sliding speed, the user's click range, etc. The basic data of the user mentioned herein may refer to basic identity information of the user, for example, a city in which the user is located, an IP address of the user, etc. The application data of the user mentioned herein may refer to application data of a service performed by the user, for example, resource information owned by the user, an activity level of the user, and the like. The device data of the terminal device of the user mentioned here may refer to the state of the terminal device of the user at the time of executing the service, for example, the screen brightness of the terminal device, the program name of the terminal device running, etc.
S102: and constructing each sample data according to each acquired service data.
In the embodiment of the present specification, the server may construct each sample data according to each acquired service data.
The tag data corresponding to the sample data may refer to single-hot tag data, which is used to indicate whether the user applies the cheating software or script. For example, the user applies the cheating software or script, the tag data is (1, 0), otherwise the tag data is (0, 1). Of course, specific tag data can be set according to actual service requirements.
In practical application, part of users can download or purchase cheating software or scripts on the internet and continuously and regularly run in the background of the mobile phone so as to simulate the behavior of the users to participate in games or marketing activities, thereby obtaining illegal benefits. To predict which users apply the cheating software or scripts, the server may construct sample data from the acquired user data for subsequent model training.
While the cheating software or script simulates the behavior of the user to participate in the game or marketing campaign, the cheating software or script cannot completely simulate the behavior of the user, e.g., the frequency of clicking of the cheating software or script is fixed at regular intervals, the sliding speed is uniform, and the range of clicking per click is constant. Based on the above, the server can determine whether the user applies the cheating software or script through the operation data of the user.
Of course, the server may also determine, through the user's basic data, whether the user has applied the cheating software or script. For example, the user's IP address logs in at multiple places simultaneously, possibly with the user applying a cheating software or script to operate multiple accounts simultaneously.
The server may also determine, via the application data of the user, whether the user has applied the cheating software or script. For example, the user's liveness is for a seven day duration login, a significantly non-human-compliant time of interest, and possibly a user's application of cheating software or scripts.
The server can also judge whether the user applies the cheating software or script through the equipment data of the terminal equipment of the user. For example, while the user is participating in the game, the screen brightness of the user's terminal device is zero, which may be software or script that the user runs cheating in the background. For another example, the program name of the illegal software running on the terminal device of the user may be the software or script of the user running in the background to cheat.
The server may determine the risk prediction result according to at least one of the operation data of the user, the base data of the user, the application data of the user, and the device data of the terminal device of the user through the prediction model. The risk prediction results referred to herein may refer to probabilities of software or scripts being used for application cheating.
S104: and mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data.
In practice, training a predictive model typically requires a large amount of user data to be obtained. When the acquired user data is less, the accuracy of risk prediction results predicted by the trained prediction model is often lower. Based on this, the server may mix the two sample data to obtain mixed sample data for training of the predictive model.
In this embodiment of the present disclosure, the server may mix tag data corresponding to at least two sample data according to a preset mixing manner, to obtain mixed tag data. And, the server may mix at least two sample data according to a mixing manner to obtain mixed sample data. The specific formula is as follows:
X ′ ＝γX i +(1―γ)X j
Y ′ ＝γY i +(1―γ)Y j
in the above formula, γ may be used to represent a preset weight. X is X i May be used to represent the traffic data in the ith sample data. X is X j May be used to represent the traffic data in the jth sample data. Y is Y i Tag data that can be used to represent the ith sample data. Y is Y j Tag data that can be used to represent the jth sample data. X is X ′ May be used to represent the mixed traffic data in the mixed sample data. Y is Y ′ May be used to represent the blended tag data in the blended sample data.
Specifically, the server may mix, according to the weight corresponding to the service dimension in the mixing manner, the service data corresponding to the service dimension included in each sample data, so as to obtain mixed service data corresponding to the service dimension. The service dimension referred to herein may refer to different categories of data in the service data, such as click frequency, sliding speed, click range, etc.
The server may then determine the mixed sample data from the mixed service data corresponding to each service dimension.
For example, the traffic data in the two sample data is (3 (click frequency), 4 (slide speed), 5 (click range)), (5 (click frequency), 6 (slide speed), 7 (click range)), and the preset weight is 0.5, and the mixed sample data is (4 (click frequency), 5 (slide speed), 6 (click range)). For another example, if the label data corresponding to the two sample data is (0, 1), (1, 0) and the preset weight is 0.2, the mixed label data is (0.2, 0.8).
It should be noted that, if the tag data in the two sample data are different, the preset weights corresponding to the two sample data are also different. For example, if the tag data in the two sample data are the same, the preset weight corresponding to the two sample data is 0.5. If the label data in the two sample data are different, the preset weight corresponding to the two sample data is 0.2. Of course, the specific numerical values of the preset weights are not limited in this specification.
In the embodiment of the present specification, a plurality of mixed sample data can be obtained by the above-described mixing manner. For example, if the number of sample data is three, the sample data a, the sample data B, and the sample data C are mixed to obtain mixed sample data AB, mixed sample data AC, and mixed sample data BC. Of course, as the number of sample data increases, the number of mixed sample data also increases.
Further, the mixed sample data may be obtained by mixing two sample data, or may be obtained by mixing any number of sample data. Similarly, the mixed tag data may be obtained by mixing two tag data, or may be obtained by mixing any number of tag data.
S106: and inputting the at least two sample data into a prediction model to be trained so as to extract service features from the at least two sample data through a feature extraction layer in the prediction model, and inputting the service features extracted respectively into a feature mixing layer in the prediction model so that the feature mixing layer mixes the service features extracted respectively according to the mixing mode to obtain mixed features.
In practical application, because the service data in the sample data are structured data, usually sparse data, the generated mixed sample data are also sparse data after the service data in the two sample data are mixed, and the accuracy of the risk prediction result predicted by the prediction model can only be partially improved. Based on the above, the server may determine the hybrid tag data in advance, and perform feature extraction on the service data in the sample data first, so that the service features corresponding to the service data are not sparse any more. And mixing the service features of the two sample data after feature extraction to obtain mixed features. The blended features are combined with the blended label data for subsequent training of the predictive model.
In this embodiment of the present disclosure, the server may input at least two sample data into a prediction model to be trained, so as to extract service features from at least two sample data through a feature extraction layer in the prediction model, and input the service features extracted respectively into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively according to a mixing manner, to obtain a mixed feature.
Specifically, the server may input the service features extracted respectively into a feature mixing layer in the prediction model, and mix, according to weights corresponding to feature dimensions in a mixing manner, feature data corresponding to the feature dimensions in each service feature, to obtain mixed feature data corresponding to the feature dimensions. Wherein the weights corresponding to the feature dimensions mentioned herein are the same as the weights corresponding to the hybrid tag data. The server may generate a look-up table based on the blended label data for the feature blending layer in the subsequent predictive model, in the specific representation (i-th sample data, j-th sample data, weight γ, blended label data).
The server may then determine the hybrid feature from the hybrid feature data corresponding to each feature dimension.
It should be noted that, in the feature mixing layer in the prediction model, a data matrix is constructed according to each service feature, one service feature is a row of data in the data matrix, and one column in the data matrix is a feature dimension. Therefore, the obtained mixed feature data corresponding to each feature dimension can construct a new row of business features in the data matrix for training of the prediction model. In the prediction layer of the prediction model, each line of data corresponds to one prediction result and one label data.
S108: and inputting the mixed characteristics to a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed characteristics.
S110: and training at least the feature extraction layer and the prediction layer in the prediction model by taking the deviation between the first risk prediction result and the mixed label data as an optimization target.
In the embodiment of the present disclosure, the server may input the mixed feature to a prediction layer in the prediction model, so that the prediction layer obtains the first risk prediction result according to the mixed feature.
The server may then train at least the feature extraction layer and the prediction layer in the prediction model with the objective of minimizing the deviation between the first risk prediction result and the hybrid tag data.
That is, the server mixes tag data of at least two sample data before inputting the two sample data to the prediction model, resulting in mixed tag data. And inputting at least two sample data into the prediction model to obtain service characteristics corresponding to the at least two sample data. And mixing service features corresponding to at least two sample data through a feature mixing layer to obtain mixed features, and training a prediction model by taking mixed tag data as tag data corresponding to the mixed features. It will be appreciated that the server builds training samples for training the predictive model from the blended features and blended label data obtained at the feature blending layer.
Further, the server may input the mixed sample data into a prediction model to be trained, so as to extract service features from the mixed sample data through a feature extraction layer in the prediction model, and input the service features extracted from the mixed sample data into the prediction layer, so as to obtain a second risk prediction result.
The server may then train at least the feature extraction layer and the prediction layer in the prediction model with the objective of minimizing the deviation between the first risk prediction result and the hybrid tag data and the objective of minimizing the deviation between the second risk prediction result and the hybrid tag data.
As can be seen from the above description, there are two ways of increasing sample data in the present method. First kind: the server mixes at least two sample data before inputting the two sample data into the predictive model to obtain mixed sample data for training of the predictive model. Second kind: the server mixes tag data of the two sample data before inputting the at least two sample data into the predictive model to obtain mixed tag data. And inputting at least two sample data into the prediction model to obtain service characteristics corresponding to the at least two sample data. And mixing service features corresponding to at least two sample data through a feature mixing layer to obtain mixed features, and training a prediction model by taking mixed tag data as tag data corresponding to the mixed features. It will be appreciated that the server constructs sample data for training the predictive model from the blended features and blended label data obtained at the feature blending layer.
Of course, the server may also obtain the third risk prediction result according to the service characteristics of the sample data by using the prediction layer of the prediction model, and train at least the feature extraction layer and the prediction layer in the prediction model with the deviation between the minimum third risk prediction result and the tag data as an optimization target.
In practical applications, since the mixed sample data is generated from each sample data, the mixed sample data may have a partial deviation from the actual situation, so that the server needs to train the prediction model through the sample data in addition to the mixed sample data, so as to improve the performance of the trained prediction model.
In the embodiment of the present specification, the server may determine the number of the obtained mixed sample data as the first number.
Secondly, the server may determine the amount of sample data input into the predictive model to be trained based on the first amount and a preset sample ratio, which may be determined empirically by an expert as the second amount.
The server may then input the first and second amounts of mixed sample data into a predictive model to be trained to extract a first amount of mixed traffic features from the first amount of mixed sample data and a second amount of traffic features from the second amount of sample data, respectively, through a feature extraction layer in the predictive model.
Further, the second number of sample data may also be used to obtain the mixed service feature through the feature mixing layer during a subsequent training process of the prediction model. Of course, the number of hybrid traffic features may be determined by traffic demand.
It should be noted that, the feature mixing layer determines the mixed service features only by the weights. That is, in the feature mixture layer, the network parameters of the model are not predicted, and thus, the weights in the feature mixture layer do not change with the model training process.
In the embodiment of the present specification, the model structure of the prediction model is as shown in fig. 2.
Fig. 2 is a schematic structural diagram of a prediction model according to an embodiment of the present disclosure.
In fig. 2, the server may input at least two sample data, mixed sample data, into the predictive model to be trained to extract service features from the at least two sample data, respectively, and from the mixed sample data, through a feature extraction layer in the predictive model.
Secondly, the server can input the service features respectively extracted into a feature mixing layer in the prediction model so that the feature mixing layer mixes the service features respectively extracted in a mixing mode to obtain mixed features.
And then, the server can input the service characteristics of at least two sample data, the service characteristics of the mixed sample data and the mixed characteristics into a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed characteristics, obtains a second risk prediction result according to the service characteristics of the mixed sample data and obtains a third risk prediction result according to the service characteristics of the sample data.
Finally, the server may train at least the feature extraction layer and the prediction layer in the prediction model with the objective of minimizing the deviation between the first risk prediction result and the hybrid tag data, minimizing the deviation between the second risk prediction result and the hybrid tag data, and minimizing the deviation between the third risk prediction result and the tag data.
According to the method, the service features extracted from at least two sample data can be input into the feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively in a mixing mode to obtain mixed features. And then, inputting the mixed features into a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed features. And finally, taking the minimized deviation between the first risk prediction result and the mixed label data as an optimization target, and training at least a feature extraction layer and a prediction layer in the prediction model. According to the method, the mixed service characteristics are obtained through the characteristic mixing layer, and the mixed service characteristics are combined with the mixed label data so as to increase training samples for training the prediction model. Therefore, the accuracy of the risk prediction result determined by the prediction model is improved.
Fig. 3 is a flow chart of a method for business wind control in the present specification, specifically including the following steps:
s300: and acquiring service data when the target user executes the service.
In the embodiment of the present disclosure, the execution subject of the method for business wind control may refer to an electronic device such as a server or a desktop computer. For convenience of description, the method of service wind control provided in the present specification will be described below with only the server as an execution body.
In the embodiment of the present specification, the server may acquire service data when the target user performs the service.
A business may refer to various applications for marketing campaigns, such as mini-games, trading cards, etc.
Because of different services, the service data of the target user when executing the service is also different, and the service data of the target user when executing the service can be changed according to actual service requirements. That is, different services correspond to different service data.
S302: and inputting the business data into a pre-trained prediction model to extract business features from the business data through a feature extraction layer in the prediction model, and inputting the business features into a prediction layer in the prediction model to enable the prediction layer to obtain a prediction result when a target user executes the business according to the business features.
S304: and carrying out service wind control according to the prediction result.
In this embodiment of the present disclosure, the server may input the service data into a pre-trained prediction model, so as to extract the service features from the service data through a feature extraction layer in the prediction model, and input the service features into a prediction layer in the prediction model, so that the prediction layer obtains a prediction result when the target user executes the service according to the service features.
And secondly, the server can perform service wind control according to the prediction result.
After the prediction model is trained, the prediction model does not use the mixed feature layer when predicting the prediction result of the target user when executing the service, so that the prediction model directly inputs the service features output by the feature extraction layer into the prediction layer by skipping the feature mixed layer, and determines the prediction result of the target user when executing the service.
The above method for training the model provided for the embodiment of the present specification further provides a corresponding device, a storage medium and an electronic apparatus based on the same idea.
Fig. 4 is a schematic structural diagram of a device for model training according to an embodiment of the present disclosure, where the device includes:
An obtaining module 400, configured to obtain service data of each user for executing a service in history, where the service data includes operation data of the user, basic data of the user, application data of the user, and device data of a terminal device of the user;
a construction module 402, configured to construct each sample data according to each acquired service data;
the mixing module 404 is configured to mix tag data corresponding to at least two sample data according to a preset mixing manner, so as to obtain mixed tag data;
the extracting module 406 is configured to input the at least two sample data into a prediction model to be trained, so as to extract service features from the at least two sample data through a feature extracting layer in the prediction model, and input the service features extracted respectively into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively according to the mixing mode to obtain mixed features;
the input module 408 is configured to input the mixed feature to a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed feature;
A training module 410, configured to train at least the feature extraction layer and the prediction layer in the prediction model with a deviation between the first risk prediction result and the mixed label data being minimized as an optimization target, where the prediction model includes a wind control model for performing service wind control.
Optionally, the mixing module 404 is specifically further configured to mix at least two sample data according to the mixing manner to obtain mixed sample data, input the mixed sample data into a prediction model to be trained, extract service features from the mixed sample data through a feature extraction layer in the prediction model, and input the service features extracted from the mixed sample data into the prediction layer to obtain a second risk prediction result, so as to minimize a deviation between the first risk prediction result and the mixed label data, and minimize a deviation between the second risk prediction result and the mixed label data as an optimization objective, and train at least the feature extraction layer and the prediction layer in the prediction model.
Optionally, the mixing module 404 is specifically configured to mix, for the service data corresponding to each service dimension included in the at least two sample data, the service data corresponding to the service dimension included in each sample data according to the weight corresponding to the service dimension in the mixing manner, to obtain mixed service data corresponding to the service dimension, and determine mixed sample data according to the mixed service data corresponding to each service dimension.
Optionally, the extracting module 406 is specifically configured to input the service features extracted respectively to a feature mixing layer in the prediction model, mix, according to the weight corresponding to the feature dimension in the mixing manner, the feature data corresponding to the feature dimension in each service feature, to obtain mixed feature data corresponding to the feature dimension, and determine a mixed feature according to the mixed feature data corresponding to each feature dimension.
Optionally, the mixing module 404 is specifically further configured to determine, as a first amount, an amount of the obtained mixed sample data, determine, according to the first amount and a preset sample ratio, an amount of the sample data input into a prediction model to be trained, input, as a second amount, the first amount of the mixed sample data and the second amount of the sample data into the prediction model to be trained, so as to extract, through a feature extraction layer in the prediction model, the first amount of mixed service features from the first amount of mixed sample data, and extract, from the second amount of sample data, the second amount of service features, respectively.
Fig. 5 is a schematic structural diagram of a device for service wind control according to an embodiment of the present disclosure, where the device includes:
the acquiring module 500 is configured to acquire service data when the target user executes a service;
the input module 502 is configured to input the service data into a pre-trained prediction model, so as to extract service features from the service data through a feature extraction layer in the prediction model, and input the service features into a prediction layer in the prediction model, so that the prediction layer obtains a prediction result when a target user executes a service according to the service features, where the prediction model is obtained by training through the model training method;
and the wind control module 504 is configured to perform service wind control according to the prediction result.
The present specification also provides a computer readable storage medium storing a computer program which when executed by a processor is operable to perform the method of model training provided in fig. 1 and the method of business wind control provided in fig. 3.
The embodiment of the specification also provides a schematic structural diagram of the electronic device shown in fig. 6. At the hardware level, as in fig. 6, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile storage, although it may include hardware required for other services. The processor reads the corresponding computer program from the nonvolatile memory into the memory and then runs the computer program to realize the method of model training provided in fig. 1 and the method of business wind control provided in fig. 3.
Of course, other implementations, such as logic devices or combinations of hardware and software, are not excluded from the present description, that is, the execution subject of the following processing flows is not limited to each logic unit, but may be hardware or logic devices.
It should be noted that, all actions for acquiring signals, information or data in the present application are performed under the condition of conforming to the corresponding data protection rule policy of the country of the location and obtaining the authorization given by the owner of the corresponding device.
In the 90 s of the 20 th century, improvements to one technology could clearly be distinguished as improvements in hardware (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software (improvements to the process flow). However, with the development of technology, many improvements of the current method flows can be regarded as direct improvements of hardware circuit structures. Designers almost always obtain corresponding hardware circuit structures by programming improved method flows into hardware circuits. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device, PLD) (e.g., field programmable gate array (Field Programmable Gate Array, FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A designer programs to "integrate" a digital system onto a PLD without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, such programming is mostly implemented by using "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before the compiling is also written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), but not just one of the hdds, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of 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, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable logic controllers, and embedded microcontrollers, 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 of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in a pure computer readable program code, it is well possible to implement the same functionality by logically programming the method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Such a controller may thus be regarded as a kind of hardware component, and means for performing various functions included therein may also be regarded as structures within the hardware component. Or even means for achieving the various functions may be regarded as either software modules implementing the methods or structures within hardware components.
The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, 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 functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present specification.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present description is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the specification. 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 one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.
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 storage media for a computer 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, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
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 one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises the element.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The 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.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.
The foregoing is merely exemplary of the present disclosure and is not intended to limit the disclosure. Various modifications and alterations to this specification will become apparent to those skilled in the art. Any modifications, equivalent substitutions, improvements, or the like, which are within the spirit and principles of the present description, are intended to be included within the scope of the claims of the present description.
Claims (10)
1. A method of model training, comprising:
acquiring service data of each user executing service in history, wherein the service data comprises operation data of the user, basic data of the user, application data of the user and equipment data of terminal equipment of the user;
constructing each sample data according to each acquired service data;
mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data;
Inputting the at least two sample data into a prediction model to be trained, so as to respectively extract service features from the at least two sample data through a feature extraction layer in the prediction model, and inputting the respectively extracted service features into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the respectively extracted service features according to the mixing mode to obtain mixed features;
inputting the mixed characteristics to a prediction layer in the prediction model, so that the prediction layer obtains a first risk prediction result according to the mixed characteristics;
and training at least the feature extraction layer and the prediction layer in the prediction model by taking the deviation between the minimized first risk prediction result and the mixed label data as an optimization target, wherein the prediction model comprises a wind control model for carrying out business wind control.
2. The method of claim 1, prior to inputting the at least two sample data into the predictive model to be trained, the method further comprising:
mixing at least two sample data according to the mixing mode to obtain mixed sample data;
With the objective of optimizing minimizing the deviation between the first risk prediction result and the mixed label data, at least before training the feature extraction layer and the prediction layer in the prediction model, the method includes:
inputting the mixed sample data into a prediction model to be trained, extracting service features from the mixed sample data through a feature extraction layer in the prediction model, and inputting the service features extracted from the mixed sample data into the prediction layer to obtain a second risk prediction result;
training at least the feature extraction layer and the prediction layer in the prediction model with a minimum deviation between the first risk prediction result and the mixed label data as an optimization target, specifically including:
and training at least the feature extraction layer and the prediction layer in the prediction model with the aim of minimizing the deviation between the first risk prediction result and the mixed label data and the aim of minimizing the deviation between the second risk prediction result and the mixed label data.
3. The method of claim 2, wherein at least two sample data are mixed according to a preset mixing manner to obtain mixed sample data, and specifically comprises:
For the service data corresponding to each service dimension contained in the at least two sample data, mixing the service data corresponding to the service dimension contained in each sample data according to the weight corresponding to the service dimension in the mixing mode to obtain mixed service data corresponding to the service dimension;
and determining mixed sample data according to the mixed service data corresponding to each service dimension.
4. The method of claim 1, inputting the extracted service features into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the extracted service features according to the mixing mode to obtain mixed features, specifically including:
inputting the respectively extracted service features to a feature mixing layer in the prediction model, and mixing the feature data corresponding to the feature dimension in each service feature according to the weight corresponding to the feature dimension in the mixing mode aiming at the feature data corresponding to each feature dimension contained in the respectively extracted service features to obtain mixed feature data corresponding to the feature dimension;
and determining the mixed characteristic according to the mixed characteristic data corresponding to each characteristic dimension.
5. The method of claim 2, prior to inputting the at least two sample data into the predictive model to be trained, the method further comprising:
determining the number of the obtained mixed sample data as a first number;
determining the number of sample data input into a prediction model to be trained as a second number according to the first number and a preset sample proportion;
inputting the at least two sample data into a prediction model to be trained so as to extract service features from the at least two sample data through a feature extraction layer in the prediction model, wherein the method specifically comprises the following steps of:
the first quantity of mixed sample data and the second quantity of sample data are input into a predictive model to be trained, so that the first quantity of mixed service features are extracted from the first quantity of mixed sample data through a feature extraction layer in the predictive model, and the second quantity of service features are respectively extracted from the second quantity of sample data.
6. A method of business wind control, comprising:
acquiring service data when a target user executes a service;
inputting the business data into a pre-trained prediction model to extract business features from the business data through a feature extraction layer in the prediction model, and inputting the business features into a prediction layer in the prediction model to enable the prediction layer to obtain a prediction result when a target user executes business according to the business features, wherein the prediction model is obtained by training the method according to any one of claims 1-5;
And carrying out service wind control according to the prediction result.
7. An apparatus for model training, comprising:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring service data of each user for executing service in history, and the service data comprises operation data of the user, basic data of the user, application data of the user and equipment data of terminal equipment of the user;
the construction module is used for constructing each sample data according to each acquired service data;
the mixing module is used for mixing the label data corresponding to at least two sample data according to a preset mixing mode to obtain mixed label data;
the extraction module is used for inputting the at least two sample data into a prediction model to be trained, extracting service features from the at least two sample data through a feature extraction layer in the prediction model, and inputting the service features extracted respectively into a feature mixing layer in the prediction model, so that the feature mixing layer mixes the service features extracted respectively according to the mixing mode to obtain mixed features;
the input module is used for inputting the mixed characteristics to a prediction layer in the prediction model so that the prediction layer obtains a first risk prediction result according to the mixed characteristics;
And the training module is used for training at least the feature extraction layer and the prediction layer in the prediction model by taking the deviation between the minimized first risk prediction result and the mixed label data as an optimization target, and the prediction model comprises a wind control model for carrying out business wind control.
8. A business air control device, comprising:
the acquisition module is used for acquiring service data when the target user executes the service;
the input module is used for inputting the business data into a pre-trained prediction model so as to extract business features from the business data through a feature extraction layer in the prediction model, and inputting the business features into a prediction layer in the prediction model so as to enable the prediction layer to obtain a prediction result when a target user executes the business according to the business features, wherein the prediction model is obtained through training by the method according to any one of claims 1-5;
and the wind control module is used for carrying out service wind control according to the prediction result.
9. A computer readable storage medium storing a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-6.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any of the preceding claims 1-6 when the program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211276984.7A CN115618962B (en) | 2022-10-18 | 2022-10-18 | Model training method, business wind control method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211276984.7A CN115618962B (en) | 2022-10-18 | 2022-10-18 | Model training method, business wind control method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115618962A CN115618962A (en) | 2023-01-17 |
| CN115618962B true CN115618962B (en) | 2023-05-23 |
Family
ID=84861788
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211276984.7A Active CN115618962B (en) | 2022-10-18 | 2022-10-18 | Model training method, business wind control method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115618962B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106548210A (en) * | 2016-10-31 | 2017-03-29 | 腾讯科技（深圳）有限公司 | Machine learning model training method and device |
| CN112686394A (en) * | 2021-01-05 | 2021-04-20 | 北京三快在线科技有限公司 | Model training method and device |
| CN113449753A (en) * | 2020-03-26 | 2021-09-28 | 中国电信股份有限公司 | Service risk prediction method, device and system |
| CN114298735A (en) * | 2021-12-07 | 2022-04-08 | 北京三快在线科技有限公司 | Model training method, information pushing method and device |
| CN114429283A (en) * | 2021-12-30 | 2022-05-03 | 中和农信项目管理有限公司 | Risk label processing method and device, wind control method and device and storage medium |
| CN114997317A (en) * | 2022-06-14 | 2022-09-02 | 蚂蚁区块链科技(上海)有限公司 | Method and device for training wind control model and predicting risk category |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10496902B2 (en) * | 2017-09-21 | 2019-12-03 | International Business Machines Corporation | Data augmentation for image classification tasks |
-
2022
- 2022-10-18 CN CN202211276984.7A patent/CN115618962B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106548210A (en) * | 2016-10-31 | 2017-03-29 | 腾讯科技（深圳）有限公司 | Machine learning model training method and device |
| CN113449753A (en) * | 2020-03-26 | 2021-09-28 | 中国电信股份有限公司 | Service risk prediction method, device and system |
| CN112686394A (en) * | 2021-01-05 | 2021-04-20 | 北京三快在线科技有限公司 | Model training method and device |
| CN114298735A (en) * | 2021-12-07 | 2022-04-08 | 北京三快在线科技有限公司 | Model training method, information pushing method and device |
| CN114429283A (en) * | 2021-12-30 | 2022-05-03 | 中和农信项目管理有限公司 | Risk label processing method and device, wind control method and device and storage medium |
| CN114997317A (en) * | 2022-06-14 | 2022-09-02 | 蚂蚁区块链科技(上海)有限公司 | Method and device for training wind control model and predicting risk category |
Non-Patent Citations (1)
| Title |
|---|
| 基于混合模型的水果需求量预测应用研究;严晨;中国优秀硕士学位论文全文数据库 经济与管理科学辑(1);J149-728 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115618962A (en) | 2023-01-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190354555A1 (en) | Dynamic faceting for personalized search and discovery | |
| CN111190718A (en) | Method, device and system for realizing task scheduling | |
| CN110457578B (en) | Customer service demand identification method and device | |
| CN118227914A (en) | Information display method and device, information processing method and device | |
| CN104504133A (en) | Application program recommending method and device | |
| CN112734498B (en) | Task rewarding acquisition method, device, terminal and storage medium | |
| CN112418920A (en) | Training method of information recommendation model, information recommendation method and device | |
| CN110489641A (en) | A kind of information recommendation data processing method and device | |
| CN108958828A (en) | Application program preloads method, apparatus, storage medium and terminal | |
| CN111241391A (en) | Task recommendation method and device | |
| CN115145801B (en) | A/B test flow distribution method, device, equipment and storage medium | |
| CN113886033A (en) | Task processing method and device | |
| CN111046156B (en) | Method, device and server for determining rewarding data | |
| CN111639269B (en) | Site recommendation method and device | |
| CN115618962B (en) | Model training method, business wind control method and device | |
| CN108595395B (en) | Nickname generation method, device and equipment | |
| CN110634024A (en) | User attribute marking method and device, electronic equipment and storage medium | |
| JP6916250B2 (en) | Virtual Assistant Domain Selection Analysis | |
| CN112307371B (en) | Applet sub-service identification method, device, equipment and storage medium | |
| CN114676272A (en) | Information processing method, device and equipment of multimedia resource and storage medium | |
| CN113672660A (en) | Data query method, device and equipment | |
| CN111461328A (en) | Neural network training method and electronic equipment | |
| CN111489163A (en) | Service processing method and device and electronic equipment | |
| CN111523866B (en) | Method and device for providing system operation enabling for user | |
| CN110659979A (en) | Method and apparatus for predicting loss rate of default, storage medium, and electronic 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 |