CN111784472A - Wind control method, device and system based on consumption data and readable storage medium - Google Patents
Wind control method, device and system based on consumption data and readable storage medium Download PDFInfo
- Publication number
- CN111784472A CN111784472A CN202010632412.2A CN202010632412A CN111784472A CN 111784472 A CN111784472 A CN 111784472A CN 202010632412 A CN202010632412 A CN 202010632412A CN 111784472 A CN111784472 A CN 111784472A
- Authority
- CN
- China
- Prior art keywords
- wind control
- consumption
- data
- equipment
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0609—Buyer or seller confidence or verification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Abstract
The invention discloses a wind control method, a wind control device, a wind control system and a readable storage medium based on consumption data, wherein the method is applied to wind control equipment, the wind control equipment is at least in communication connection with consumption equipment, and the method comprises the following steps: receiving risk subdata of a user to be predicted, which is transmitted by each consumption device, and aggregating the risk subdata to generate target risk data, wherein each risk subdata is based on a wind control model submodule in each consumption device and is used for performing risk prediction generation on the consumption subdata of the user to be predicted in each consumption device; and determining a wind control strategy of the user to be predicted according to the target risk data, and performing wind control on the user to be predicted according to the wind control strategy. The federal wind control model is generated through longitudinal federal training between each consumption device and each wind control device, so that the safety and the effectiveness of data used for training are ensured, and accurate wind control on users to be predicted is realized based on a large amount of safe and effective user data.
Description
Technical Field
The invention relates to the technical field of financial technology (Fintech), in particular to a wind control method, a wind control device, a wind control system and a readable storage medium based on consumption data.
Background
With the continuous development of financial technology (Fintech), especially internet technology and finance, more and more technologies (such as artificial intelligence, big data, cloud storage and the like) are applied to the financial field, but the financial field also puts higher requirements on various technologies, such as the requirement on improving the accuracy of user wind control.
Currently, the wind control of the user is generally performed by performing credit rating and evaluating a credit line by means of consumption behavior data of the user, and performing wind control on the credit rating and the credit line. However, the consumption behavior data of the user belongs to the privacy data of the user and is difficult to obtain in a large amount through a safe and effective way, so that the consumption behavior data for credit rating and credit line assessment is too small, and accurate wind control on the user is difficult to achieve.
Therefore, how to perform accurate wind control based on a large amount of safe and effective user data is a technical problem to be solved urgently at present.
Disclosure of Invention
The invention mainly aims to provide a consumption data-based wind control method, a consumption data-based wind control device, a consumption data-based wind control system and a readable storage medium, and aims to solve the technical problem of how to perform accurate wind control based on a large amount of safe and effective user data in the prior art.
In order to achieve the above object, the present invention provides a consumption data-based wind control method, which is applied to consumption data-based wind control equipment, where the wind control equipment is in communication connection with at least one consumption equipment, and the consumption data-based wind control method includes the following steps:
receiving risk subdata of a user to be predicted, which is transmitted by each consumption device, and aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and determining a wind control strategy of the user to be predicted according to the target risk data, and performing wind control on the user to be predicted according to the wind control strategy.
Optionally, after the step of aggregating the risk sub-data to generate the target risk data, the method includes:
determining a model type corresponding to each wind control sub-module, and if the model type is the wind control type, executing a step of determining a wind control strategy of the user to be predicted according to the target risk data;
and if the model type is an abnormality detection type, performing abnormality detection on the user to be predicted according to the abnormality score corresponding to the target risk data.
Optionally, before the step of receiving the risk sub-data of the user to be predicted transmitted by each consumption device, and aggregating each risk sub-data to generate the target risk data, the method further includes:
carrying out sample alignment or encrypted sample alignment with each consumption device, and receiving sample characteristic data determined by the sample alignment in each consumption device to respectively train a preset wind control sub-module in each consumption device to generate an intermediate result;
aggregating the intermediate results to generate gradient information, and transmitting the gradient information back to the consumption equipment so as to perform federal training by combining sample label data determined by sample alignment in the wind control equipment and sample characteristic data of the consumption equipment;
and when detecting that the federal training meets a preset finishing condition, finishing the federal training, and respectively generating preset wind control sub-modules in the consumption equipment into the wind control sub-modules.
Optionally, the preset ending condition includes a preset training frequency, a preset training duration, or a preset loss function convergence of the federal model, and when detecting that the federal training satisfies the preset ending condition, the step of completing the federal training includes:
when the counted number of times of the federal training reaches the preset number of times, completing the federal training;
or when the statistical result shows that the training time of the federal training reaches the preset training time, completing the federal training;
or when the convergence of the loss function of the preset federated model is detected, completing the federated training.
Further, in order to achieve the above object, the present invention further provides a consumption data-based wind control method, which is applied to consumption equipment, and the consumption data-based wind control method includes the following steps:
carrying out sample alignment or encrypted sample alignment with a wind control device, and determining sample characteristic data of the consumption device;
training a preset wind control submodule in the consumption equipment based on the sample characteristic data, generating an intermediate result and transmitting the intermediate result to the wind control equipment;
receiving gradient information generated after the wind control equipment aggregates the intermediate results, and performing federal model training on sample label data determined by sample alignment in the wind control equipment and sample characteristic data in the consumption equipment based on the gradient information;
and when a federal training end instruction is received, stopping the federal training, and generating a preset wind control submodule in the consumption equipment into a wind control submodule in the consumption equipment so as to perform wind control based on the wind control submodule.
Optionally, after the step of generating the preset wind control submodule in the consumer device into the wind control submodule in the consumer device, the method further includes:
and when the data volume of the updating data in the consumption equipment is detected to be larger than a preset threshold value, performing sample alignment or encrypted sample alignment with the wind control equipment based on the updating data, and determining sample characteristic data of the consumption equipment so as to update the wind control submodule in the consumption equipment based on the updating data.
Optionally, before the step of sample aligning or encrypted sample aligning with a wind control device and determining sample characteristic data of the consumption device, the method comprises:
according to a preset data derivation model, deriving the consumption subdata in the consumption equipment to generate derived data corresponding to the consumption subdata, and determining sample characteristic data of the consumption equipment based on the derived data and the consumption subdata after sample alignment is carried out on the consumption subdata and the wind control equipment.
Further, to achieve the above object, the present invention further provides a consumption data-based wind control device, including:
the aggregation module is used for receiving risk subdata of the user to be predicted transmitted by each consumption device and aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and the wind control module is used for determining the wind control strategy of the user to be predicted according to the target risk data and carrying out wind control on the user to be predicted according to the wind control strategy.
Further, to achieve the above object, the present invention also provides a consumption data based wind control system, which includes a memory, a processor, and a consumption data based wind control program stored in the memory and executable on the processor, wherein the consumption data based wind control program, when executed by the processor, implements the steps of the consumption data based wind control method as described above.
Further, to achieve the above object, the present invention also provides a readable storage medium, on which a consumption data based wind control program is stored, and the consumption data based wind control program, when executed by a processor, implements the steps of the consumption data based wind control method as described above.
The consumption data-based wind control method is applied to wind control equipment, and the wind control equipment is in communication connection with at least one consumption equipment. Compared with the prior art that the consumption behavior data for credit rating and credit line assessment are too little to cause difficulty in accurate wind control, the method and the system have the advantages that after the risk subdata of the user to be predicted, which is transmitted by each consumption device, is received, the risk subdata is aggregated to generate target risk data; the method comprises the steps that a wind control submodule is trained in each consumption device through a longitudinal federation in advance, a user to be predicted generates different types of consumption subdata in each consumption device, the risk prediction is carried out on the respective consumption subdata through the wind control submodule in each consumption device, and the generated risk subdata is transmitted to the wind control device to be aggregated; and after the target risk data are obtained through aggregation, determining a wind control strategy of the user to be predicted according to the risk represented by the target risk data, and further performing wind control on the user to be predicted according to the wind control strategy. Because the wind control submodule combines a large number of sample characteristic data of users at each consumption device and sample label data of the wind control device, longitudinal federal training generation is carried out, and the sample characteristic data of each consumption device and the sample label data of the wind control device are not transmitted out of the device where the wind control submodule is originally arranged, so that the safety and the effectiveness of the data used for training are ensured, and the accuracy of wind control is facilitated by a large number of training data. The defect that accurate wind control is difficult due to the fact that consumption behavior data used for credit rating and credit line assessment are too little in the prior art is overcome; the accurate wind control on the user to be predicted based on a large amount of safe and effective user data is realized.
Drawings
FIG. 1 is a schematic structural diagram of a hardware operating environment of a device according to an embodiment of a consumption data-based wind control system of the present invention;
FIG. 2 is a schematic flow chart of a first embodiment of a consumption data-based wind control method according to the present invention;
FIG. 3 is a flowchart illustrating a first embodiment of a consumption data-based wind control method according to the present invention;
FIG. 4 is a flowchart illustrating a first embodiment of a consumption data based wind control method according to the present invention;
FIG. 5 is a functional block diagram of a preferred embodiment of a consumption data based wind control apparatus according to the present invention;
fig. 6 is a schematic diagram of communication connection between the wind control device and the consumption device for federal training in the consumption data based wind control method of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a consumption data-based wind control system, which comprises wind control equipment and at least one consumption equipment in communication connection with the wind control equipment, and referring to fig. 1, fig. 1 is a schematic structural diagram of an equipment hardware operating environment related to an embodiment of the consumption data-based wind control system.
As shown in fig. 1, the consumption data-based wind control system may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a memory device separate from the processor 1001 described above.
Those skilled in the art will appreciate that the hardware configuration of the consumption data based wind control system shown in fig. 1 does not constitute a limitation of the consumption data based wind control system, and may include more or less components than those shown, or some components in combination, or a different arrangement of components.
As shown in fig. 1, an operating system, a network communication module, a user interface module, and a consumption data-based wind control program may be included in a memory 1005, which is a readable storage medium. The operating system is a program for managing and controlling the wind control system and the software resources based on the consumption data, and supports the operation of a network communication module, a user interface module, the wind control program based on the consumption data and other programs or software; the network communication module is used to manage and control the network interface 1004; the user interface module is used to manage and control the user interface 1003.
In the hardware structure of the consumption data-based wind control system shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server and performing data communication with the background server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; the processor 1001 may call a consumption data based wind control program stored in the memory 1005 and perform the following operations:
receiving risk subdata of a user to be predicted, which is transmitted by each consumption device, and aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and determining a wind control strategy of the user to be predicted according to the target risk data, and performing wind control on the user to be predicted according to the wind control strategy.
Further, after the step of aggregating the risk sub-data to generate the target risk data, the processor 1001 may call the consumption data-based wind control program stored in the memory 1005, and perform the following operations:
determining a model type corresponding to each wind control sub-module, and if the model type is the wind control type, executing a step of determining a wind control strategy of the user to be predicted according to the target risk data;
and if the model type is an abnormality detection type, performing abnormality detection on the user to be predicted according to the abnormality score corresponding to the target risk data.
Further, before the step of receiving the risk sub-data of the user to be predicted transmitted by each consumption device, and aggregating each risk sub-data to generate the target risk data, the processor 1001 may invoke the consumption data-based wind control program stored in the memory 1005, and perform the following operations:
carrying out sample alignment or encrypted sample alignment with each consumption device, and receiving sample characteristic data determined by the sample alignment in each consumption device to respectively train a preset wind control sub-module in each consumption device to generate an intermediate result;
aggregating the intermediate results to generate gradient information, and transmitting the gradient information back to the consumption equipment so as to perform federal training by combining sample label data determined by sample alignment in the wind control equipment and sample characteristic data of the consumption equipment;
and when detecting that the federal training meets a preset finishing condition, finishing the federal training, and respectively generating preset wind control sub-modules in the consumption equipment into the wind control sub-modules.
Further, the preset ending condition includes preset training times, preset training duration or preset loss function convergence of the federal model, and when detecting that the federal training meets the preset ending condition, the step of completing the federal training includes:
when the counted number of times of the federal training reaches the preset number of times, completing the federal training;
or when the statistical result shows that the training time of the federal training reaches the preset training time, completing the federal training;
or when the convergence of the loss function of the preset federated model is detected, completing the federated training.
Further, the processor 1001 may call a consumption data based wind control program stored in the memory 1005 and perform the following operations:
carrying out sample alignment or encrypted sample alignment with a wind control device, and determining sample characteristic data of the consumption device;
training a preset wind control submodule in the consumption equipment based on the sample characteristic data, generating an intermediate result and transmitting the intermediate result to the wind control equipment;
receiving gradient information generated after the wind control equipment aggregates the intermediate results, and performing federal model training on sample label data determined by sample alignment in the wind control equipment and sample characteristic data in the consumption equipment based on the gradient information;
and when a federal training end instruction is received, stopping the federal training, and generating a preset wind control submodule in the consumption equipment into a wind control submodule in the consumption equipment so as to perform wind control based on the wind control submodule.
Further, after the step of generating the preset wind control sub-module in the consumption device as the wind control sub-module in the consumption device, the processor 1001 may call the wind control program based on the consumption data stored in the memory 1005, and perform the following operations:
and when the data volume of the updating data in the consumption equipment is detected to be larger than a preset threshold value, performing sample alignment or encrypted sample alignment with the wind control equipment based on the updating data, and determining sample characteristic data of the consumption equipment so as to update the wind control submodule in the consumption equipment based on the updating data.
Further, before the step of performing sample alignment or encrypted sample alignment with the wind control device and determining the sample characteristic data of the consumption device, the processor 1001 may call a consumption data-based wind control program stored in the memory 1005, and perform the following operations:
according to a preset data derivation model, deriving the consumption subdata in the consumption equipment to generate derived data corresponding to the consumption subdata, and determining sample characteristic data of the consumption equipment based on the derived data and the consumption subdata after sample alignment is carried out on the consumption subdata and the wind control equipment.
The specific implementation of the consumption data-based wind control system of the present invention is substantially the same as the following embodiments of the consumption data-based wind control method, and is not described herein again.
The invention also provides a wind control method based on the consumption data.
Referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of a consumption data-based wind control method according to the present invention.
While a logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than presented herein. Specifically, the consumption data based wind control method in this embodiment is applied to a wind control device, where the wind control device is in communication connection with at least one consumption device, and the consumption data based wind control method includes:
step S10, receiving risk subdata of a user to be predicted transmitted by each consumption device, aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
the consumption data-based wind control method in this embodiment is applied to wind control equipment, and the wind control equipment is deployed in an organization with wind control requirements, such as a financial institution like a bank, an insurance company, a financial technology company, and the like. Meanwhile, the wind control device is in communication connection with at least one consumption device, and the consumption device is deployed in an organization which embodies user consumption data, such as a water supply company, an electric power company, a gas company, and the like, and refer to fig. 6 specifically. And carrying out federal training on data between the wind control equipment and the consumption equipment jointly to obtain an intelligent wind control model for risk control. The federal training is preferably longitudinal federal training, and in the longitudinal federal training process, data in the consumption equipment are located in the consumption equipment and are not transmitted out of the consumption equipment; the data in the wind control equipment are located in the wind control equipment and do not transmit out of the wind control equipment, so that the safety of the data in each equipment is ensured. And the intelligent wind control model formed by training is composed of wind control submodules distributed in each device, namely after the training is finished, the wind control submodules are formed in the consumption devices, the wind control submodules are also formed in the wind control devices, and the intelligent wind control model is formed by all the wind control submodules together.
Further, when the financial institution receives an application item of a certain user, such as a loan item, and has a wind control demand for the user, the financial institution uses the user as a user to be predicted, and performs risk prediction on the consumption subdata of the user to be predicted in each consumption device by the wind control sub-module in each consumption device to generate the risk subdata. The consumer equipment such as the consumer equipment which is docked with the wind control equipment comprises first consumer equipment positioned at a tap water company, second consumer equipment positioned at an electric power company and third consumer equipment positioned at a gas company; carrying out risk prediction on tap water data of a user to be predicted in the first consumption terminal by a wind control submodule in the first consumption terminal to generate first risk subdata; the wind control submodule in the second consumption terminal carries out risk prediction on the power data of the user to be predicted in the second consumption terminal to generate second risk subdata; and carrying out risk prediction on the gas data of the user to be predicted in the third consumption terminal by a wind control submodule in the third consumption terminal to generate third risk subdata. And the wind control sub-modules transmit the generated risk subdata to the wind control sub-modules in the wind control equipment, and the wind control sub-modules in the wind control equipment receive the various risk subdata and aggregate the various risk subdata to generate target risk data representing the risk of the user to be predicted. The aggregation mode can be set according to requirements, if mean aggregation is set to be performed on each risk subdata, mean processing is performed on each risk subdata, a mean result is obtained and serves as target risk data, and the average risk of the user to be predicted on various consumption data is represented. Or setting the maximum aggregation of the risk subdata, comparing the risk subdata, searching the maximum value of the risk subdata, using the maximum value as target risk data, and representing the maximum risk of the user to be predicted on various consumption data.
Step S20, determining the wind control strategy of the user to be predicted according to the target risk data, and performing wind control on the user to be predicted according to the wind control strategy.
Furthermore, after the wind control sub-module of the wind control equipment generates the target risk data, the wind control strategy of the user to be predicted is determined according to the target risk data. The target risk data can exist in a vector form or a numerical form and represent the risk score. Different risk scores correspond to different credit ratings and/or credit lines, which characterize the credit rating of the user to be predicted, and the rights on the requested item, such as a loan line on a loan item. The wind control strategy represents control strategies for different credit ratings and credit lines, including strategies before, in the middle of affairs and after affairs. When the credit rating is low and the credit line is low corresponding to higher risk value and the risk representing the user to be predicted is higher, the wind control strategy can be pre-refusal, such as refusal of application, refusal of request and the like; or prompt in time, such as sending prompt information at regular time; and the method can also be disabled afterwards, such as stopping use, locking and the like. The corresponding relation among risk data, credit rating, credit line and wind control strategy can be preset; after the target risk data are generated, according to the corresponding relation, the credit rating and the credit line can be determined, and according to the credit rating and the credit line, the wind control strategy is determined. And further carrying out wind control on the user to be predicted according to the wind control strategy so as to avoid financial institution risks caused by paying of the user to be predicted.
Further, during the process of data joint training in the wind control device and the consumption device, the tag data representing the risk, such as the data tag Y in fig. 6, may be provided by the wind control device; feature data characterizing the user features are provided by each consumer device, such as X1, X2, and X3 in fig. 6, where X1, X2, and X3 may exist in vector form or scalar form, which is not limited thereto. And performing longitudinal federal training by combining the label data in the wind control equipment and various characteristic data in each consumption equipment to obtain an intelligent wind control model for performing wind control on the user with consumption sub data in each consumption terminal. In addition, longitudinal federal training may also be provided with feature data only by consumer devices, with only aggregation by wind control devices, and no tag data. And performing longitudinal training by combining various characteristic data in each consumption device to obtain an abnormality detection model for detecting the abnormality of the user with the consumption sub data in each consumption terminal. The anomaly detection model and the intelligent wind control model have different processing modes for target risk data. Specifically, after the step of aggregating the risk sub-data to generate the target risk data, the method further includes:
a1, determining a model type corresponding to each wind control submodule, and if the model type is the wind control type, executing a step of determining a wind control strategy of the user to be predicted according to the target risk data;
step a2, if the model type is an abnormality detection type, performing abnormality detection on the user to be predicted according to the abnormality score corresponding to the target risk data.
Furthermore, after the target risk data are generated, the target risk data are processed in different modes according to different model types which are commonly corresponding to the wind control sub-modules. The model type corresponding to each wind control submodule is characterized in that the model formed by each wind control submodule is an intelligent wind control model or an abnormal detection model. And different models are distinguished by different identifiers, the intelligent wind control model is distinguished as a wind control type through the identifier, and the abnormal detection model is distinguished as an abnormal detection model through the identifier. And after the target risk data are generated, determining the model type which corresponds to each wind control submodule through the identification in the wind control equipment. And if the determined model type is the wind control type, the model obtained by the longitudinal federal training is an intelligent wind control model, and the generated target risk data is data for wind control, so that the wind control strategy of the user to be predicted is determined, and wind control is performed on the user to be predicted according to the wind control strategy.
Further, if the model type determined by the identification is an anomaly detection type, the model obtained by representing the longitudinal federal training is an anomaly detection model, and the generated target risk data is data used for anomaly detection, so that anomaly detection is performed on the user to be predicted according to the anomaly score corresponding to the target risk data. Wherein the target risk data is in the form of vectors or values, different vectors or values characterizing different anomaly scores. And presetting a corresponding relation between the vector or the numerical value and the score, searching an abnormal score corresponding to the target risk data according to the corresponding relation, and representing the abnormality of the user to be predicted by the abnormal score. The detected abnormality includes at least fraud identification, user identity abnormality identification, and the like.
The wind control method based on the consumption data is applied to wind control equipment, and the wind control equipment is in communication connection with at least one consumption equipment. Compared with the prior art that the consumption behavior data for credit rating and credit line assessment are too little to cause difficulty in accurate wind control, the method and the system have the advantages that after the risk subdata of the user to be predicted, which is transmitted by each consumption device, is received, the risk subdata is aggregated to generate target risk data; the method comprises the steps that a wind control submodule is trained in each consumption device through a longitudinal federation in advance, a user to be predicted generates different types of consumption subdata in each consumption device, the risk prediction is carried out on the respective consumption subdata through the wind control submodule in each consumption device, and the generated risk subdata is transmitted to the wind control device to be aggregated; and after the target risk data are obtained through aggregation, determining a wind control strategy of the user to be predicted according to the risk represented by the target risk data, and further performing wind control on the user to be predicted according to the wind control strategy. Because the wind control submodule combines a large number of sample characteristic data of users at each consumption device and sample label data of the wind control device, longitudinal federal training generation is carried out, and the sample characteristic data of each consumption device and the sample label data of the wind control device are not transmitted out of the device where the wind control submodule is originally arranged, so that the safety and the effectiveness of the data used for training are ensured, and the accuracy of wind control is facilitated by a large number of training data. The defect that accurate wind control is difficult due to the fact that consumption behavior data used for credit rating and credit line assessment are too little in the prior art is overcome; the accurate wind control on the user to be predicted based on a large amount of safe and effective user data is realized.
Further, referring to fig. 3, a second embodiment of the consumption data-based wind control method according to the present invention is provided based on the first embodiment of the consumption data-based wind control method according to the present invention.
The second embodiment of the consumption data-based wind control method is different from the first embodiment of the consumption data-based wind control method in that, before the steps of receiving the risk subdata of the user to be predicted transmitted by each consumption device, aggregating the risk subdata, and generating the target risk data, the method further includes:
step S30, performing sample alignment or encrypted sample alignment with each consumption device, and receiving sample characteristic data determined by the sample alignment in each consumption device to respectively train a preset wind control sub-module in each consumption device to generate an intermediate result;
step S40, aggregating the intermediate results to generate gradient information and transmitting the gradient information back to the consumption equipment so as to perform federal training by combining the sample label data determined by sample alignment in the wind control equipment and the sample characteristic data of the consumption equipment;
and step S50, when detecting that the federal training meets the preset end conditions, finishing the federal training, and respectively generating the preset wind control sub-modules in the consumer equipment into the wind control sub-modules.
The embodiment relates to a training part performed by the wind control equipment in the process of performing longitudinal federal training by combining label data in the wind control equipment with feature data in each consumer equipment. Specifically, sample alignment is performed between the wind control device and each consumer device, and the same user is determined between the mechanisms through the sample alignment. In addition, to further ensure the privacy of data between the organizations, the sample alignment may be performed in an encrypted manner, such as in a hash encryption alignment to determine the same users between the organizations. Meanwhile, a preset wind control submodule is preset in each consumption device, consumption subdata of the same user in each consumption device is respectively transmitted to the preset wind control submodule of each consumption device, a preset federal model of each consumption device is trained, and a training intermediate result is obtained and transmitted to the wind control device. And aggregating the intermediate result by the wind control equipment, generating gradient information, and transmitting the gradient information back to each consumption equipment, so that the label value after sample alignment in the wind control equipment is used as sample label data, and the characteristic data after sample alignment in the consumption equipment is used as sample characteristic data, thereby realizing the longitudinal federal training by combining the sample label data and each item of sample characteristic data. It should be noted that the intermediate result may exist in a vector form or a scalar form; the intermediate results u1, u2, and u3 shown in fig. 6 are in vector form. And the gradient information obtained by aggregation is divided into gradient information of the loss function of the preset federal model formed by each wind control submodule in the training process to each intermediate result, such as the gradient information of the loss function to u1, the gradient information to u2 and the gradient information to u3 shown in fig. 6, which are respectively transmitted back to a water company, an electric power company and a gas company.
Further, preset ending conditions of the federal training are preset, and whether the federal training reaches the preset ending conditions or not is judged in the process of performing longitudinal federal training by combining the sample label data and the characteristic data of each sample. If the preset ending condition is met, ending the federal training; otherwise, continuing training until reaching the preset end condition. Specifically, the preset termination condition includes any one of a preset training frequency, a preset training duration, or a preset loss function convergence of the federal model; when detecting that the federal training meets the preset end condition, the step of completing the federal training comprises the following steps:
b1, when the number of times of federal training reaches the preset number of times, completing the federal training;
or step b2, when the training time of the federal training reaches the preset training time, the federal training is completed;
or, step b3, when detecting the convergence of the loss function of the preset federal model, completing the federal training.
Furthermore, when the preset ending condition is the preset training times, aggregation is performed on the wind control equipment once, and counting statistics is performed on the aggregation times to obtain the training times of the federal training. And comparing the obtained training times with the preset training times, and judging whether the training times reach the preset training times. And if the preset training times are reached, indicating that the federal training meets the preset end condition, and finishing the federal training. And if the training times do not reach the preset training times, continuing to perform longitudinal federal training by combining the sample label data of the wind control equipment and the sample characteristic data of the consumption equipment until the counted training times reach the preset training times, and finishing the federal training.
In addition, for the situation that the preset ending condition is the preset training duration, the joint training time between the wind control equipment and each consumption equipment is counted to obtain the training duration of the federal training. And comparing the obtained training time with the preset training time, and judging whether the training time reaches the preset training time. And if the preset training time is reached, indicating that the federal training meets the preset ending condition, and finishing the federal training. And if the training time length does not reach the preset training time length, continuing to perform longitudinal federal training by combining the sample label data of the wind control equipment and the sample characteristic data of the consumption equipment until the counted training time length reaches the preset training time length, and finishing the federal training.
Further, for the case that the preset ending condition is the convergence of the loss function of the preset federal model, the model formed by each wind control submodule in the training process is the preset federal model, and the intelligent wind control model formed by each wind control submodule after the training is the intelligent wind control model formed by the preset federal model. And presetting a loss function of the federal model as a system loss function between the wind control equipment and a system formed by each consumer equipment. And after the wind control equipment aggregates the intermediate results, calculating the model gradient of the loss function on the intermediate results transmitted by the consumption equipment respectively. And judging whether the preset loss function converges or not through the stability of the model gradient. When the gradient of the model is stable within the continuous times, judging that the loss function of the preset federal model is converged, and completing the federal training when the federal training meets the preset end condition; and when the gradient of the model is unstable within the continuous times, judging that the preset loss function is not converged, and the federal training does not meet the preset ending condition, and continuing the federal training until the loss function of the preset federal model is converged to finish the federal training. It should be noted that, because of the large number of consumption devices, the preset loss function may be stable for the model gradient of the intermediate result transmitted by one consumption device, and unstable for the model gradient of the intermediate result transmitted by other consumption devices; at the moment, the federal training does not meet the preset end condition, the stable model gradient is matched with the unstable model gradient to continue the training until the model gradient of each consumption device reaches a relatively stable state, and the federal training is judged to meet the preset end condition, so that the federal training is completed.
In the embodiment, the sample label data of the wind control equipment and the sample characteristic data of each consumption equipment are jointly subjected to longitudinal federal training, so that an intelligent wind control model containing a wind control submodule inside each consumption equipment is obtained for wind control. Because the sample label data of the wind control equipment and the sample characteristic data of each consumption equipment are not transmitted out of the equipment where the wind control equipment and the consumption equipment are located, the privacy of each item of data is protected. Meanwhile, due to the fact that a large amount of sample label data and sample characteristic data are combined for training, the accuracy of the intelligent wind control model of the user wind control obtained through training is guaranteed.
Further, referring to fig. 4, fig. 4 is a flowchart illustrating a consumption data-based wind control method according to a third embodiment of the present invention. Specifically, the consumption data-based wind control method in this embodiment is applied to a consumption device, and includes:
step S60, carrying out sample alignment or encrypted sample alignment with the wind control equipment, and determining sample characteristic data of the consumption equipment;
the embodiment relates to a training part of consumer equipment in a longitudinal federal training process of combining label data in wind control equipment and feature data in the consumer equipment. Because training between each consumption device has similarity, only the combined characteristic data types have difference; for example, the type of the characteristic data used for the consumer in the water company is water fee data, and the type of the characteristic data used for the consumer in the electric power company is electric power fee data. The present embodiment is described by taking the training of feature data provided in a consumer device as an example. In particular, sample alignment is performed between the wind control device and the consumer device, or in an encrypted manner, such as hash encryption alignment. Determining users who are the same among the organizations through sample alignment or encrypted sample alignment; and further determining feature data of the same user in the consumption equipment as sample feature data, and performing longitudinal federal training by combining the sample label data in the wind control equipment.
In addition, in order to further enrich sample feature data for embodying user features, the present embodiment is provided with a mechanism that expands on the basis of consumption sub data. Specifically, before the step of sample aligning with the wind control device and determining sample characteristic data of the consumer device, the method comprises:
and c, according to a preset data derivative model, deriving the consumption subdata in the consumption equipment to generate derivative data corresponding to the consumption subdata, and determining sample characteristic data of the consumption equipment based on the derivative data and the consumption subdata after sample alignment is carried out on the consumption subdata and the wind control equipment.
Further, a preset data derivative model for deriving the sample characteristic data is preset, and the derived data comprises but is not limited to a living address, whether the living room is owned, family population, age scale, business trip condition, home condition, work and rest habits, eating habits and the like. The method comprises the steps of obtaining derived data of a user from which consumer data is obtained in advance, forming a corresponding relation between the consumer data and derived numerical values as training data, training a preset data derived model through the training data, and forming a preset data derived model which can be finally used for deriving the derived data. Before determining that the sample characteristic data of the consumption equipment is subjected to longitudinal federal training, calling a trained preset data derivative model, and deriving the consumption subdata by using the preset data derivative model to generate derivative data corresponding to the consumption subdata. If the number of the associated family population is b1 to b2 for the range of the water fee from a1 to a2 in the consumption subdata in the preset data derivative model, if the water fee of the consumption subdata in the consumption device is a3 between a1 to a2, the derived family population is a value from b1 to b 2. Therefore, a plurality of items of derivative data corresponding to the consumption subdata are obtained through derivation, and the sample characteristic data is determined based on the derivative data and the consumption subdata, so that the sample characteristic data used for training is expanded, and the trained wind control submodule in the consumption equipment is more accurate.
Step S70, training a preset wind control submodule in the consumption equipment based on the sample characteristic data, generating an intermediate result and transmitting the intermediate result to the wind control equipment;
step S80, receiving gradient information generated by the wind control equipment after the intermediate results are aggregated, and performing federal model training based on the gradient information in combination with sample label data determined by sample alignment in the wind control equipment and sample characteristic data in the consumption equipment;
and step S90, when a federal training end instruction is received, stopping the federal training, and generating a preset wind control submodule in the consumption equipment into the wind control submodule in the consumption equipment so as to perform wind control based on the wind control submodule.
Furthermore, a preset wind control submodule is preset in the consumption equipment, after sample characteristic data are obtained through sample alignment, the sample characteristic data are transmitted to the preset wind control submodule, the preset wind control submodule is trained, and an intermediate result is generated and transmitted to the wind control equipment. And the wind control equipment carries out aggregation processing on the intermediate results from the consumption equipment to generate gradient information and transmits the gradient information back to the consumption equipment. After the consumption equipment receives the gradient information, the preset wind control submodule is updated through the gradient information, and the updated preset wind control submodule is trained through the sample characteristic data. The wind control equipment is provided with sample label data after sample alignment, and aggregation of intermediate results has correlation with the sample label data, so that the gradient information reflects the sample label data, and accordingly training of a preset federal model updated by the gradient information by using the sample characteristic data reflects a process of federal training combining the sample label data in the wind control equipment and the sample characteristic data in the consumption equipment.
Further, preset ending conditions for representing the ending of the federal training are preset in the wind control equipment, and the preset ending conditions include, but are not limited to, preset training times, preset training duration or preset loss function convergence of the federal model. And when the wind control equipment judges that the federal training meets the preset end condition, the federal training is finished, and a federal training end instruction is generated and sent to the consumption equipment. When the consumption equipment receives the federal training end instruction, the federal training is stopped, and the preset wind control submodule in the consumption terminal is generated into the wind control submodule, so that wind control can be performed according to the wind control submodule subsequently, wind control can be performed by combining sample characteristic data in each consumption terminal, and accuracy of the wind control is improved.
Understandably, the consumption subdata in each consumption terminal has an updating characteristic, so that in order to further improve the accuracy of the wind control submodule, the embodiment is provided with an updating training mechanism for the trained wind control submodule. Specifically, after the step of generating the preset federal model in the consumer device into the wind control submodule in the consumer device, the method further includes:
and d, when the fact that the data volume of the updating data in the consumption equipment is larger than a preset threshold value is detected, performing sample alignment or encrypted sample alignment with the wind control equipment based on the updating data, and determining sample characteristic data of the consumption equipment so as to update the wind control submodule in the consumption equipment based on the updating data.
Further, a trigger condition for updating the training is preset, where the trigger condition may be a preset threshold or a preset period, and the preset threshold is taken as an example for description in this embodiment. Specifically, the data volume of the update data in the consumer device is counted, and whether the data volume of the update data is larger than a preset threshold is detected. If the value is larger than the preset threshold value, the triggering condition of the updating training is reached, and the updating training is triggered. At this time, because the consumption subdata in the consumption equipment is changed due to the updated data, sample alignment or encrypted sample alignment is performed between the consumption equipment and the wind control equipment, sample characteristic data is determined from the updated consumption subdata, and longitudinal federal training is performed on the sample characteristic data and the sample label data in the wind control equipment, so that the wind control submodule in the consumption equipment is updated. Through the updating of the wind control sub-modules, the intelligent wind control model jointly formed by the wind control sub-modules in the consumption equipment is updated, and the wind control accuracy of the intelligent wind control model is improved through richer training data.
It should be noted that, an intelligent wind control model composed of trained wind control submodules in each consumer device is provided with feedback data for receiving and representing wind control accuracy in the use process. Therefore, in this embodiment, in addition to triggering the update training by the consumption sub data updated by each consumption device, the update training may also be triggered by the feedback data. Similarly, a preset threshold value or a preset period is preset, when the data volume of the feedback data reaches the preset threshold value or the service life of the intelligent wind control model reaches the preset period, updating training is triggered, sample alignment is carried out between the consumption equipment and the wind control equipment, sample characteristic data and sample label data in the wind control equipment are determined to carry out longitudinal federal training, so that the wind control sub-modules are updated, and the wind control accuracy of the intelligent wind control model formed by the wind control sub-modules is improved.
In the embodiment, in the process of carrying out the vertical federal training on the sample label data of the wind control equipment and the sample characteristic data of each consumer equipment to obtain the intelligent wind control model containing the wind control sub-modules positioned in each consumer equipment to carry out the wind control, the derivative data corresponding to the consumption sub-data are derived through the preset derivative model, so that the characteristics contained in each sample characteristic data are more diverse, and the accuracy of the wind control sub-modules obtained through the federal training and the accuracy of the intelligent wind control model consisting of the wind control sub-modules are ensured. Meanwhile, an updating training mechanism is set for the trained wind control sub-modules, and the wind control sub-modules are updated by using a large amount of updating sample characteristic data, so that the accuracy of the wind control sub-modules and an intelligent wind control model formed by the wind control sub-modules is further improved.
The invention also provides a wind control device based on the consumption data.
Referring to fig. 5, fig. 5 is a functional block diagram of a first embodiment of a consumption data-based wind control device according to the present invention. The consumption data-based wind control device comprises:
the aggregation module 10 is configured to receive risk subdata of a user to be predicted, which is transmitted by each consumption device, and aggregate the risk subdata to generate target risk data, where each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and the wind control module 20 is configured to determine a wind control strategy of the user to be predicted according to the target risk data, and perform wind control on the user to be predicted according to the wind control strategy.
Further, the consumption data-based wind control device further comprises:
the determining module is used for determining a model type corresponding to each wind control submodule, and if the model type is the wind control type, the step of determining the wind control strategy of the user to be predicted according to the target risk data is executed;
and the detection module is used for carrying out abnormality detection on the user to be predicted according to the abnormality score corresponding to the target risk data if the model type is an abnormality detection type.
Further, the consumption data-based wind control device further comprises:
the receiving module is used for carrying out sample alignment or encrypted sample alignment with each consumption device and receiving an intermediate result generated by training a preset wind control sub-module in each consumption device respectively through sample characteristic data determined by the sample alignment in each consumption device;
the generating module is used for aggregating the intermediate results, generating gradient information and transmitting the gradient information back to the consuming equipment so as to perform federal training by combining the sample label data determined by sample alignment in the wind control equipment and the sample characteristic data of the consuming equipment;
and the completion module is used for completing the federal training when detecting that the federal training meets a preset completion condition, and respectively generating preset wind control sub-modules in the consumer equipment into the wind control sub-modules.
Further, the preset termination condition includes a preset training frequency, a preset training duration, or a preset loss function convergence of the federal model, and the generation module is further configured to:
when the counted number of times of the federal training reaches the preset number of times, completing the federal training;
or when the statistical result shows that the training time of the federal training reaches the preset training time, completing the federal training;
or when the convergence of the loss function of the preset federated model is detected, completing the federated training.
Further, the present invention also provides a consumption data-based wind control device, which includes:
the sample alignment module is used for carrying out sample alignment or encrypted sample alignment with the wind control equipment and determining sample characteristic data of the consumption equipment;
the training module is used for training a preset wind control submodule in the consumption equipment based on the sample characteristic data, generating an intermediate result and transmitting the intermediate result to the wind control equipment;
the training module is further used for receiving gradient information generated by the wind control equipment after the intermediate results are aggregated, and performing federal model training based on the gradient information in combination with sample label data determined by sample alignment in the wind control equipment and sample characteristic data in the consumption equipment;
and the stopping module is used for stopping the federal training when a federal training ending instruction is received, and generating a preset wind control submodule in the consumption equipment into the wind control submodule in the consumption equipment so as to perform wind control based on the wind control submodule.
Further, the consumption data-based wind control device further comprises:
and the execution module is used for executing sample alignment or encrypted sample alignment with the wind control equipment based on the update data when detecting that the data volume of the update data in the consumption equipment is larger than a preset threshold value, and determining sample characteristic data of the consumption equipment so as to update the wind control submodule in the consumption equipment based on the update data.
Further, the consumption data-based wind control device further comprises:
and the derivation module is used for deriving the consumption subdata in the consumption equipment according to a preset data derivation model, generating derived data corresponding to the consumption subdata, and determining sample characteristic data of the consumption equipment based on the derived data and the consumption subdata after sample alignment is carried out on the wind control equipment.
The specific implementation of the consumption data-based wind control device of the present invention is substantially the same as that of the above-mentioned embodiments of the consumption data-based wind control method, and is not described herein again.
In addition, the embodiment of the invention also provides a readable storage medium.
The readable storage medium stores a consumption data-based wind control program, and the consumption data-based wind control program realizes the steps of the consumption data-based wind control method when being executed by the processor.
The readable storage medium of the present invention may be a computer readable storage medium, and the specific implementation manner of the readable storage medium of the present invention is substantially the same as that of each embodiment of the wind control method based on consumption data, and will not be described herein again.
The present invention is described in connection with the accompanying drawings, but the present invention is not limited to the above embodiments, which are only illustrative and not restrictive, and those skilled in the art can make various changes without departing from the spirit and scope of the invention as defined by the appended claims, and all changes that come within the meaning and range of equivalency of the specification and drawings that are obvious from the description and the attached claims are intended to be embraced therein.
Claims (10)
1. A consumption data-based wind control method is applied to wind control equipment, the wind control equipment is in communication connection with at least one consumption equipment, and the consumption data-based wind control method comprises the following steps:
receiving risk subdata of a user to be predicted, which is transmitted by each consumption device, and aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and determining a wind control strategy of the user to be predicted according to the target risk data, and performing wind control on the user to be predicted according to the wind control strategy.
2. The consumption data-based wind control method according to claim 1, wherein after the step of aggregating each of the risk sub-data to generate target risk data, the method comprises:
determining a model type corresponding to each wind control sub-module, and if the model type is the wind control type, executing a step of determining a wind control strategy of the user to be predicted according to the target risk data;
and if the model type is an abnormality detection type, performing abnormality detection on the user to be predicted according to the abnormality score corresponding to the target risk data.
3. The consumption data-based wind control method according to claim 1, wherein before the step of receiving the risk subdata of the user to be predicted transmitted by each consumption device, and aggregating each risk subdata to generate the target risk data, the method further comprises:
carrying out sample alignment or encrypted sample alignment with each consumption device, and receiving sample characteristic data determined by the sample alignment in each consumption device to respectively train a preset wind control sub-module in each consumption device to generate an intermediate result;
aggregating the intermediate results to generate gradient information, and transmitting the gradient information back to the consumption equipment so as to perform federal training by combining sample label data determined by sample alignment in the wind control equipment and sample characteristic data of the consumption equipment;
and when detecting that the federal training meets a preset finishing condition, finishing the federal training, and respectively generating preset wind control sub-modules in the consumption equipment into the wind control sub-modules.
4. The consumption data-based wind control method according to claim 3, wherein the preset ending condition includes a preset training time, a preset training duration, or a preset loss function convergence of a federal model, and when detecting that the federal training satisfies the preset ending condition, the step of completing the federal training includes:
when the counted number of times of the federal training reaches the preset number of times, completing the federal training;
or when the statistical result shows that the training time of the federal training reaches the preset training time, completing the federal training;
or when the convergence of the loss function of the preset federated model is detected, completing the federated training.
5. A consumption data-based wind control method is applied to consumption equipment, and is characterized by comprising the following steps:
carrying out sample alignment or encrypted sample alignment with a wind control device, and determining sample characteristic data of the consumption device;
training a preset wind control submodule in the consumption equipment based on the sample characteristic data, generating an intermediate result and transmitting the intermediate result to the wind control equipment;
receiving gradient information generated after the wind control equipment aggregates the intermediate results, and performing federal model training on sample label data determined by sample alignment in the wind control equipment and sample characteristic data in the consumption equipment based on the gradient information;
and when a federal training end instruction is received, stopping the federal training, and generating a preset wind control submodule in the consumption equipment into a wind control submodule in the consumption equipment so as to perform wind control based on the wind control submodule.
6. The consumption data-based wind control method according to claim 5, wherein after the step of generating a preset wind control submodule within the consumption device as the wind control submodule within the consumption device, the method further comprises:
and when the data volume of the updating data in the consumption equipment is detected to be larger than a preset threshold value, performing sample alignment or encrypted sample alignment with the wind control equipment based on the updating data, and determining sample characteristic data of the consumption equipment so as to update the wind control submodule in the consumption equipment based on the updating data.
7. The consumption data-based wind control method according to claim 5, wherein the step of sample aligning or encrypted sample aligning with a wind control device to determine sample characteristic data of the consumption device is preceded by the method comprising:
according to a preset data derivation model, deriving the consumption subdata in the consumption equipment to generate derived data corresponding to the consumption subdata, and determining sample characteristic data of the consumption equipment based on the derived data and the consumption subdata after sample alignment is carried out on the consumption subdata and the wind control equipment.
8. A consumption data based wind control device, comprising:
the aggregation module is used for receiving risk subdata of the user to be predicted transmitted by each consumption device and aggregating the risk subdata to generate target risk data, wherein each risk subdata is generated by performing risk prediction on the consumption subdata of the user to be predicted in each consumption device based on a wind control submodule in each consumption device;
and the wind control module is used for determining the wind control strategy of the user to be predicted according to the target risk data and carrying out wind control on the user to be predicted according to the wind control strategy.
9. A consumption data based wind control system, characterized in that the consumption data based wind control system comprises a wind control device and at least one consumption device communicatively connected to the wind control device, the consumption data based wind control system further comprises a memory, a processor and a consumption data based wind control program stored on the memory and executable on the processor, the consumption data based wind control program when executed by the processor implements the steps of the consumption data based wind control method according to any one of claims 1 to 4 or any one of claims 5 to 7.
10. A readable storage medium, characterized in that the readable storage medium has stored thereon a consumption data based wind control program, which when executed by a processor implements the steps of the consumption data based wind control method according to any of claims 1-4 or any of claims 5-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010632412.2A CN111784472A (en) | 2020-07-03 | 2020-07-03 | Wind control method, device and system based on consumption data and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010632412.2A CN111784472A (en) | 2020-07-03 | 2020-07-03 | Wind control method, device and system based on consumption data and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111784472A true CN111784472A (en) | 2020-10-16 |
Family
ID=72757952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010632412.2A Pending CN111784472A (en) | 2020-07-03 | 2020-07-03 | Wind control method, device and system based on consumption data and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111784472A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112270597A (en) * | 2020-11-10 | 2021-01-26 | 恒安嘉新（北京）科技股份公司 | Business processing and credit evaluation model training method, device, equipment and medium |
| CN113269431A (en) * | 2021-05-20 | 2021-08-17 | 深圳易财信息技术有限公司 | Inventory risk prediction method, apparatus, medium, and computer program product |
| CN113421136A (en) * | 2021-08-25 | 2021-09-21 | 深圳兆瑞优品科技有限公司 | Online shopping wind control method, device and system |
| CN113538127A (en) * | 2021-07-16 | 2021-10-22 | 四川新网银行股份有限公司 | Method, system, equipment and medium for supporting simultaneous joint wind control test of multiple partners |
| CN113569263A (en) * | 2021-07-30 | 2021-10-29 | 拉扎斯网络科技（上海）有限公司 | Secure processing method and device for cross-private-domain data and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110307396A1 (en) * | 2010-06-15 | 2011-12-15 | Masteryconnect Llc | Education Tool for Assessing Students |
| CN111008709A (en) * | 2020-03-10 | 2020-04-14 | 支付宝(杭州)信息技术有限公司 | Federal learning and data risk assessment method, device and system |
-
2020
- 2020-07-03 CN CN202010632412.2A patent/CN111784472A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110307396A1 (en) * | 2010-06-15 | 2011-12-15 | Masteryconnect Llc | Education Tool for Assessing Students |
| CN111008709A (en) * | 2020-03-10 | 2020-04-14 | 支付宝(杭州)信息技术有限公司 | Federal learning and data risk assessment method, device and system |
Non-Patent Citations (1)
| Title |
|---|
| 杨强: "ＡI 与数据隐私保护：联邦学习的破解之道", 《信息安全研究》, vol. 5, no. 11, pages 961 - 965 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112270597A (en) * | 2020-11-10 | 2021-01-26 | 恒安嘉新（北京）科技股份公司 | Business processing and credit evaluation model training method, device, equipment and medium |
| CN113269431A (en) * | 2021-05-20 | 2021-08-17 | 深圳易财信息技术有限公司 | Inventory risk prediction method, apparatus, medium, and computer program product |
| CN113269431B (en) * | 2021-05-20 | 2023-12-05 | 深圳易财信息技术有限公司 | Inventory risk prediction method, apparatus, medium and computer program product |
| CN113538127A (en) * | 2021-07-16 | 2021-10-22 | 四川新网银行股份有限公司 | Method, system, equipment and medium for supporting simultaneous joint wind control test of multiple partners |
| CN113538127B (en) * | 2021-07-16 | 2023-06-23 | 四川新网银行股份有限公司 | Method, system, equipment and medium for supporting simultaneous combined wind control test of multiple partners |
| CN113569263A (en) * | 2021-07-30 | 2021-10-29 | 拉扎斯网络科技（上海）有限公司 | Secure processing method and device for cross-private-domain data and electronic equipment |
| CN113421136A (en) * | 2021-08-25 | 2021-09-21 | 深圳兆瑞优品科技有限公司 | Online shopping wind control method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111784472A (en) | Wind control method, device and system based on consumption data and readable storage medium | |
| WO2020238415A1 (en) | Method and apparatus for monitoring model training | |
| CN109760041B (en) | Chat robot-based cloud management system and operation method thereof | |
| CN110147925B (en) | Risk decision method, device, equipment and system | |
| US11558420B2 (en) | Detection of malicious activity within a network | |
| US20210049259A1 (en) | Threshold determining and identity verification method, apparatus, electronic device, and storage medium | |
| CN104541293A (en) | Architecture for client-cloud behavior analyzer | |
| US20200285994A1 (en) | Determination system, determination method and program | |
| CN109376050A (en) | A kind of APP monitoring method, computer readable storage medium and terminal device | |
| US20200272849A1 (en) | Estimating system, estimating method and program | |
| CN110991871A (en) | Risk monitoring method, device, equipment and computer readable storage medium | |
| CN106878236A (en) | A kind of user's request processing method and equipment | |
| CN110807209B (en) | Data processing method, device and storage medium | |
| US11763189B2 (en) | Method for tracking lack of bias of deep learning AI systems | |
| US20170213283A1 (en) | Device and a computer software for provisioning a user with requested services | |
| CN112968796A (en) | Network security situation awareness method and device and computer equipment | |
| CN111539020A (en) | Material purchasing management system and method | |
| US20150295952A1 (en) | Service Provisioning with Improved Authentication Processing | |
| KR20100127624A (en) | Method and apparatus for management information | |
| CN111489175B (en) | Online identity authentication method, device, system and storage medium | |
| CN112184429A (en) | User information processing method and block chain link point | |
| US9183595B1 (en) | Using link strength in knowledge-based authentication | |
| CN112069222A (en) | Enterprise policy query system based on big data | |
| CN111311102A (en) | Resource ratio adjusting method, device, equipment and computer readable storage medium | |
| US20240086923A1 (en) | Entity profile for access control |
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 |