CN112017061A

CN112017061A - Financial risk prediction method and device based on Bayesian deep learning and electronic equipment

Info

Publication number: CN112017061A
Application number: CN202010682561.XA
Authority: CN
Inventors: 王骞
Original assignee: Beijing Qilu Information Technology Co Ltd
Current assignee: Beijing Qilu Information Technology Co Ltd
Priority date: 2020-07-15
Filing date: 2020-07-15
Publication date: 2020-12-01

Abstract

The invention provides a financial risk prediction method and device based on Bayesian deep learning and electronic equipment. The method comprises the following steps: acquiring a historical user data set, and establishing a training data set and a test data set according to the historical user data set; constructing a Bayes deep learning model, and performing parameter estimation on parameters to be optimized of a deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers; training the Bayes deep learning model after the parameter optimization by using a training data set; acquiring user characteristic data of a target user, and calculating a financial risk assessment value by using a Bayesian deep learning model; and according to the preset judgment rule and the calculated financial risk assessment value, performing financial risk prediction on the target user. The method of the invention enhances the generalization prediction capability of the model, improves the accuracy of the model prediction and further reduces the financial risk.

Description

Financial risk prediction method and device based on Bayesian deep learning and electronic equipment

Technical Field

The invention relates to the field of computer information processing, in particular to a financial risk prediction method and device based on Bayesian deep learning and electronic equipment.

Background

Risk control (wind control for short) refers to the risk manager taking various measures and methods to eliminate or reduce the various possibilities of occurrence of a risk event, or the risk controller reducing the losses caused when a risk event occurs. The risk control is generally applied to the financial industry, such as risk control on company transactions, merchant transactions or personal transactions and the like.

In the prior art, the main purpose of financial risk assessment is how to distinguish good customers from bad customers, and assess the risk condition of users, so as to reduce credit risk and realize profit maximization. At present, a Logistic regression statistical method is mainly adopted to calculate the risk score, for example, a Logistic regression method selects 10-20 features as primers, and the effect is not good when high-dimensional data is processed. In addition, with the development of machine learning technology, especially the XGBoost model in the tree model is widely applied in the field of financial risk assessment, but the model has many related parameters, and after training to a certain degree, the prediction effect is difficult to be improved.

Therefore, it is necessary to provide a financial risk prediction method with higher accuracy.

Disclosure of Invention

In order to improve the model prediction precision, accurately evaluate the risk condition of a user and further reduce the financial risk, the invention provides a financial risk prediction method, which comprises the following steps: acquiring a historical user data set, and establishing a training data set and a testing data set according to the historical user data set, wherein the training data set comprises user characteristic data and financial performance data, and the testing data set comprises a parameter combination set; constructing a Bayes deep learning model, and performing parameter estimation on parameters to be optimized of a deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers; training the parameter-optimized Bayesian deep learning model by using the training data set; acquiring user characteristic data of a target user, and calculating a financial risk assessment value by using the Bayesian deep learning model; and predicting the financial risk of the target user according to a preset judgment rule and the calculated financial risk assessment value.

Preferably, the performing parameter estimation on the parameter to be optimized of the deep neural network by using the bayesian statistical method includes: setting prior probability of the parameter to be optimized to obey standard normal distribution, and determining posterior probability distribution of the parameter to be optimized through likelihood and prior probability: sampling the distribution of the weight parameters and the bias parameters for multiple times by using an MCMC method to obtain a parameter combination set; and training the Bayesian deep learning model pair by using the parameter combination set and the training data set.

Preferably, the KL-divergence is defined by minimizing the KL-divergence to obtain a posterior probability distribution of the weight parameter, the bias parameter.

Preferably, a posterior probability distribution of the weight parameters and the bias parameters is obtained by using a Monte Carlo dropout method; when the parameter combination set is used for testing, multiple times of forward propagation are carried out on the input same user characteristics, and the weight parameters, the average number of the bias parameters and the statistical variance value are calculated so as to optimize the weight parameters and the bias parameters.

Preferably, the deep neural network comprises a dropout layer and the data in the training dataset and the parameter combination set are fitted using a Relu activation function.

Preferably, the parameters to be optimized further include the number of layers of the deep neural network, the number of iterations, and the learning rate.

Preferably, the financial performance data includes a probability of breach and/or a probability of overdue.

Preferably, the predicting the financial risk of the target user according to the preset judgment rule and the calculated financial risk assessment value includes: the preset judgment rule comprises setting a first threshold value and a second threshold value; when the calculated financial evaluation value is larger than the first threshold value, judging that the user is a first target user; when the calculated financial evaluation value is larger than a second threshold value and smaller than a first threshold value, judging that the user is a second target user; when the calculated financial evaluation value is less than a second threshold value, the user is judged to be a non-target user.

Preferably, the method further comprises the following steps: and customizing a resource allocation strategy, an increase strategy, a decrease strategy or a limit strategy for maximizing profits for different users.

In addition, the present invention also provides a financial risk prediction apparatus, comprising: the system comprises a data acquisition module, a data acquisition module and a data analysis module, wherein the data acquisition module is used for acquiring a historical user data set, establishing a training data set and a test data set according to the historical user data set, the training data set comprises user characteristic data and financial performance data, and the test data set comprises a parameter combination set; the building module is used for building a Bayes deep learning model and carrying out parameter estimation on parameters to be optimized of the deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers; the training module is used for training the Bayes deep learning model after parameter optimization by using the training data set; the calculation module is used for acquiring user characteristic data of a target user and calculating a financial risk assessment value by using the Bayesian deep learning model; and the prediction module is used for predicting the financial risk of the target user according to a preset judgment rule and the calculated financial risk assessment value.

Preferably, the method further comprises the following steps: the setting module is used for setting the prior probability of the parameter to be optimized to obey standard normal distribution and determining the posterior probability distribution of the parameter to be optimized according to the likelihood and the prior probability; and the sampling module is used for sampling the distribution of the weight parameters and the bias parameters for multiple times by using an MCMC method to obtain a parameter combination set, and training the Bayesian deep learning model pair by using the parameter combination set and the training data set.

Preferably, the preset judgment rule includes setting a first threshold and a second threshold; when the calculated financial evaluation value is larger than the first threshold value, judging that the user is a first target user; when the calculated financial evaluation value is larger than a second threshold value and smaller than a first threshold value, judging that the user is a second target user; when the calculated financial evaluation value is less than a second threshold value, the user is judged to be a non-target user.

Preferably, the system further comprises a customization information module, wherein the customization information module is used for customizing a resource allocation strategy, an increase strategy, a decrease strategy or a limit strategy for maximizing profits for different users.

In addition, the present invention also provides an electronic device, wherein the electronic device includes: a processor; and a memory storing computer executable instructions that, when executed, cause the processor to perform the financial risk prediction method of the present invention.

Further, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more programs which, when executed by a processor, implement the financial risk prediction method of the present invention.

Advantageous effects

Compared with the prior art, the financial risk prediction method provided by the invention constructs the Bayes deep learning model by combining the Bayes method and the deep learning method, wherein the weight and the bias are changed into distribution through Bayes deep learning, the learning effect is more stable, the total information, the sample information and the prior information of a user are fully utilized, the deep neural network is utilized, and the dropout (or MCMC drop out) method is adopted through multiple layers and multiple neurons, so that the overfitting is prevented, the model fitting precision is improved, the generalization prediction capability of the model is enhanced, the model prediction precision is improved, the risk condition of the user is accurately evaluated, and the financial risk is further reduced.

Drawings

In order to make the technical problems solved by the present invention, the technical means adopted and the technical effects obtained more clear, the following will describe in detail the embodiments of the present invention with reference to the accompanying drawings. It should be noted, however, that the drawings described below are only illustrations of exemplary embodiments of the invention, from which other embodiments can be derived by those skilled in the art without inventive faculty.

Fig. 1 is a flowchart of an example of a financial risk prediction method of embodiment 1 of the present invention.

Fig. 2 is a flowchart of another example of the financial risk prediction method of embodiment 1 of the present invention.

Fig. 3 is a flowchart of still another example of the financial risk prediction method of embodiment 1 of the present invention.

Fig. 4 is a schematic diagram of an example of a financial risk prediction apparatus according to embodiment 2 of the present invention.

Fig. 5 is a schematic diagram of another example of the financial risk prediction apparatus according to embodiment 2 of the present invention.

Fig. 6 is a schematic diagram of still another example of the financial risk prediction apparatus of embodiment 2 of the present invention.

Fig. 7 is a block diagram of an exemplary embodiment of an electronic device according to the present invention.

Fig. 8 is a block diagram of an exemplary embodiment of a computer-readable medium according to the present invention.

Detailed Description

Exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. The exemplary embodiments, however, may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these exemplary embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the invention to those skilled in the art. The same reference numerals denote the same or similar elements, components, or parts in the drawings, and thus their repetitive description will be omitted.

Features, structures, characteristics or other details described in a particular embodiment do not preclude the fact that the features, structures, characteristics or other details may be combined in a suitable manner in one or more other embodiments in accordance with the technical idea of the invention.

In describing particular embodiments, the present invention has been described with reference to features, structures, characteristics or other details that are within the purview of one skilled in the art to provide a thorough understanding of the embodiments. One skilled in the relevant art will recognize, however, that the invention may be practiced without one or more of the specific features, structures, characteristics, or other details.

The flow charts shown in the drawings are merely illustrative and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.

The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. I.e. these functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor means and/or microcontroller means.

It will be understood that, although the terms first, second, third, etc. may be used herein to describe various elements, components, or sections, these terms should not be construed as limiting. These phrases are used to distinguish one from another. For example, a first device may also be referred to as a second device without departing from the spirit of the present invention.

The term "and/or" and/or "includes any and all combinations of one or more of the associated listed items.

In order to improve the model prediction precision, accurately evaluate the risk condition of the user and further reduce the financial risk, the invention uses a combination mode of a Bayes method and a deep learning method to construct a Bayes deep learning model and calculate the financial evaluation value of the user so as to predict the financial risk of the user, and the detailed evaluation process is described in detail below.

Example 1

Hereinafter, an embodiment of the financial risk prediction method of the present invention will be described with reference to fig. 1 to 3.

FIG. 1 is a flow chart of a financial risk prediction method of the present invention. As shown in fig. 1, a financial risk prediction method includes the following steps.

Step S101, a historical user data set is obtained, a training data set and a testing data set are established according to the historical user data set, the training data set comprises user characteristic data and financial performance data, and the testing data set comprises a parameter combination set.

And S102, constructing a Bayes deep learning model, and performing parameter estimation on parameters to be optimized of the deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers.

And step S103, training the Bayes deep learning model after parameter optimization by using the training data set.

And step S104, acquiring user characteristic data of the target user, and calculating a financial risk assessment value by using the Bayesian deep learning model.

And step S105, performing financial risk prediction on the target user according to a preset judgment rule and the calculated financial risk assessment value.

First, in step S101, a historical user data set is obtained, a training data set and a test data set are established according to the historical user data set, the training data set includes user feature data and financial performance data, and the test data set includes a parameter combination set.

In this example, the user characteristic data includes user basic information data, social behavior data, and the like. Specifically, for example, the user's age, sex, occupation, monthly income/annual income, and the like.

It should be noted that the financial performance data includes data related to the performance of the financial product by the user. In this example, the financial performance data includes a probability of breach and/or a probability of overdue. However, the present invention is not limited thereto, and the above description is only by way of example and is not to be construed as limiting the present invention.

In this example, for the training data set, good and bad samples are defined, and the label is 0, 1, where 1 represents a sample in which the user's overdue probability (or default probability) is Y or more, and 0 represents a sample in which the user's overdue probability (or default probability) is less than Y. Generally, the lower the user's probability of overdue (or probability of default), the better the loan is to recover principal, the better the efficiency of the use of funds, the lower the risk level of the property, and vice versa.

Next, in step 102, a bayesian deep learning model is constructed, and parameters to be optimized of the deep neural network are estimated by using a bayesian statistical method, where the parameters to be optimized include weight parameters and bias parameters between layers.

To solve the nonlinear separable problem, a multi-layer functional neuron, i.e., a Deep Neural Network (DNN) is used, wherein the Deep Neural Network specifically includes an input layer, an hidden layer, and an output layer, and both the hidden layer and the output layer neurons are functional neurons that possess activation functions.

In this example, the deep neural network further includes a dropout layer and fits the training dataset and the data in the combined set of parameters using a Relu activation function.

Preferably, the parameters to be optimized include weight parameters and bias parameters between layers of the deep neural network.

It should be noted that, for the selection of the parameters to be optimized, adaptive selection can be performed according to specific scenarios and service requirements, and the above description is only used as a preferred example, and is not to be construed as limiting the present invention. In other examples, the parameters to be optimized also include the number of layers of the deep neural network, the number of iterations, and the learning rate.

Further, the weight parameter between the layers of the neural network is set not simply as a value but from a distribution normal distribution. And recognizing the weight parameter w by using a Bayes method, adjusting the weight parameter w from p (w) to p (w | x, w), and determining p (w | x, w) according to a Bayes formula. The calculation principle using the bayesian method will be described in detail below.

The bayesian formula is as follows.

Wherein p (z | x) is called posterior probability (posterior), p (x, z) is called joint probability, p (x | z) is called likelihood (likelihood), p (z) is called prior probability (prior), and p (x) is called evidence.

Setting for the introduction of the whole probability formula p (x) ═ p (x | z) p (z) dz (2)

When z is a discrete variable, the denominator integral sign ^ in expression (2) is changed to the summation sign ^.

The formula for the posterior probability of the weight parameter w is as follows:

as can be seen from equation (3), to obtain the posterior probability p (w | x, y) of w, the prior probability p (w) needs to be determined.

Specifically, the prior probability of the parameter to be optimized is set to obey standard normal distribution, and the posterior probability distribution of the parameter to be optimized is determined through the likelihood and the prior probability. For example, p (w) is initially set to a standard normal distribution (i.e., p (w) ═ N (μ, Σ)), and the likelihood p (y | x, w) is a function of w. When w is equal to a certain value, the numerator of equation (3) can be easily calculated, and the denominator is also calculated in order to obtain the posterior probability p (w | x, y).

It should be noted that, the integral of the denominator in the formula (3) is performed on the value space of w, the value space of the single weight of the deep neural network may be the real number set R, and the space formed by the weights together is quite complex. The calculation of the denominator will be described below.

As shown in fig. 2, the method of the present invention further includes a step S201 of sampling the distribution of the weight parameter and the bias parameter a plurality of times using the MCMC method.

In step S201, the distributions of the weight parameters and the bias parameters are sampled a plurality of times using the MCMC method, thereby obtaining a set of parameter combinations, and the denominator is approximately calculated for determining the posterior probability of w.

Specifically, the Bayesian deep learning model pair is trained using the parameter combination set and the training data set to optimize parameters.

Preferably, to enhance the fitting ability of the model, the deep neural network comprises a dropout layer and uses a Relu function as an activation function to fit the training data set and the data in the parameter combination set. The Relu function is as follows.

In another example, the distribution p of the posterior probabilities is approximated, for example, with a relatively simple distribution q, i.e., the difference between distributions q and p is directly minimized regardless of denominator integration, for example using a KL divergence calculation.

Specifically, the KL divergence is defined by minimizing the KL divergence. Further, if the overall distribution of the prior normal distribution q θ (z) is approximated to the posterior distribution, ELBO ═ E_qθ[log p(x|z)]-KL[qθ(z)||p(z)]The loss function is KL (p, q) ═ E_z～p[log p(z)]-E_z～p[log q(z)]. Thereby, the posterior probability distribution of the weight parameter and the bias parameter is obtained.

In another example, the posterior probability distribution of the weight parameter and the bias parameter is obtained by using a monte carlo dropout method.

Specifically, when a parameter combination set is used for testing, multiple forward propagation is performed on the input same user characteristic, the weight parameter, the average number of the bias parameters and the statistical variance value are calculated to optimize the weight parameter and the bias parameters, and finally, the neural network after parameter optimization is output.

Next, in step S103, the bayesian deep learning model after parameter optimization is trained using the training data set.

Specifically, the data training set D { (x) established in step S101 is used₁，y₁)，(x₂，y₂)，...，(x_m，y_m) And training the Bayes deep learning model after parameter optimization by using the data training set, thereby realizing classification of overdue probability (or default probability) of the historical users based on the historical user characteristic data.

Next, in step S104, user feature data of the target user is acquired, and the financial risk assessment value is calculated using the bayesian deep learning model.

It should be noted that the specific meaning and the obtaining manner of the user characteristic data and the financial risk assessment value are the same as those of the user characteristic data and the financial risk assessment value in step S101, and therefore, the description thereof is omitted.

Specifically, the acquired user feature data of the target user is input into a trained Bayesian deep learning model, and the financial risk assessment value of the target user is calculated.

In other examples, the method of the present invention further includes step S301 of presetting the determination rule, see fig. 3.

In step S301, a determination rule is set in advance.

Specifically, the determination rule includes setting a first threshold value and a second threshold value.

In the present example, when the calculated financial evaluation value is greater than the first threshold value, it is determined that the user is a first target user; when the calculated financial evaluation value is larger than a second threshold value and smaller than a first threshold value, judging that the user is a second target user; when the calculated financial evaluation value is less than a second threshold value, the user is judged to be a non-target user.

It should be noted that the above-mentioned embodiments are only preferred embodiments, and should not be construed as limiting the present invention. In other embodiments, the user risk control model may also be used to calculate the risk level of the user, or as a prediction module in other risk prediction models, and so on.

Those skilled in the art will appreciate that all or part of the steps to implement the above-described embodiments are implemented as programs (computer programs) executed by a computer data processing apparatus. When the computer program is executed, the method provided by the invention can be realized. Furthermore, the computer program may be stored in a computer readable storage medium, which may be a readable storage medium such as a magnetic disk, an optical disk, a ROM, a RAM, or a storage array composed of a plurality of storage media, such as a magnetic disk or a magnetic tape storage array. The storage medium is not limited to centralized storage, but may be distributed storage, such as cloud storage based on cloud computing.

Compared with the prior art, the financial risk prediction method provided by the invention constructs the Bayes deep learning model by combining the Bayes method with the deep learning method, wherein the weight and the bias are changed into distribution through the Bayes deep learning, the learning effect is more stable, the total information, the sample information and the prior information of the user are fully utilized, the deep neural network is utilized, and the dropout (or MCMC drop out) method is adopted through multiple layers and multiple neurons, so that the overfitting is prevented, the model fitting precision is improved, the generalization prediction capability of the model is enhanced, the model prediction precision is improved, the risk condition of the user is accurately evaluated, and the financial risk is further reduced.

Example 2

Embodiments of the apparatus of the present invention are described below, which may be used to perform method embodiments of the present invention. The details described in the device embodiments of the invention should be regarded as complementary to the above-described method embodiments; reference is made to the above-described method embodiments for details not disclosed in the apparatus embodiments of the invention.

Referring to fig. 4, 5 and 6, the present invention also provides a financial risk prediction apparatus 400, including: a data obtaining module 401, configured to obtain a historical user data set, and establish a training data set and a test data set according to the historical user data set, where the training data set includes user feature data and financial performance data, and the test data set includes a parameter combination set; a building module 402, configured to build a bayesian deep learning model, and perform parameter estimation on parameters to be optimized of the deep neural network by using a bayesian statistical method, where the optimized parameters include weight parameters and bias parameters between layers; a training module 403, configured to train the parameter-optimized bayesian deep learning model using the training data set; a calculating module 404, configured to obtain user feature data of a target user, and calculate a financial risk assessment value using the bayesian deep learning model; and the prediction module 405 is configured to perform financial risk prediction on the target user according to a preset judgment rule and the calculated financial risk assessment value.

As shown in fig. 5, the method further includes: a setting module 501, configured to set a prior probability of a parameter to be optimized to comply with a standard normal distribution, and determine a posterior probability distribution of the parameter to be optimized according to the likelihood and the prior probability; a sampling module 502, configured to perform multiple sampling on the distribution of the weight parameter and the bias parameter by using an MCMC method to obtain a parameter combination set, and train the bayesian deep learning model pair by using the parameter combination set and the training data set.

Preferably, the optimization parameters further include the number of layers of the deep neural network, the number of iterations, and the learning rate.

As shown in fig. 6, the system further comprises a customization information module 601, wherein the customization information module 601 is used for customizing a resource allocation strategy, an increase strategy, a decrease strategy or a limit strategy for maximizing profit for different users.

In embodiment 2, the same portions as those in embodiment 1 are not described.

Those skilled in the art will appreciate that the modules in the above-described embodiments of the apparatus may be distributed as described in the apparatus, and may be correspondingly modified and distributed in one or more apparatuses other than the above-described embodiments. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.

Compared with the prior art, the financial risk prediction device provided by the invention constructs the Bayes deep learning model by combining the Bayes method and the deep learning method, wherein the weight and the bias are changed into distribution through the Bayes deep learning, the learning effect is more stable, the total information, the sample information and the prior information of a user are fully utilized, the deep neural network is utilized, and the dropout (or MCMC drop out) method is adopted through multiple layers and multiple neurons, so that the overfitting is prevented, the model fitting precision is improved, the generalization prediction capability of the model is enhanced, the model prediction precision is improved, the risk condition of the user is accurately evaluated, and the financial risk is further reduced.

Example 3

In the following, embodiments of the electronic device of the present invention are described, which may be regarded as specific physical implementations for the above-described embodiments of the method and apparatus of the present invention. Details described in the embodiments of the electronic device of the invention should be considered supplementary to the embodiments of the method or apparatus described above; for details which are not disclosed in embodiments of the electronic device of the invention, reference may be made to the above-described embodiments of the method or the apparatus.

Fig. 7 is a block diagram of an exemplary embodiment of an electronic device according to the present invention. An electronic apparatus 200 according to this embodiment of the present invention is described below with reference to fig. 7. The electronic device 200 shown in fig. 7 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.

As shown in fig. 7, the electronic device 200 is embodied in the form of a general purpose computing device. The components of the electronic device 200 may include, but are not limited to: at least one processing unit 210, at least one memory unit 220, a bus 230 connecting different system components (including the memory unit 220 and the processing unit 210), a display unit 240, and the like.

Wherein the storage unit stores program code executable by the processing unit 210 to cause the processing unit 210 to perform steps according to various exemplary embodiments of the present invention described in the processing method section of the electronic device described above in this specification. For example, the processing unit 210 may perform the steps as shown in fig. 1.

The memory unit 220 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)2201 and/or a cache memory unit 2202, and may further include a read only memory unit (ROM) 2203.

The storage unit 220 may also include a program/utility 2204 having a set (at least one) of program modules 2205, such program modules 2205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.

Bus 230 may be one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

The electronic device 200 may also communicate with one or more external devices 300 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 200, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 200 to communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 250. Also, the electronic device 200 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the Internet) via the network adapter 260. The network adapter 260 may communicate with other modules of the electronic device 200 via the bus 230. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments of the present invention described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a computer-readable storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to make a computing device (which can be a personal computer, a server, or a network device, etc.) execute the above-mentioned method according to the present invention. The computer program, when executed by a data processing apparatus, enables the computer readable medium to carry out the above-described methods of the invention.

As shown in fig. 8, the computer program may be stored on one or more computer readable media. The computer readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).

In summary, the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functionality of some or all of the components in embodiments in accordance with the invention may be implemented in practice using a general purpose data processing device such as a microprocessor or a Digital Signal Processor (DSP). The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.

While the foregoing embodiments have described the objects, aspects and advantages of the present invention in further detail, it should be understood that the present invention is not inherently related to any particular computer, virtual machine or electronic device, and various general-purpose machines may be used to implement the present invention. The invention is not to be considered as limited to the specific embodiments thereof, but is to be understood as being modified in all respects, all changes and equivalents that come within the spirit and scope of the invention.

Claims

1. A financial risk prediction method based on Bayesian deep learning is characterized by comprising the following steps:

acquiring a historical user data set, and establishing a training data set and a testing data set according to the historical user data set, wherein the training data set comprises user characteristic data and financial performance data, and the testing data set comprises a parameter combination set;

constructing a Bayes deep learning model, and performing parameter estimation on parameters to be optimized of a deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers;

training the parameter-optimized Bayesian deep learning model by using the training data set;

acquiring user characteristic data of a target user, and calculating a financial risk assessment value by using the Bayesian deep learning model;

and predicting the financial risk of the target user according to a preset judgment rule and the calculated financial risk assessment value.

2. The financial risk prediction method of claim 1, wherein the performing parameter estimation on the parameters to be optimized of the deep neural network using a bayesian statistical method comprises:

setting prior probability of the parameter to be optimized to obey standard normal distribution, and determining posterior probability distribution of the parameter to be optimized through likelihood and prior probability:

sampling the distribution of the weight parameters and the bias parameters for multiple times by using an MCMC method to obtain a parameter combination set;

and training the Bayesian deep learning model pair by using the parameter combination set and the training data set.

3. The financial risk prediction method of any one of claims 1-2,

and defining KL divergence, and obtaining posterior probability distribution of the weight parameters and the bias parameters by minimizing the KL divergence.

4. The financial risk prediction method of any one of claims 1-3,

obtaining posterior probability distribution of the weight parameters and the bias parameters by using a Monte Carlo dropout method;

when the parameter combination set is used for testing, multiple times of forward propagation are carried out on the input same user characteristics, and the weight parameters, the average number of the bias parameters and the statistical variance value are calculated so as to optimize the weight parameters and the bias parameters.

5. The financial risk prediction method of any one of claims 1-4,

the deep neural network includes a dropout layer and fits data in the training dataset and the parameter combination set using a Relu activation function.

6. The financial risk prediction method of any one of claims 1-5,

the parameters to be optimized further comprise the number of layers of the deep neural network, iteration times and a learning rate.

7. The financial risk prediction method of any one of claims 1-6,

the financial performance data includes a probability of breach and/or a probability of overdue.

8. A financial risk prediction device based on Bayesian deep learning is characterized by comprising:

the system comprises a data acquisition module, a data acquisition module and a data analysis module, wherein the data acquisition module is used for acquiring a historical user data set, establishing a training data set and a test data set according to the historical user data set, the training data set comprises user characteristic data and financial performance data, and the test data set comprises a parameter combination set;

the building module is used for building a Bayes deep learning model and carrying out parameter estimation on parameters to be optimized of the deep neural network by using a Bayes statistical method, wherein the parameters to be optimized comprise weight parameters and bias parameters among layers;

the training module is used for training the Bayes deep learning model after parameter optimization by using the training data set;

the calculation module is used for acquiring user characteristic data of a target user and calculating a financial risk assessment value by using the Bayesian deep learning model;

and the prediction module is used for predicting the financial risk of the target user according to a preset judgment rule and the calculated financial risk assessment value.

9. An electronic device, wherein the electronic device comprises:

a processor; and the number of the first and second groups,

a memory storing computer-executable instructions that, when executed, cause the processor to perform the bayesian deep learning based financial risk prediction method of any of claims 1-7.

10. A computer readable storage medium, wherein the computer readable storage medium stores one or more programs which, when executed by a processor, implement the bayesian deep learning based financial risk prediction method of any of claims 1-7.