CN116416069A - Potential user screening method and device - Google Patents

Potential user screening method and device Download PDF

Info

Publication number
CN116416069A
CN116416069A CN202211650073.6A CN202211650073A CN116416069A CN 116416069 A CN116416069 A CN 116416069A CN 202211650073 A CN202211650073 A CN 202211650073A CN 116416069 A CN116416069 A CN 116416069A
Authority
CN
China
Prior art keywords
user
sample information
information
characteristic
target
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
Application number
CN202211650073.6A
Other languages
Chinese (zh)
Inventor
曹圳杰
常鹏
朱益兴
李飞
林星凯
朱恩东
王步青
赖众程
黎利
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Bank Co Ltd
Original Assignee
Ping An Bank Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Bank Co Ltd filed Critical Ping An Bank Co Ltd
Priority to CN202211650073.6A priority Critical patent/CN116416069A/en
Publication of CN116416069A publication Critical patent/CN116416069A/en
Pending legal-status Critical Current

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
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/06—Asset management; Financial planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

The application relates to the technical field of data mining, and provides a potential user screening method and device. The method comprises the following steps: acquiring user characteristic information of a target user; inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user; according to the user grade, a screening result for judging whether the target user is a potential user is obtained; the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information. The potential user screening method provided by the embodiment of the invention can effectively find whether the user is a possible potential client.

Description

Potential user screening method and device
Technical Field
The application relates to the technical field of data mining, in particular to a potential user screening method and device.
Background
In traditional financial services, the customer's asset information is highly sensitive and is in principle invisible to business personnel during various links of the financial transaction, such as selling financial products. Therefore, how to screen out potential financial customers is a technical problem that needs to be solved in the present time under the condition that customer asset information cannot be known.
Disclosure of Invention
The present application aims to solve at least one of the technical problems existing in the related art. Therefore, the potential user screening method is provided, and whether the user is a possible potential client or not can be effectively found.
The application also provides a potential user screening device.
The application also provides electronic equipment.
The present application also proposes a computer-readable storage medium.
The potential user screening method according to the embodiment of the first aspect of the application comprises the following steps:
acquiring user characteristic information of a target user;
inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user;
according to the user grade, a screening result for judging whether the target user is a potential user is obtained;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
According to the potential user screening method, the user characteristic information of the target user is input into the deep learning neural network trained by the characteristic sample information consistent with the data type of the user characteristic information, so that the user grade corresponding to the target user is obtained, whether the target user is the potential user is judged according to the user grade, whether the target user is the potential user can be effectively screened based on the user characteristic information of the target user, and further screening of the potential user can be achieved.
According to one embodiment of the present application, further comprising:
acquiring target sample information corresponding to the same preset user grade from the characteristic sample information;
and sequentially inputting the target sample information into the deep learning neural network for training, and adjusting a loss function of the deep learning neural network after each training until the user grade output after each target sample information is input into the deep learning neural network is the preset user grade.
According to one embodiment of the present application, adjusting the loss function of the deep learning neural network includes:
acquiring current target sample information input into the deep learning neural network, and acquiring first cosine similarity of the current target sample information and each piece of residual sample information in each piece of characteristic sample information;
according to the first cosine similarity and the second cosine similarity between the current target sample information and the negative sample information, adjusting the loss function;
the residual sample information is characteristic sample information except the current target sample information in the characteristic sample information;
the negative sample information is characteristic sample information of different preset user grades corresponding to the current target sample information in the characteristic sample information.
According to one embodiment of the present application, adjusting the loss function further according to each of the first cosine similarities and each of the second cosine similarities between the current target sample information and each of the negative sample information includes:
inputting each first cosine similarity and each second cosine similarity into a loss function operation model
Figure BDA0004010071290000031
Adjusting the loss function L;
wherein i represents the current target sample information, j represents the remaining sample information, s im (z i ，z j ) Represents the first cosine similarity, τ is a temperature coefficient, x represents the negative sample information, s im (z i ，z x ) And representing the second cosine similarity between the current target sample information and the negative sample information, wherein N represents the number of preset user levels, k represents the number of samples of each preset user level, and N represents the number of the first cosine similarities.
According to one embodiment of the present application, according to the user class, obtaining a screening result for determining whether the target user is a potential user includes:
and determining that the user grade is not lower than a preset target user grade, and determining that the target user is a potential user.
According to one embodiment of the application, the user characteristic information includes user portrait information.
According to one embodiment of the present application, the user characteristic information includes environment information in which the user portrait information is located.
According to an embodiment of the second aspect of the present application, a potential user screening apparatus includes:
the feature information acquisition module is used for acquiring user feature information of a target user;
the user grade acquisition module is used for inputting the user characteristic information into a trained deep learning neural network to acquire a user grade corresponding to the target user;
the potential user screening module is used for acquiring and judging whether the target user is a screening result of the potential user according to the user grade;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
An electronic device according to an embodiment of a third aspect of the present application includes a processor and a memory storing a computer program, the processor implementing the potential user screening method according to any of the above embodiments when executing the computer program.
A computer readable storage medium according to an embodiment of a fourth aspect of the present application, having stored thereon a computer program which, when executed by a processor, implements the potential user screening method according to any of the embodiments described above.
A computer program product according to an embodiment of the fifth aspect of the present application, comprising: the computer program, when executed by a processor, implements a potential user screening method as described in any of the embodiments above.
The above technical solutions in the embodiments of the present application have at least one of the following technical effects:
the user characteristic information of the target user is input into the deep learning neural network trained by the characteristic sample information consistent with the data type of the user characteristic information to obtain the user grade corresponding to the target user, so that whether the target user is a potential user or not is judged according to the user grade, whether the target user is the potential user or not can be effectively screened based on the user characteristic information of the target user, and further screening of the potential user can be achieved.
Furthermore, the target sample information of the same preset user grade is sequentially input into the deep learning neural network for training, so that the deep learning neural network can better identify the characteristic information corresponding to the preset user grade, and the accuracy of the subsequent user grade classification of the user characteristic information is improved.
Drawings
For a clearer description of the present application or of the prior art, the drawings that are used in the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description below are some embodiments of the present application, and that other drawings may be obtained from these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of a potential user screening method provided in an embodiment of the present application;
FIG. 2 is a flow chart illustrating further refinement of training of the deep learning neural network in the potential user screening method of FIG. 1 in an embodiment of the present application;
FIG. 3 is a flow chart illustrating further refinement of the adjustment of the loss function network in the potential user screening method of FIG. 2 in an embodiment of the present application;
fig. 4 is a schematic structural diagram of a potential user screening apparatus according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
The method and apparatus for screening potential users provided in the embodiments of the present application will be described and illustrated in detail below by using several specific embodiments.
In traditional financial services, the customer's asset information is highly sensitive and is in principle invisible to business personnel during various links of the financial transaction, such as selling financial products. On the other hand, business personnel may wish to gain more information about the customer's assets to mine potential customers. Thus, in order to be able to discover potential customers who may purchase financial products in a timely manner among a large number of customers transacting financial services, in one embodiment, a potential user screening method is provided, which is applied to a terminal device, for screening potential customers who purchase financial products. The terminal device may be a user terminal or a server, the user terminal may be a desktop terminal or a portable terminal, such as a desktop computer, a notebook computer, etc., and the server may be an independent server or a server cluster formed by a plurality of servers, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, and basic cloud computing services such as big data and artificial intelligent sampling point devices.
As shown in fig. 1, the method for screening potential users provided in this embodiment includes:
step 101, obtaining user characteristic information of a target user;
102, inputting the user characteristic information into a trained deep learning neural network to obtain a user grade corresponding to the target user;
step 103, obtaining a screening result for judging whether the target user is a potential user according to the user grade;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
The user characteristic information of the target user is input into the deep learning neural network trained by the characteristic sample information consistent with the data type of the user characteristic information to obtain the user grade corresponding to the target user, so that whether the target user is a potential user or not is judged according to the user grade, whether the target user is the potential user or not can be effectively screened based on the user characteristic information of the target user, and further screening of the potential user can be achieved.
In one embodiment, the user characteristic information is non-private information of the target user. In order to make the final filtering more accurate, the user characteristic information may be user image information of the target user, such as face information of the user. The user portrait information of the user can be extracted from the image information obtained by the camera during the authentication process of the user for financial service.
Because the living environment of the user can reflect the user habit of the user to a certain extent, the user characteristic information comprises the environment information of the user portrait information besides the user portrait information. The image information of the user, such as face information or human body information, can be extracted from the image information obtained by the camera. And extracting background information in the image information as environment information in which the user portrait information is located.
After the user characteristic information of the target user is obtained, the user characteristic information can be input into the trained deep learning neural network. The deep learning neural network may be at least one of a convolutional neural network CNN, a recurrent neural network RNN, and a neural network generating an countermeasure network (GAN).
The training of the deep learning neural network is achieved by acquiring each piece of characteristic sample information which is assigned with a preset user grade, and then sequentially inputting each piece of characteristic sample information into the deep learning neural network for training.
The training of the deep learning neural network may be, for example, inputting the feature sample information into the deep learning neural network, obtaining a user level output by the deep learning neural network, and then matching the user level with a preset user level specified by the feature sample information to determine whether the user level and the preset user level are the same. If the user classes are different, the loss function of the deep learning neural network is adjusted, then the next characteristic sample information is continuously input for training until each characteristic sample information is input into the deep learning neural network, and the output user class is the same as the preset user class designated by the characteristic sample information.
For example, assuming that each feature sample information is (a 1, a2, … … an), a1 is preset to have a preset user rank p5, a2 is preset to have a preset user rank p2 … … an is preset to have a preset user rank p3, a1 is input into the deep learning neural network first, and an output user rank is obtained. If the user grade is p3, the user grade p5 corresponding to a1 is different, at the moment, the loss function of the deep learning neural network is adjusted, and then the next characteristic sample information a2 is input; if the output user level is p5, the user level p5 corresponding to a1 is the same, and the next feature sample information a2 is directly input.
The optimization of the loss function can be realized by optimizing gradient descent based on an adam algorithm, and cross entropy is used as an optimization function of the loss function in the training process.
When training the deep learning neural network, the training method can be used according to 8:1:1, randomly dividing each characteristic sample information into a training set, a verification set and a test set. The feature sample information in the training set is used for training the deep learning neural network, the feature sample information in the verification set is used for verifying whether the deep learning neural network is completely trained, and the feature sample information in the test set is used for testing the training effect of the deep learning neural network.
In order to make the user grade classification result of the trained deep learning neural network more accurate, the characteristic sample information also comprises user portrait information and environment information where the user portrait information is located.
Because the user portrait information is sensitive information, after the feature sample information is input into the deep learning neural network for training, the feature sample information is deleted, and the feature sample information for training is not stored locally.
To improve the user-level classification effect of the deep learning neural network, in one embodiment, as shown in fig. 2, training the deep learning neural network includes:
step 201, obtaining each target sample information corresponding to the same preset user level from each characteristic sample information;
step 202, sequentially inputting the target sample information into the deep learning neural network for training, and adjusting a loss function of the deep learning neural network after each training until the user grade output after each target sample information is input into the deep learning neural network is the preset user grade.
In an embodiment, since each piece of feature sample information has a corresponding preset user level, the feature sample information can be classified first, that is, the feature sample information corresponding to the same preset user level is used as the same class, and at this time, the feature sample information belonging to the same class is the target sample information.
After the target sample information of the same class is obtained, the target sample information is sequentially input into the deep learning neural network, and the user grade output by the deep learning neural network is obtained. And then matching the user grade with a preset user grade corresponding to the category of target sample information, and judging whether the user grade and the preset user grade are the same. If the two types of target sample information are different, the loss function of the deep learning neural network is adjusted, then the next target sample information is continuously input for training until the user grade output after each target sample information is input into the deep learning neural network is the same as the preset user grade corresponding to the target sample information.
For example, assuming that each target sample information is (b 1, b2, … … bn), and the preset user level corresponding to the target sample information is p5, b1 is input into the deep learning neural network first, and the output user level is obtained. If the user grade is p3, and the preset user grade p5 corresponding to the target sample information is different, at the moment, the loss function of the deep learning neural network is adjusted, and then the next characteristic sample information b2 is input; if the output user level is p5, the user level is the same as the preset user level p5 corresponding to the target sample information, and the next characteristic sample information b2 is directly input.
In this way, the target sample information of the same preset user grade is sequentially input into the deep learning neural network for training, so that the deep learning neural network can better identify the characteristic information corresponding to the preset user grade, and the accuracy of the subsequent user grade classification of the user characteristic information is improved.
To further improve the accuracy of the deep learning neural network in classifying the user class, in an embodiment, as shown in fig. 3, adjusting the loss function of the deep learning neural network includes:
step 301, obtaining the first cosine similarity between the current target sample information input into the deep learning neural network and each piece of residual sample information in each piece of characteristic sample information;
step 302, adjusting the loss function according to each first cosine similarity and each second cosine similarity between the current target sample information and each negative sample information;
the residual sample information is characteristic sample information except the current target sample information in the characteristic sample information;
the negative sample information is characteristic sample information of different preset user grades corresponding to the current target sample information in the characteristic sample information.
In an embodiment, assuming that each characteristic sample information includes N sample information of preset user levels, the number of each target sample information is k, and in each target sample information, the current target sample information input to the deep learning neural network is i, dividing the current target sample information i in the target sample information to be used as each remaining sample information j, and then forming positive sample pairs by the current target sample information i and each remaining sample information j respectively, so as to obtain k-1 positive sample pairs (i, j). After k-1 positive sample pairs (i, j) are obtained, calculating the first cosine similarity of each positive sample pair (i, j), thereby obtaining each first cosine similarity corresponding to each positive sample pair (i, j) one by one. And simultaneously, the current target sample information is i, and negative sample information x with different preset user grades corresponding to the current target sample information is i in each piece of characteristic sample information, namely N-k negative sample information x except k pieces of target sample information in each piece of characteristic sample information respectively form negative sample pairs, so that (N-1) k negative sample pairs (i, x) can be obtained. After obtaining (N-1) k negative sample pairs (i, x), calculating the second cosine similarity of each negative sample pair (i, x), thereby obtaining each second cosine similarity corresponding to each negative sample pair (i, x) one by one.
After each first cosine similarity and each second cosine similarity are obtained, the loss function can be adjusted according to each first cosine similarity and each second cosine similarity.
Wherein, according to each first cosine similarity and each second cosine similarity, adjust the loss function, include:
inputting each first cosine similarity and each second cosine similarity into a loss function operation model
Figure BDA0004010071290000111
Adjusting the loss function L;
wherein i represents the current target sample information, j represents the remaining sample information, s im (z i ，z j ) Represents the first cosine similarity, τ is a temperature coefficient, x represents the negative sample information, s im (z i ，z x ) And representing the second cosine similarity between the current target sample information and the negative sample information, wherein N represents the number of preset user levels, k represents the number of samples of each preset user level, and N represents the number of the first cosine similarities.
In one embodiment, in the loss function model, the temperature coefficient is used to adjust the degree of interest in difficult samples, and the smaller the temperature coefficient, the more interest in separating this sample from the most similar other samples. l (L) i The numerator of (a) is the similarity of a positive sample pair i and j, and the denominator is the sum of the similarity of the positive sample pair and the (N-1) k negative sample pairs containing i. After obtaining l i Then, the calculated l is calculated for n positive sample pairs i Taking the negative logarithm and then averaging to obtain the final loss function L.
After training the deep learning neural network, user characteristic information of a target user to be identified can be input into the trained deep learning neural network to obtain a user grade corresponding to the target user.
After the user grade corresponding to the target user is obtained, whether the user grade is not lower than a pre-designated target grade is detected. The target level may be set according to practical situations, for example, the user level may generally include five levels P1-P5, P1-P3 may represent low net value clients, and P4-P5 may represent high net value clients, so the target level may be set to P4. If the user grade is not lower than the pre-designated target grade, the user can be judged to belong to a high net value client, and the target user can be determined to be a potential user at the moment; otherwise, it may be determined that the user belongs to a low net value client, at which point the target user may be determined to be a non-potential user.
The potential user screening device provided by the application is described below, and the potential user screening device described below and the potential user screening method described above can be referred to correspondingly.
In one embodiment, as shown in fig. 4, there is provided a potential user screening apparatus comprising:
a feature information obtaining module 210, configured to obtain user feature information of a target user;
the user grade obtaining module 220 is configured to input the user characteristic information into a trained deep learning neural network, and obtain a user grade corresponding to the target user;
a potential user screening module 230, configured to obtain a screening result for determining whether the target user is a potential user according to the user class;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
The user characteristic information of the target user is input into the deep learning neural network trained by the characteristic sample information consistent with the data type of the user characteristic information to obtain the user grade corresponding to the target user, so that whether the target user is a potential user or not is judged according to the user grade, whether the target user is the potential user or not can be effectively screened based on the user characteristic information of the target user, and further screening of the potential user can be achieved.
In an embodiment, the user level acquisition module 220 is further configured to:
acquiring target sample information corresponding to the same preset user grade from the characteristic sample information;
and sequentially inputting the target sample information into the deep learning neural network for training, and adjusting a loss function of the deep learning neural network after each training until the user grade output after each target sample information is input into the deep learning neural network is the preset user grade.
In one embodiment, the user level acquisition module 220 is specifically configured to:
acquiring current target sample information input into the deep learning neural network, and acquiring first cosine similarity of the current target sample information and each piece of residual sample information in each piece of characteristic sample information;
according to the first cosine similarity and the second cosine similarity between the current target sample information and the negative sample information, adjusting the loss function;
the residual sample information is characteristic sample information except the current target sample information in the characteristic sample information;
the negative sample information is characteristic sample information of different preset user grades corresponding to the current target sample information in the characteristic sample information.
In one embodiment, the user level acquisition module 220 is specifically configured to:
inputting each first cosine similarity and each second cosine similarity into a loss function operation model
Figure BDA0004010071290000141
Adjusting the loss function L;
wherein i represents the current target sample information, j represents the remaining sample information, s im (z i ，z j ) Represents the first cosine similarity, τ is a temperature coefficient, x represents the negative sample information, s im (z i ，z x ) And representing the second cosine similarity between the current target sample information and the negative sample information, wherein N represents the number of preset user levels, k represents the number of samples of each preset user level, and N represents the number of the first cosine similarities.
In one embodiment, the potential user screening module 230 is specifically configured to:
and determining that the user grade is not lower than a preset target user grade, and determining that the target user is a potential user.
In one embodiment, the user characteristic information includes user portrait information.
In an embodiment, the user characteristic information includes environment information in which the user portrait information is located.
Fig. 5 illustrates a physical schematic diagram of an electronic device, as shown in fig. 5, which may include: processor 810, communication interface (Commun i cat i on I nterface) 820, memory 830, and communication bus 840, wherein processor 810, communication interface 820, memory 830 accomplish communication with each other through communication bus 840. Processor 810 may invoke a computer program in memory 830 to perform potential user screening methods including, for example:
acquiring user characteristic information of a target user;
inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user;
according to the user grade, a screening result for judging whether the target user is a potential user is obtained;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-On-i-memory (ROM), a random-access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, embodiments of the present application further provide a storage medium, where the storage medium includes a computer program, where the computer program may be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer is capable of executing the potential user screening method provided in the foregoing embodiments, for example, including:
acquiring user characteristic information of a target user;
inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user;
according to the user grade, a screening result for judging whether the target user is a potential user is obtained;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
In another aspect, embodiments of the present application further provide a processor-readable storage medium storing a computer program, where the computer program is configured to cause a processor to perform a method provided in the foregoing embodiments, for example, including:
acquiring user characteristic information of a target user;
inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user;
according to the user grade, a screening result for judging whether the target user is a potential user is obtained;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
The processor-readable storage medium may be any available medium or data storage device that can be accessed by a processor including, but not limited to, magnetic memory (e.g., floppy disk, hard disk, tape, magneto-optical disk (MO), etc.), optical memory (e.g., CD, DVD, BD, HVD, etc.), and semiconductor memory (e.g., ROM, EPROM, EEPROM, nonvolatile memory (NAND FLASH), solid State Disk (SSD)), etc.
The apparatus embodiments described above are merely illustrative, wherein elements illustrated as separate elements may or may not be physically separate, and elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims (10)

1. A method of screening potential users, comprising:
acquiring user characteristic information of a target user;
inputting the user characteristic information into a trained deep learning neural network, and obtaining a user grade corresponding to the target user;
according to the user grade, a screening result for judging whether the target user is a potential user is obtained;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
2. The potential user screening method of claim 1, further comprising:
acquiring target sample information corresponding to the same preset user grade from the characteristic sample information;
and sequentially inputting the target sample information into the deep learning neural network for training, and adjusting a loss function of the deep learning neural network after each training until the user grade output after each target sample information is input into the deep learning neural network is the preset user grade.
3. The potential user screening method of claim 2, wherein adjusting the loss function of the deep learning neural network comprises:
acquiring current target sample information input into the deep learning neural network, and acquiring first cosine similarity of the current target sample information and each piece of residual sample information in each piece of characteristic sample information;
according to the first cosine similarity and the second cosine similarity between the current target sample information and the negative sample information, adjusting the loss function;
the residual sample information is characteristic sample information except the current target sample information in the characteristic sample information;
the negative sample information is characteristic sample information of different preset user grades corresponding to the current target sample information in the characteristic sample information.
4. The method of claim 3, wherein adjusting the loss function further based on each of the first cosine similarities and each of the second cosine similarities between the current target sample information and each of the negative sample information comprises:
inputting each first cosine similarity and each second cosine similarity into a loss function operation model
Figure FDA0004010071280000021
Adjusting the loss function L;
wherein i represents the current target sample information, j represents the remaining sample information, sim (z i ，z j ) Represents the first cosine similarity, τ is a temperature coefficient, x represents the negative sample information, sim (z) i ，z x ) And representing the second cosine similarity between the current target sample information and the negative sample information, wherein N represents the number of preset user levels, k represents the number of samples of each preset user level, and N represents the number of the first cosine similarities.
5. The method for screening potential users according to claim 1, wherein obtaining a screening result for determining whether the target user is a potential user according to the user class comprises:
and determining that the user grade is not lower than a preset target user grade, and determining that the target user is a potential user.
6. The potential user screening method of any one of claims 1-5, wherein the user characteristic information comprises user portrait information.
7. The method of claim 5, wherein the user characteristic information comprises environmental information in which the user profile information is located.
8. A potential user screening apparatus, comprising:
the feature information acquisition module is used for acquiring user feature information of a target user;
the user grade acquisition module is used for inputting the user characteristic information into a trained deep learning neural network to acquire a user grade corresponding to the target user;
the potential user screening module is used for acquiring and judging whether the target user is a screening result of the potential user according to the user grade;
the deep learning neural network is trained by each piece of characteristic sample information, and the characteristic sample information is consistent with the data type of the user characteristic information.
9. An electronic device comprising a processor and a memory storing a computer program, wherein the processor implements the potential user screening method of any one of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the potential user screening method of any one of claims 1 to 7.
CN202211650073.6A 2022-12-21 2022-12-21 Potential user screening method and device Pending CN116416069A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211650073.6A CN116416069A (en) 2022-12-21 2022-12-21 Potential user screening method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211650073.6A CN116416069A (en) 2022-12-21 2022-12-21 Potential user screening method and device

Publications (1)

Publication Number Publication Date
CN116416069A true CN116416069A (en) 2023-07-11

Family

ID=87058717

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211650073.6A Pending CN116416069A (en) 2022-12-21 2022-12-21 Potential user screening method and device

Country Status (1)

Country Link
CN (1) CN116416069A (en)

Similar Documents

Publication Publication Date Title
US11475143B2 (en) Sensitive data classification
CN112417439B (en) Account detection method, device, server and storage medium
US11403643B2 (en) Utilizing a time-dependent graph convolutional neural network for fraudulent transaction identification
CN111681091B (en) Financial risk prediction method and device based on time domain information and storage medium
US20170018030A1 (en) System and Method for Determining Credit Worthiness of a User
CN109840413B (en) Phishing website detection method and device
CN110084609B (en) Transaction fraud behavior deep detection method based on characterization learning
CN110162958B (en) Method, apparatus and recording medium for calculating comprehensive credit score of device
CN113221104A (en) User abnormal behavior detection method and user behavior reconstruction model training method
CN113052577A (en) Method and system for estimating category of virtual address of block chain digital currency
CN111967503A (en) Method for constructing multi-type abnormal webpage classification model and abnormal webpage detection method
CN112990989B (en) Value prediction model input data generation method, device, equipment and medium
CN114202336A (en) Risk behavior monitoring method and system in financial scene
CN111245815B (en) Data processing method and device, storage medium and electronic equipment
CN112487284A (en) Bank customer portrait generation method, equipment, storage medium and device
CN113177603B (en) Training method of classification model, video classification method and related equipment
CN116416069A (en) Potential user screening method and device
CN110570301B (en) Risk identification method, device, equipment and medium
CN110472680B (en) Object classification method, device and computer-readable storage medium
CN112529319A (en) Grading method and device based on multi-dimensional features, computer equipment and storage medium
CN114170000A (en) Credit card user risk category identification method, device, computer equipment and medium
CN113706258A (en) Product recommendation method, device, equipment and storage medium based on combined model
CN113239075A (en) Construction data self-checking method and system
CN113706279A (en) Fraud analysis method and device, electronic equipment and storage medium
CN117786121B (en) File identification method and system based on artificial intelligence

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