CN117437010A

CN117437010A - Resource borrowing level prediction method, device, equipment, storage medium and program product

Info

Publication number: CN117437010A
Application number: CN202311255090.4A
Authority: CN
Inventors: 卢翔
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2023-09-26
Filing date: 2023-09-26
Publication date: 2024-01-23

Abstract

The application relates to a resource borrowing level prediction method, a resource borrowing level prediction device, resource borrowing level prediction equipment, a storage medium and a program product. The method comprises the following steps: the method comprises the steps of extracting characteristics of user data by using a first convolution layer in a level prediction model to obtain a first convolution result, carrying out residual processing on the first convolution result by using a first residual layer in the level prediction model, determining a target residual result, determining the creditworthiness level and the net value level of a user according to the target residual result and the first intensive convolution layer in the level prediction model, and obtaining a resource borrowing level of the user based on a quantized value corresponding to the creditworthiness level of the user and a quantized value corresponding to the net value level of the user, so that the accuracy of predicting the resource borrowing level can be improved.

Description

Resource borrowing level prediction method, device, equipment, storage medium and program product

Technical Field

The present disclosure relates to the field of artificial intelligence, and in particular, to a resource borrowing level prediction method, apparatus, device, storage medium, and program product.

Background

When a bank performs resource borrowing service, the sustainable resource borrowing grade of a customer needs to be considered, different customers generally correspond to different resource borrowing grades, and the bank executes the resource borrowing service aiming at the different resource borrowing grades.

In the prior art, the resource borrowing level of a client is usually predicted manually according to client information and the historical resource borrowing behavior of the client.

However, the above-described resource borrowing level prediction method is low in accuracy.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a resource borrowing level prediction method, apparatus, device, storage medium, and program product that can improve the accuracy of resource borrowing level prediction.

In a first aspect, the present application provides a resource borrowing level prediction method. The method comprises the following steps:

extracting features of the user data by using a first convolution layer in the level prediction model to obtain a first convolution result;

performing residual processing on the first convolution result by using a first residual layer in the level prediction model to determine a target residual result;

determining the credibility level and the net value level of the user according to the target residual error result and the first dense convolution layer in the level prediction model;

and obtaining the resource borrowing grade of the user based on the quantized value corresponding to the credit grade of the user and the quantized value corresponding to the net value grade of the user.

In one embodiment, the performing residual processing on the first convolution result by using the first residual layer in the level prediction model to determine a target residual result includes:

Carrying out residual processing on the first convolution result by utilizing a first residual layer in the level prediction model to obtain a first residual result;

inputting the first residual result to a second residual layer in the level prediction model to obtain a second residual result;

and compressing the second residual result by using a first full-connection layer in the level prediction model to obtain a target residual result.

In one embodiment, the first residual layer includes a second convolution layer and a third convolution layer, and the performing residual processing on the first convolution result by using the first residual layer in the level prediction model to obtain a first residual result includes:

performing feature mapping on the first convolution result to obtain a first mapping result;

obtaining a second convolution result according to the first convolution result, the second convolution layer and the third convolution layer;

and summing the first mapping result and the second convolution result to obtain the first residual result.

In one embodiment, the determining the user's reputation level and net level based on the target residual result and the first dense convolutional layer in the level prediction model comprises:

performing feature extraction on the first convolution result by using a first dense convolution layer in the level prediction model to determine a target dense convolution result;

And determining the credibility level and the net value level of the user according to the target dense convolution result and the target residual error result.

In one embodiment, the feature extraction of the first convolution result with the first dense convolution layer in the level prediction model to determine a target dense convolution result includes:

extracting features of the first convolution result by using a first dense convolution layer in the level prediction model to obtain a first dense convolution result;

inputting the first dense convolution result to a first pooling layer in the level prediction model to obtain a first pooling result;

inputting the first pooling result to a second dense convolution layer in the level prediction model to obtain a second dense convolution result;

inputting the second dense convolution result to a second pooling layer in the level prediction model to obtain a second pooling result;

and compressing the second pooling result by using a second full-connection layer in the level prediction model to obtain a target dense convolution result.

In one embodiment, the first dense convolution layer includes a fourth convolution layer, a fifth convolution layer, and a sixth convolution layer, and the feature extracting the first convolution result by using the first dense convolution layer in the level prediction model to obtain a first dense convolution result includes:

Inputting the first convolution result into the fourth convolution layer to obtain a third convolution result;

inputting the first convolution result and the third convolution result into the fifth convolution layer to obtain a fourth convolution result;

and inputting the first convolution result, the third convolution result and the fourth convolution result into the sixth convolution layer to obtain the first dense convolution result.

In one embodiment, the determining the reputation level and the net value level of the user based on the target dense convolution result and the target residual result comprises:

determining a first average between the first probability value and the second probability value; the first probability value is the probability value of each first credibility level in the target dense convolution result, and the second probability value is the probability value of a second credibility level corresponding to the first credibility level in the target residual error result;

taking the reputation level corresponding to the maximum first mean value as the reputation level of the user;

determining a second average between the third probability value and the fourth probability value; the third probability value is the probability value of each first net value grade in the target dense convolution result, and the fourth probability value is the probability value of a second net value grade corresponding to the first net value grade in the target residual error result;

And taking the net value grade corresponding to the maximum second mean value as the net value grade of the user.

In one embodiment, the obtaining the resource borrowing level of the user based on the quantized value corresponding to the reputation level of the user and the quantized value corresponding to the net value level of the user includes:

determining a first product result of a quantized value corresponding to the user's reputation level and a weight of the reputation level;

determining a second product of a quantized value corresponding to the net worth grade of the user and a weight corresponding to the net worth grade;

determining a summation result of the first product result and the second product result;

searching a mapping relation table according to the summation result to obtain the resource borrowing grade of the user; the mapping relation table comprises mapping relations between different summation results and different resource borrowing levels of users.

In a second aspect, the present application further provides a resource borrowing level prediction apparatus. The device comprises:

the extraction module is used for extracting characteristics of the user data by utilizing a first convolution layer in the grade prediction model to obtain a first convolution result;

the processing module is used for carrying out residual processing on the first convolution result by utilizing a first residual layer in the level prediction model and determining a target residual result;

The first determining module is used for determining the credibility level and the net value level of the user according to the target residual error result and the first dense convolution layer in the level prediction model;

and the second determining module is used for obtaining the resource borrowing grade of the user based on the quantized value corresponding to the credit grade of the user and the quantized value corresponding to the net value grade of the user.

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the steps of the above method when the processor executes the computer program.

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of the above method.

In a fifth aspect, the present application also provides a computer program product. The computer program product comprising a computer program which, when executed by a processor, implements the steps of the above method.

The resource borrowing grade prediction method, the device, the equipment, the storage medium and the program product are characterized in that the first convolution layer in the grade prediction model is utilized to extract the characteristics of the user data to obtain a first convolution result, the first residual layer in the grade prediction model is utilized to carry out residual processing on the first convolution result, the target residual result is determined, the credibility grade and the net value grade of the user are determined according to the target residual result and the first intensive convolution layer in the grade prediction model, and the resource borrowing grade of the user is obtained based on the quantized value corresponding to the credibility grade of the user and the quantized value corresponding to the net value grade of the user. In the conventional technology, the resource borrowing level of the client is usually predicted manually according to the client information and the historical resource borrowing behavior of the client, but the prediction accuracy is low. In the embodiment of the application, the credibility level and the net value level are obtained by using the pre-designed level prediction model, and then the resource borrowing level of the user is obtained according to the quantized value corresponding to the credibility level of the user and the quantized value corresponding to the net value level of the user, so that the accuracy of predicting the resource borrowing level can be improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the related art, the drawings that are required to be used in the embodiments or the related technical descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to the drawings without inventive effort for a person having ordinary skill in the art.

Fig. 1 is an application environment diagram of a resource borrowing level prediction method provided in an embodiment of the present application;

fig. 2 is a flow chart of a resource borrowing level prediction method according to an embodiment of the present application;

fig. 3 is a flow chart of a target residual error result determining method provided in an embodiment of the present application;

fig. 4 is a flowchart of a first residual result determining method provided in an embodiment of the present application;

FIG. 5 is a flowchart of a method for determining a reputation level and a net value level according to an embodiment of the present application;

FIG. 6 is a flowchart of a method for determining a target dense convolution result according to an embodiment of the present application;

fig. 7 is a flowchart of a first dense convolution result determining method according to an embodiment of the present application;

FIG. 8 is a second flow chart of a method for determining a reputation level and a net value level according to an embodiment of the present application;

fig. 9 is a flowchart of a resource borrowing level determining method according to an embodiment of the present application;

fig. 10 is a block diagram of a resource borrowing level prediction apparatus according to an embodiment of the present application;

fig. 11 is an internal structure diagram of a computer device according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

The voice recognition method provided by the embodiment of the application can be applied to an application environment shown in fig. 1, and fig. 1 is an application environment diagram of a resource borrowing level prediction method provided by the embodiment of the application. Wherein the terminal 101 communicates with the server 102 via a network. The data storage system may store data that the server 102 needs to process. The data storage system may be integrated on the server 102. The terminal 101 may send user data to the server 102, where the user data may be data acquired based on the internet of things, and the server 102 may perform feature extraction on the user data by using a first convolution layer in the level prediction model to obtain a first convolution result; residual processing is carried out on the first convolution result by utilizing a first residual layer in the grade prediction model, and a target residual result is determined; determining the credibility level and the net value level of the user according to the target residual error result and the first dense convolution layer in the level prediction model; and obtaining the resource borrowing grade of the user based on the quantized value corresponding to the credit grade of the user and the quantized value corresponding to the net value grade of the user. The terminal 101 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices and portable wearable devices, and the internet of things devices may be smart speakers, smart televisions, smart air conditioners, smart vehicle devices, and the like. The portable wearable device may be a smart watch, smart bracelet, headset, or the like. The server 102 may be implemented as a stand-alone server or as a server cluster of multiple servers.

In one embodiment, fig. 2 is a flowchart of a resource borrowing level prediction method provided in the embodiment of the present application, as shown in fig. 2, and a resource borrowing level prediction method is provided, and the method is applied to the server 102 in fig. 1 for illustration, and includes the following steps:

and S201, extracting characteristics of the user data by using a first convolution layer in the level prediction model to obtain a first convolution result.

The user data may include data acquired through an internet of things technology, for example, industries such as manufacturing factories, logistics, agriculture/pasture industries, photovoltaic industries, health industries, electric power energy sources, motorcades and the like, and the industries have the common characteristic that the user data are suitable for facilitating acquisition and scene monitoring of various business scene data in an industry chain by applying the internet of things technology or laying out an internet of things open cloud platform; the bank integrates and shares channel data, product data, scene data, terminal data and the like by constructing a data infrastructure platform and supporting access to multi-channel, multi-product and multi-mode system architecture, and extracts the data of the Internet of things. The user data may also include information left by the user when transacting banking.

Specifically, the level prediction model may include a first convolution layer, and the feature extraction may be performed on the user data by using the first convolution layer in the level prediction model to obtain a first convolution result. It should be noted that, the user data may be a text vector obtained by word segmentation of text data.

It should be noted that, the collection of the user data may be performed in real time through various channels, and the user data may be stored in a relational database, a NoSQL database, a distributed file system, or the like. The historical user data can be constructed as a comprehensive data set, and the initial level prediction model is trained to obtain the level prediction model. The data are acquired through the Internet of things, so that the data have more dimensions, the prediction accuracy can be improved, and the risk control effect is further improved.

S202, carrying out residual processing on the first convolution result by using a first residual layer in the level prediction model, and determining a target residual result.

Specifically, the level prediction model may further include a first volume of residual error layer, residual error processing may be performed on the first convolution result by using the first residual error layer in the level prediction model, so as to obtain a first residual error result, and the first residual error result is used as a target residual error result; the first residual result may also be processed using a self-attention mechanism algorithm to obtain a result as a target residual result.

S203, determining the creditworthiness level and the net value level of the user according to the target residual error result and the first dense convolution layer in the level prediction model.

Specifically, the level prediction model may further include a first dense convolution layer, the first dense convolution layer in the level prediction model may be used to perform dense feature extraction on the first convolution result to obtain a first dense convolution result, the first dense convolution result is used as a target dense convolution result, and the reputation level and the net value level of the user are determined according to the similarity between the target residual error result and the target dense convolution result.

S204, obtaining the resource borrowing grade of the user based on the quantized value corresponding to the credit grade of the user and the quantized value corresponding to the net grade of the user.

Specifically, a first product result of a quantized value corresponding to the reputation level of the user and a weight of the reputation level may be determined, a second product of a quantized value corresponding to the net level of the user and a weight corresponding to the net level may be determined, a summation result of the first product result and the second product result may be determined, and a resource borrowing level of the user may be determined according to a resource borrowing level corresponding to the summation result.

In the above embodiment, the feature extraction may be performed on the user data by using the first convolution layer in the level prediction model to obtain a first convolution result, the residual processing may be performed on the first convolution result by using the first residual layer in the level prediction model to determine a target residual result, and the reputation level and the net value level of the user may be determined according to the target residual result and the first dense convolution layer in the level prediction model, and the resource borrowing level of the user may be obtained based on the quantized value corresponding to the reputation level of the user and the quantized value corresponding to the net value level of the user. In the conventional technology, the resource borrowing level of the client is usually predicted manually according to the client information and the historical resource borrowing behavior of the client, but the prediction accuracy is low. In the embodiment of the application, the credibility level and the net value level are obtained by using the pre-designed level prediction model, and then the resource borrowing level of the user is obtained according to the quantized value corresponding to the credibility level of the user and the quantized value corresponding to the net value level of the user, so that the accuracy of predicting the resource borrowing level can be improved, and the effective risk control is performed.

In an embodiment, fig. 3 is a flow chart of a target residual result determining method provided in the embodiment of the present application, as shown in fig. 3, where the embodiment relates to how to use a first residual layer in a level prediction model to perform residual processing on a first convolution result, and determine a possible implementation manner of the target residual result, on the basis of the embodiment, the step S202 includes:

s301, carrying out residual processing on the first convolution result by using a first residual layer in the level prediction model to obtain a first residual result.

In an embodiment of the present application, a first residual layer may be included in the level prediction model, and the first residual layer may include a second convolution layer and a third convolution layer.

Optionally, fig. 4 is a flow chart of a first residual result determining method provided in the embodiment of the present application, as shown in fig. 4, where the embodiment relates to how a first residual layer includes a second convolution layer and a third convolution layer, and how to perform residual processing on the first convolution result by using the first residual layer in the level prediction model to obtain a possible implementation manner of the first residual result, where, on the basis of the foregoing embodiment, S301 includes:

S401, performing feature mapping on the first convolution result to obtain a first mapping result.

Specifically, the first convolution result may be subjected to eigen feature mapping, and the obtained mapping result is the same as the first convolution result, and the mapping result is used as the first mapping result. The result obtained when the feature mapping is performed on the first convolution result by using the preset convolution check may also be used as the first mapping result.

S402, obtaining a second convolution result according to the first convolution result, the second convolution layer and the third convolution layer.

Specifically, the second convolution layer in the level prediction model may be used to perform convolution processing on the first convolution result to obtain a first intermediate convolution result, and the third convolution layer in the level prediction model may be used to perform convolution processing on the first intermediate convolution result to obtain a second intermediate convolution result, where the second intermediate convolution result may be used as the second convolution result. And normalizing the second intermediate convolution result by utilizing batch normalization operation to obtain a normalization result, and taking the normalization result as the second convolution result.

S403, summing the first mapping result and the second convolution result to obtain a first residual result.

Specifically, a summation result obtained by summing the first mapping result and the second convolution result may be used as the first residual result. The first mapping result and the second convolution result are added to obtain a summation result, and the summation result is corrected by using a preset correction function to obtain a first residual error result; the preset correction function may be a linear rectification function.

S302, inputting the first residual result into a second residual layer in the grade prediction model to obtain a second residual result.

Specifically, the level prediction model may further include a second residual layer, and the first residual result may be input to the second residual layer in the level prediction model, and after residual processing, the second residual result is obtained. The specific process is similar to the method for performing residual processing on the first convolution result by using the first residual layer in the level prediction model provided in S401-S403 to obtain the first residual result.

And S303, compressing the second residual result by using a first full-connection layer in the grade prediction model to obtain a target residual result.

Specifically, the level prediction model may further include a first full-connection layer, and the first full-connection layer in the level prediction model may be used to compress the second residual result to obtain a compressed result, and the compressed result is used as the target residual result.

In one embodiment, fig. 5 is one of the flow diagrams of a method for determining a reputation level and a net value level according to the embodiment of the present application, as shown in fig. 5, where the embodiment relates to a possible implementation manner of determining a reputation level and a net value level of a user according to a target residual result and a first dense convolution layer in a level prediction model, and on the basis of the embodiment, the step S203 includes:

S501, extracting features of the first convolution result by using a first dense convolution layer in the level prediction model to determine a target dense convolution result.

Optionally, fig. 6 is a flowchart of a method for determining a target dense convolution result according to the embodiment of the present application, as shown in fig. 6, where the embodiment relates to a possible implementation manner of performing feature extraction on a first convolution result by using a first dense convolution layer in a level prediction model to determine the target dense convolution result, and on the basis of the embodiment, S501 includes:

and S601, extracting features of the first convolution result by using a first dense convolution layer in the level prediction model to obtain the first dense convolution result.

Specifically, the level prediction model may include a first dense convolution layer, and the feature extraction may be performed on the first convolution result by using the first dense convolution layer in the level prediction model to obtain the first dense convolution result.

Optionally, fig. 7 is a flow chart of a first dense convolution result determining method provided in the embodiment of the present application, as shown in fig. 7, where the embodiment relates to how to use the first dense convolution layer in the level prediction model to perform feature extraction on the first convolution result to obtain a possible implementation manner of the first dense convolution result, and on the basis of the embodiment, the step S601 includes:

S701, inputting the first convolution result into a fourth convolution layer to obtain a third convolution result.

Specifically, the level prediction model may further include a fourth convolution layer, the first convolution result may be input into the fourth convolution layer, the fourth convolution layer is used to perform convolution processing on the first convolution result, and a result obtained after the convolution processing is a third convolution result.

S702, inputting the first convolution result and the third convolution result into a fifth convolution layer to obtain a fourth convolution result.

Specifically, the level prediction model may further include a fifth convolution layer, the first convolution result and the third convolution result may be input into the fifth convolution layer together, and the fifth convolution layer is used to process the first convolution result and the third convolution result to obtain a fourth convolution result.

S703, inputting the first convolution result, the third convolution result and the fourth convolution result into a sixth convolution layer to obtain a first dense convolution result.

Specifically, the level prediction model may further include a sixth convolution layer, the first convolution result, the third convolution result and the fourth convolution result may be input into the sixth convolution layer together, the sixth convolution layer is used to process the first convolution result, the third convolution result and the fourth convolution result, and the processed result is used as the first dense convolution result.

S602, inputting the first dense convolution result to a first pooling layer in the level prediction model to obtain a first pooling result.

Specifically, the level prediction model may further include a first pooling layer. The first dense convolution result can be input to a first pooling layer in the level prediction model, and pooling processing is carried out on the first dense convolution result by using the first pooling layer to obtain a first pooling result. Wherein the first pooling layer may be one of maximum pooling, average pooling, global average pooling, hybrid pooling, random pooling, power average pooling.

S603, inputting the first pooling result into a second dense convolution layer in the level prediction model to obtain a second dense convolution result.

Specifically, the level prediction model may further include a second dense convolution layer, the first pooled result may be input to the second dense convolution layer in the level prediction model, and the second dense convolution layer is used to process the first pooled result to obtain a second dense convolution set. The first pooled result is input to the second dense convolution layer in the level prediction model, and the second dense convolution result is the same as the above-mentioned S701-S703.

S604, inputting the second dense convolution result to a second pooling layer in the level prediction model to obtain a second pooling result.

Specifically, the level prediction model may further include a second pooling layer, and the second dense convolution result may be input to the second pooling layer in the level prediction model, and the second pooling layer is used to process the second dense convolution result to obtain a second pooled result.

And S605, compressing the second pooling result by using a second full-connection layer in the level prediction model to obtain a target dense convolution result.

Specifically, the level prediction model may further include a second full-connection layer, and the second pooled result may be compressed by using the second full-connection layer in the level prediction model to obtain a compressed result, and the compressed result is used as a target dense convolution result. And the second full-connection layer in the grade prediction model is used for compressing the second pooling result to obtain a compression result, the compression result is corrected to obtain a corrected compression result, and the corrected compression result is used as a target dense convolution result.

S502, determining the creditworthiness level and the net value level of the user according to the target dense convolution result and the target residual error result.

Optionally, fig. 8 is a second flow chart of a method for determining a reputation level and a net value level according to the embodiment of the present application, as shown in fig. 8, where the embodiment relates to a possible implementation manner of determining a reputation level and a net value level of a user according to a target dense convolution result and a target residual result, and on the basis of the embodiment, S502 includes:

S801, determining a first average value between a first probability value and a second probability value; the first probability value is the probability value of each first credibility level in the target dense convolution result, and the second probability value is the probability value of a second credibility level corresponding to the first credibility level in the target residual error result.

The first probability value is the probability value of each first credibility level in the target dense convolution result, and the second probability value is the probability value of the second credibility level corresponding to the first credibility level in the target residual error result. For example, if the first reputation level in the target dense convolution result is high reputation, the probability value corresponding to the high reputation is 0.6, the second reputation level in the target residual result is high reputation, the probability value corresponding to the high reputation is 0.4, then it may be determined that the first average between the first reputation level and the second reputation level is 0.5.

S802, taking the reputation level corresponding to the maximum first mean value as the reputation level of the user.

In the embodiment of the present application, there are a plurality of first means, and the reputation level corresponding to the largest first means may be used as the reputation level of the user. For example, if the first reputation level in the target dense convolution result is high reputation, the probability value corresponding to the high reputation is 0.6, the second reputation level in the target residual result is high reputation, the probability value corresponding to the high reputation is 0.4, then it may be determined that the first average between the first reputation level and the second reputation level is 0.5. The first reputation level in the target dense convolution result is the middle reputation, the probability value corresponding to the middle reputation is 0.3, the second reputation level in the target residual result is the middle reputation, the probability value corresponding to the high reputation is 0.3, and then the first average value between the first reputation level and the second reputation level can be determined to be 0.3. The first reputation level in the target dense convolution result is low reputation, the probability value corresponding to the low reputation is 0.1, the second reputation level in the target residual result is high reputation, the probability value corresponding to the high reputation is 0.3, and then the first average value between the first reputation level and the second reputation level can be determined to be 0.2. The corresponding reputation level with the first mean value of 0.5 is taken as the user reputation level, i.e. the high reputation is taken as the user reputation level.

S803, determining a second average value between the third probability value and the fourth probability value; the third probability value is the probability value of each first net value grade in the target dense convolution result, and the fourth probability value is the probability value of a second net value grade corresponding to the first net value grade in the target residual error result.

The third probability value is the probability value of each first net value grade in the target dense convolution result, and the fourth probability value is the probability value of a second net value grade corresponding to the first net value grade in the target residual error result. For example, if the first net value level in the target dense convolution result is a high net value, the probability value corresponding to the high net value is 0.6, the second net value level in the target residual result is a high net value, the probability value corresponding to the high net value is 0.4, it may be determined that the first average value between the first net value level and the second net value level is 0.5.

S804, taking the net value grade corresponding to the maximum second mean value as the net value grade of the user.

In this embodiment of the present application, there are a plurality of second means, and the net value level corresponding to the largest second means may be used as the net value level of the user. For example, if the first net value level in the target dense convolution result is a high net value, the probability value corresponding to the high net value is 0.6, the second net value level in the target residual result is a high net value, the probability value corresponding to the high net value is 0.4, it may be determined that the second average value between the first net value level and the second net value level is 0.5. The first net value grade in the target dense convolution result is the middle net value, the probability value corresponding to the middle net value is 0.3, the second net value grade in the target residual result is the middle net value, the probability value corresponding to the high net value is 0.3, and then the second average value between the first net value grade and the second net value grade can be determined to be 0.3. The first net value grade in the target dense convolution result is a low net value, the probability value corresponding to the low net value is 0.1, the second net value grade in the target residual result is a high net value, the probability value corresponding to the high net value is 0.3, and then the second average value between the first net value grade and the second net value grade can be determined to be 0.2. The net worth of the user is taken as the net worth of the user corresponding to the second mean value of 0.5, i.e., the net worth of the user is taken as the net worth of the user.

In one embodiment, fig. 9 is a flowchart of a resource borrowing level determining method provided in the embodiment of the present application, as shown in fig. 9, where the embodiment relates to how to obtain a possible implementation manner of a resource borrowing level of a user based on a quantized value corresponding to a reputation level of the user and a quantized value corresponding to a net value level of the user, and on the basis of the above embodiment, the step S204 includes:

s901, determining a first product result of a quantized value corresponding to the credibility level of the user and the weight of the credibility level.

Specifically, a quantized value corresponding to the reputation level of the user may be multiplied by a weight of the reputation level to obtain a first product result. The method can be as follows: the quantized value corresponding to the high reputation level is 6, and the weight of the reputation level is 6, and the first product result of the quantized value corresponding to the user's reputation level (high reputation level) and the weight of the reputation level is 36.

S902, determining a second product of the quantized value corresponding to the net value grade of the user and the weight corresponding to the net value grade.

Specifically, the quantized value corresponding to the net worth of the user may be multiplied by the weight of the net worth, to obtain a second product result. The method can be as follows: the quantized value corresponding to the net worth class is 3, the weight of the reputation class is 4, and the second product of the quantized value corresponding to the net worth class of the user (quantized value corresponding to the net worth class) and the weight corresponding to the net worth class is 12.

S903, determining a summation result of the first product result and the second product result.

The first product result and the second product result may be added to obtain a sum result of the first product result and the second product result. For example, the sum of the first product result and the second product result is 48.

S904, searching a mapping relation table according to the summation result to obtain a resource borrowing level of the user; the mapping relation table comprises mapping relations between different summation results and different resource borrowing levels of users.

The mapping relation table may include mapping relations between different summation results and different resource borrowing levels of the users.

Specifically, a mapping relation table can be searched according to the summation result to obtain the resource borrowing grade of the user. For example, the mapping relation table includes: and the resource borrowing level corresponding to the summation result of 0-10 is none, namely, the resource borrowing is not carried out. The summation result is a first grade of resource borrowing grade corresponding to 11-20, the summation result is a second grade of resource borrowing grade corresponding to 21-30, the summation result is a third grade of resource borrowing grade corresponding to 31-40, and so on, the higher the quantization value of the summation result is, the higher the corresponding resource borrowing grade is. The higher the resource borrowing level, the more resources that represent the user can be borrowed.

In the embodiment of the application, a first product result of a quantized value corresponding to the credit level of the user and a weight of the credit level can be determined, a second product of the quantized value corresponding to the net value level of the user and a weight corresponding to the net value level is determined, a summation result of the first product result and the second product result is determined, and finally a mapping relation table is searched according to the summation result to obtain the resource borrowing level of the user; the mapping relation table comprises mapping relations between different summation results and different resource borrowing levels of users. According to the quantized value corresponding to the credit rating of the user and the quantized value corresponding to the net rating of the user, the resource borrowing rating of the user can be obtained, and the accuracy of resource borrowing rating prediction can be improved.

It should be understood that, although the steps in the flowcharts related to the above embodiments are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.

Based on the same inventive concept, the embodiment of the application also provides a resource borrowing level prediction device for realizing the above-mentioned resource borrowing level prediction method. The implementation of the solution provided by the apparatus is similar to the implementation described in the above method, so the specific limitation in the embodiments of the resource borrowing level prediction apparatus provided in the following may be referred to the limitation of the resource borrowing level prediction method hereinabove, and will not be repeated herein.

In one embodiment, fig. 10 is a block diagram of a resource borrowing level prediction device provided in the embodiment of the present application, and as shown in fig. 10, a resource borrowing level prediction device 1000 is provided, including: an extraction module 1001, a processing module 1002, a first determination module 1003, and a second determination module 1004, wherein:

the extracting module 1001 is configured to perform feature extraction on the user data by using a first convolution layer in the level prediction model, so as to obtain a first convolution result.

The processing module 1002 is configured to perform residual processing on the first convolution result by using a first residual layer in the level prediction model, and determine a target residual result.

A first determining module 1003, configured to determine a reputation level and a net value level of the user according to the target residual result and the first dense convolution layer in the level prediction model.

The second determining module 1004 is configured to obtain a resource borrowing level of the user based on the quantized value corresponding to the reputation level of the user and the quantized value corresponding to the net value level of the user.

In one embodiment, the processing module 1002 includes:

and the first processing submodule is used for carrying out residual processing on the first convolution result by utilizing a first residual layer in the grade prediction model to obtain a first residual result.

The first determination submodule is used for inputting the first residual result to a second residual layer in the grade prediction model to obtain a second residual result.

And the second processing sub-module is used for compressing the second residual error result by using the first full-connection layer in the grade prediction model to obtain a target residual error result.

In one embodiment, the first residual layer comprises a second convolution layer and a third convolution layer, and the first processing submodule comprises:

and the mapping unit is used for performing feature mapping on the first convolution result to obtain a first mapping result.

The first determining unit is used for obtaining a second convolution result according to the first convolution result, the second convolution layer and the third convolution layer.

And the second determining unit is used for summing the first mapping result and the second convolution result to obtain a first residual result.

In one embodiment, the first determining module 1003 includes:

the second determining submodule is used for extracting features of the first convolution result by using the first dense convolution layer in the grade prediction model so as to determine a target dense convolution result;

and the third determining submodule is used for determining the credibility level and the net value level of the user according to the target dense convolution result and the target residual error result.

In one embodiment, the second determination submodule includes:

and the extraction unit is used for extracting the characteristics of the first convolution result by using a first dense convolution layer in the grade prediction model to obtain the first dense convolution result.

And the third determining unit is used for inputting the first dense convolution result to a first pooling layer in the level prediction model to obtain a first pooling result.

And the fourth determining unit is used for inputting the first pooling result into a second dense convolution layer in the level prediction model to obtain a second dense convolution result.

And a fifth determining unit, configured to input the second dense convolution result to a second pooling layer in the level prediction model, to obtain a second pooling result.

And the processing unit is used for compressing the second pooling result by using a second full-connection layer in the grade prediction model to obtain a target dense convolution result.

In one embodiment, the first dense convolution layer includes a fourth convolution layer, a fifth convolution layer, and a sixth convolution layer, and the extracting unit is specifically configured to input the first convolution result into the fourth convolution layer to obtain a third convolution result; inputting the first convolution result and the third convolution result into a fifth convolution layer to obtain a fourth convolution result; and inputting the first convolution result, the third convolution result and the fourth convolution result into a sixth convolution layer to obtain a first dense convolution result.

In one embodiment, the third determination submodule includes:

a sixth determining unit configured to determine a first average value between the first probability value and the second probability value; the first probability value is the probability value of each first credibility level in the target dense convolution result, and the second probability value is the probability value of a second credibility level corresponding to the first credibility level in the target residual error result.

And a seventh determining unit, configured to take the reputation level corresponding to the maximum first mean value as the reputation level of the user.

An eighth determining unit configured to determine a second average value between the third probability value and the fourth probability value; the third probability value is the probability value of each first net value grade in the target dense convolution result, and the fourth probability value is the probability value of a second net value grade corresponding to the first net value grade in the target residual error result.

And the ninth determining unit is used for taking the net value grade corresponding to the largest second mean value as the net value grade of the user.

In one embodiment, the second determining module 1004 is specifically configured to determine a first product of a quantized value corresponding to the reputation level of the user and a weight of the reputation level; determining a second product of the quantized value corresponding to the net value grade of the user and the weight corresponding to the net value grade; determining a summation result of the first product result and the second product result; searching a mapping relation table according to the summation result to obtain a resource borrowing grade of the user; the mapping relation table comprises mapping relations between different summation results and different resource borrowing levels of users.

The above-mentioned various modules in the resource borrowing level prediction apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, where the computer device may be a server, and an internal structure diagram of the computer device may be shown in fig. 11, and fig. 11 is an internal structure diagram of the computer device provided in an embodiment of the present application. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is for storing user data. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements a resource borrowing level prediction method.

It will be appreciated by those skilled in the art that the structure shown in fig. 11 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the computer device to which the present application applies, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.

In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:

residual processing is carried out on the first convolution result by utilizing a first residual layer in the grade prediction model, and a target residual result is determined;

In one embodiment, the processor when executing the computer program further performs the steps of:

Residual processing is carried out on the first convolution result by using a first residual layer in the grade prediction model, so as to obtain a first residual result;

inputting the first residual result into a second residual layer in the grade prediction model to obtain a second residual result;

and compressing the second residual error result by using a first full-connection layer in the grade prediction model to obtain a target residual error result.

and summing the first mapping result and the second convolution result to obtain a first residual result.

extracting features of the first convolution result by using a first dense convolution layer in the level prediction model to determine a target dense convolution result;

and determining the creditworthiness level and the net value level of the user according to the target dense convolution result and the target residual error result.

Extracting features of the first convolution result by using a first dense convolution layer in the level prediction model to obtain the first dense convolution result;

inputting the first pooling result into a second dense convolution layer in the level prediction model to obtain a second dense convolution result;

and compressing the second pooling result by using a second full-connection layer in the grade prediction model to obtain a target dense convolution result.

inputting the first convolution result into a fourth convolution layer to obtain a third convolution result;

inputting the first convolution result and the third convolution result into a fifth convolution layer to obtain a fourth convolution result;

and inputting the first convolution result, the third convolution result and the fourth convolution result into a sixth convolution layer to obtain a first dense convolution result.

determining a first product result of a quantized value corresponding to the reputation level of the user and a weight of the reputation level;

determining a second product of the quantized value corresponding to the net value grade of the user and the weight corresponding to the net value grade;

searching a mapping relation table according to the summation result to obtain a resource borrowing grade of the user; the mapping relation table comprises mapping relations between different summation results and different resource borrowing levels of users.

In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:

In one embodiment, the computer program when executed by the processor further performs the steps of:

In one embodiment, a computer program product is provided comprising a computer program which, when executed by a processor, performs the steps of:

It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use, and processing of the related data are required to meet the related regulations.

Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The foregoing examples represent only a few embodiments of the present application, which are described in more detail and are not thereby to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims

1. A method for predicting a resource borrowing level, the method comprising:

determining the creditworthiness level and the net value level of the user according to the target residual error result and a first dense convolution layer in the level prediction model;

And obtaining the resource borrowing level of the user based on the quantized value corresponding to the credit level of the user and the quantized value corresponding to the net value level of the user.

2. The method of claim 1, wherein the performing residual processing on the first convolution result with the first residual layer in the level prediction model to determine a target residual result comprises:

residual processing is carried out on the first convolution result by utilizing a first residual layer in the grade prediction model, so as to obtain a first residual result;

inputting the first residual result to a second residual layer in the grade prediction model to obtain a second residual result;

3. The method according to claim 2, wherein the first residual layer includes a second convolution layer and a third convolution layer, and the performing residual processing on the first convolution result with the first residual layer in the level prediction model to obtain a first residual result includes:

4. The method of claim 1, wherein said determining the user's reputation level and net value level from the target residual result and the first dense convolutional layer in the level prediction model comprises:

5. The method of claim 4, wherein the feature extracting the first convolution result with the first dense convolution layer in the level prediction model to determine a target dense convolution result comprises:

Inputting the first pooling result to a second dense convolution layer in the grade prediction model to obtain a second dense convolution result;

6. The method of claim 5, wherein the first dense convolution layer comprises a fourth convolution layer, a fifth convolution layer, and a sixth convolution layer, wherein the feature extracting the first convolution result using the first dense convolution layer in the level prediction model to obtain a first dense convolution result comprises:

7. The method of any of claims 4-6, wherein the determining the user's reputation level and net value level from the target dense convolution result and the target residual result comprises:

8. The method of claim 1, wherein the obtaining the resource borrowing level of the user based on the quantized value corresponding to the user's reputation level and the quantized value corresponding to the user's net value level comprises:

determining a second product of a quantized value corresponding to the net value grade of the user and a weight corresponding to the net value grade;

9. A resource borrowing level prediction apparatus, the apparatus comprising:

the processing module is used for carrying out residual processing on the first convolution result by utilizing a first residual layer in the grade prediction model and determining a target residual result;

the first determining module is used for determining the creditworthiness level and the net value level of the user according to the target residual error result and the first dense convolution layer in the level prediction model;

10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 8 when the computer program is executed.

11. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 8.

12. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.