Transaction risk factor prediction method, device, electronic device and storage medium

Abstract

The present disclosure provides a transaction risk factor prediction method, device, electronic device and storage medium, which can be applied to the field of artificial intelligence technology and the financial field. The transaction risk factor prediction method includes: in response to receiving a transaction risk factor prediction request, obtaining transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include concentration risk factor, risk weight and initial sensitivity; inputting the transaction risk factor prediction parameters into a support vector machine module in a trained transaction risk factor prediction model, and outputting an initial transaction risk factor prediction value; inputting the initial transaction risk factor prediction value into a radial basis function neural network module in a trained transaction risk factor prediction model, and outputting a transaction risk factor prediction value.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology and the financial field, and more specifically, to a method, device, electronic device, computer-readable storage medium and a computer program product for predicting transaction risk factors. Background Art

[0002] With the development of artificial intelligence technology, the optimization calculation of transaction risk factors based on relationship networks can obtain more reliable transaction risk factor calculation results, thereby providing more convincing data support for financial decision makers.

[0003] In the process of realizing the concept of the present disclosure, the inventors found that there are at least the following problems in the related art: the existing transaction risk factor prediction method cannot guarantee the accuracy of the prediction in the rapidly changing financial market environment. Summary of the invention

[0004] In view of this, the present disclosure provides a method, device, electronic device, computer-readable storage medium and computer program product for predicting transaction risk factors.

[0005] According to one aspect of the present disclosure, a method for predicting a transaction risk factor is provided, comprising:

[0006] In response to receiving a transaction risk factor prediction request, obtaining transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include the concentration risk factor, the risk weight, and the initial sensitivity;

[0007] Inputting the transaction risk factor prediction parameters into the support vector machine module in the trained transaction risk factor prediction model to output the initial transaction risk factor prediction value; and

[0008] The above initial transaction risk factor prediction value is input into the radial basis function neural network module in the trained transaction risk factor prediction model, and the transaction risk factor prediction value is output.

[0009] According to an embodiment of the present disclosure, the method for predicting a transaction risk factor further includes:

[0010] According to the predicted value of the transaction risk factor, the transaction to be predicted corresponding to the transaction data to be predicted is evaluated to obtain the margin amount of the transaction to be predicted;

[0011] When the margin amount meets the preset warning condition, outputting prompt information indicating that the predicted transaction is allowed to be executed; and

[0012] In the case where the margin amount does not satisfy the preset warning condition, a warning message indicating that the execution of the transaction to be predicted is prohibited is output.

[0013] According to an embodiment of the present disclosure, the training method of the above-mentioned trained transaction risk factor prediction model includes:

[0014] Extracting training sample data corresponding to the sample transaction data from the source database, wherein the training sample data includes a sample concentration risk factor, a sample risk weight, a sample initial sensitivity, and a true value of the transaction risk factor;

[0015] Input the training sample data into the transaction risk factor prediction model, and output the sample transaction risk factor prediction value; and

[0016] The model parameters of the transaction risk factor prediction model are adjusted by using the predicted values ​​of the sample transaction risk factors and the actual values ​​of the transaction risk factors to obtain the trained transaction risk factor prediction model.

[0017] According to an embodiment of the present disclosure, the transaction risk factor prediction model includes the support vector machine module and the radial basis function neural network module;

[0018] The above training sample data is input into the transaction risk factor prediction model, and the output sample transaction risk factor prediction value includes:

[0019] Input the training sample data into the support vector machine module to output the initial sample transaction risk factor prediction value; and

[0020] Inputting the above-mentioned initial sample transaction risk factor prediction value into the above-mentioned radial basis function neural network module, and outputting the above-mentioned sample transaction risk factor prediction value;

[0021] The above-mentioned use of the above-mentioned sample transaction risk factor prediction value and the above-mentioned transaction risk factor true value to adjust the model parameters of the above-mentioned transaction risk factor prediction model to obtain the above-mentioned trained transaction risk factor prediction model includes:

[0022] Using the initial sample transaction risk factor prediction value and the transaction risk factor true value, adjusting the model parameters of the support vector machine module to obtain the support vector machine module in the trained transaction risk factor prediction model; and

[0023] While keeping the model parameters of the support vector machine module in the above-mentioned trained transaction risk factor prediction model unchanged, the model parameters of the above-mentioned radial basis function neural network module are adjusted using the above-mentioned sample transaction risk factor prediction value and the above-mentioned transaction risk factor true value to obtain the radial basis function neural network module in the above-mentioned trained transaction risk factor prediction model.

[0024] According to an embodiment of the present disclosure, the radial basis function neural network module in the above-mentioned transaction risk factor prediction model includes an input layer, a hidden layer and an output layer, and the kernel function of the above-mentioned hidden layer includes a Gaussian function.

[0025] According to an embodiment of the present disclosure, the radial basis function neural network module in the trained transaction risk factor prediction model is obtained by adjusting the model parameters of the radial basis function neural network module using the sample transaction risk factor prediction value and the transaction risk factor true value. The radial basis function neural network module includes:

[0026] Determine the central parameter of the kernel function according to the orthogonal least squares method, wherein the central parameter includes the number of central nodes of the hidden layer and the position parameter of the central node;

[0027] According to the above center parameter, the variance parameter of the above kernel function is determined;

[0028] Determine the weight parameters between the hidden layer and the output layer according to the least mean square algorithm and the gradient descent method;

[0029] Determine whether an error parameter between the actual value of the transaction risk factor and the predicted value of the sample transaction risk factor meets a preset condition according to at least one of the mean absolute error, the mean square error, the mean absolute percentage error and the root mean square error; and

[0030] The model obtained when the above error parameters meet the preset conditions is determined as the radial basis function neural network module in the above trained transaction risk factor prediction model.

[0031] According to an embodiment of the present disclosure, the above method further includes:

[0032] In response to the error parameter not satisfying the preset condition, updating the target learning rate of the radial basis function neural network module;

[0033] According to the target learning rate, update the predicted value of the risk factor of the sample transaction; and

[0034] According to the true value of the above-mentioned transaction risk factor and the updated sample risk factor prediction value, the updated error parameter is determined.

[0035] According to another aspect of the present disclosure, a device for predicting transaction risk factors is provided, comprising:

[0036] an acquisition module, configured to, in response to receiving a transaction risk factor prediction request, acquire transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include a concentration risk factor, a risk weight, and an initial sensitivity;

[0037] a first processing module, configured to input the transaction risk factor prediction parameter into a support vector machine module in a trained transaction risk factor prediction model, and output an initial transaction risk factor prediction value; and

[0038] The second processing module is used to input the above-mentioned initial transaction risk factor prediction value into the radial basis function neural network module in the trained transaction risk factor prediction model, and output the transaction risk factor prediction value.

[0039] According to another aspect of the present disclosure, there is provided an electronic device, including:

[0040] one or more processors;

[0041] a memory for storing one or more instructions,

[0042] When the one or more instructions are executed by the one or more processors, the one or more processors implement the method as described above.

[0043] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which executable instructions are stored. When the executable instructions are executed by a processor, the processor implements the method described above.

[0044] According to another aspect of the present disclosure, a computer program product is provided. The computer program product includes computer executable instructions. When the computer executable instructions are executed, they are used to implement the method described above.

[0045] According to the embodiments of the present disclosure, by using a transaction risk factor prediction model including a support vector machine and a radial basis function neural network, the concentration risk factor, risk weight and initial sensitivity of the transaction data to be predicted are processed to obtain a transaction risk factor prediction value. Through the above technical means, the historical limitations of transaction risk factors caused by manual calculations in related technologies are at least partially overcome, and the technical problem that the accuracy of prediction cannot be guaranteed in the environment of rapidly changing financial markets is overcome. Based on artificial intelligence technology, the prediction method of transaction risk factors is optimized by using a trained transaction risk factor prediction model, thereby improving the prediction efficiency of transaction risk factors and ensuring the accuracy of transaction risk factor prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0047] Figure 1A system architecture diagram schematically shows a method for predicting transaction risk factors that can be applied according to an embodiment of the present disclosure;

[0048] Figure 2 A flowchart schematically shows a method for predicting transaction risk factors according to an embodiment of the present disclosure;

[0049] Figure 3 A flowchart schematically shows a method for training a transaction risk factor prediction model according to an embodiment of the present disclosure;

[0050] Figure 4 A flowchart schematically shows a method for training a transaction risk factor prediction model according to another embodiment of the present disclosure;

[0051] Figure 5 A flowchart of a method for determining a transaction risk factor prediction model according to an embodiment of the present disclosure is schematically shown;

[0052] Figure 6 A flowchart of a method for determining whether an error parameter meets a preset condition according to an embodiment of the present disclosure is schematically shown;

[0053] Figure 7 The structure block diagram of the prediction device of transaction risk factor according to the embodiment of the present disclosure is schematically shown;

[0054] Figure 8 A block diagram schematically shows a structure of a training device for a transaction risk factor prediction model according to an embodiment of the present disclosure; and

[0055] Fig. 9 A block diagram of an electronic device suitable for implementing a method for predicting a transaction risk factor according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0056] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0057] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0058] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0059] In the case of using expressions such as "at least one of A, B, and C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.). In the case of using expressions such as "at least one of A, B, or C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, or C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0060] In the technical solution disclosed in the present invention, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations, necessary confidentiality measures are taken, and do not violate public order and good morals.

[0061] In the technical solution of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0062] With the development of artificial intelligence technology, research on risk factor optimization calculation based on relationship networks can obtain more reliable risk factor calculation results, thereby providing decision makers with more convincing data support.

[0063] The calculation of the traditional initial margin involves a variety of risk factors, and the calculation methods of these risk factors are relatively primitive. In the prior art, the Heston stochastic volatility model is generally used to obtain the historical volatility, and the relevant parameters required for calculating the weighted sensitivity are simulated under a 99% confidence interval.

[0064] However, the parameters obtained by using the Heston model only refer to the historical fluctuations of 10 days, while in fact the relevant parameters are usually updated once a quarter or half a year. The Heston model cannot meet the needs of the rapidly changing financial market and cannot be frequently corrected through the changes in historical data risks. Moreover, the correction of the sub-curve correlation and term correlation parameters involved is generally completed through manual updates by developers. From the release of parameters to the correction of parameters, and then to the calculation of parameters for margin in production, the overall trend of financial derivatives is not accurately judged, which can easily lead to errors in the calculation of margin, thereby causing potential credit risks.

[0065] In order to at least partially solve the technical problems existing in the related art, the present disclosure provides a transaction risk factor prediction method, device, electronic device and storage medium, which can be applied to the field of artificial intelligence technology and the financial field. The transaction risk factor prediction method includes: in response to receiving a transaction risk factor prediction request, obtaining transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include concentration risk factor, risk weight and initial sensitivity; inputting the empirical risk factor prediction parameters corresponding to the transaction data to be predicted into the support vector machine module in the trained transaction risk factor prediction model, and outputting the initial transaction risk factor prediction value; inputting the initial transaction risk factor prediction value into the radial basis function neural network module in the trained transaction risk factor prediction model, and outputting the transaction risk factor prediction value.

[0066] It should be noted that the transaction risk factor prediction method and device provided in the embodiments of the present disclosure can be used in the field of artificial intelligence technology and the financial field, for example, it can be applied to the risk factor calculation of the margin. The transaction risk factor prediction method and device provided in the embodiments of the present disclosure can also be used in any field other than the field of artificial intelligence technology and the financial field, for example, it can be applied to insurance claims. The application field of the transaction risk factor prediction method and device provided in the embodiments of the present disclosure is not limited.

[0067] Figure 1 The system architecture diagram of the prediction method of transaction risk factors according to the embodiment of the present disclosure is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0068] like Figure 1As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0069] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software, etc. (only as examples).

[0070] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0071] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0072] It should be noted that the transaction risk factor prediction method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the transaction risk factor prediction device provided in the embodiment of the present disclosure can generally be set in the server 105. The transaction risk factor prediction method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Correspondingly, the transaction risk factor prediction device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Alternatively, the transaction risk factor prediction method provided in the embodiment of the present disclosure can also be executed by the terminal devices 101, 102, or 103, or can also be executed by other terminal devices different from the terminal devices 101, 102, or 103. Accordingly, the transaction risk factor prediction device provided in the embodiment of the present disclosure may also be disposed in the terminal device 101 , 102 , or 103 , or in other terminal devices different from the terminal device 101 , 102 , or 103 .

[0073] For example, the training sample data and / or the transaction risk factor prediction parameters may be originally stored in any one of the terminal devices 101, 102, or 103 (for example, the terminal device 101, but not limited thereto), or stored in an external storage device and may be imported into the terminal device 101. Then, the terminal device 101 may locally execute the transaction risk factor prediction method provided by the embodiment of the present disclosure, or send the training sample data and / or the transaction risk factor prediction parameters to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the training sample data and / or the transaction risk factor prediction parameters may execute the transaction risk factor prediction method provided by the embodiment of the present disclosure.

[0074] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0075] Figure 2 The flowchart of the method for predicting transaction risk factors according to an embodiment of the present disclosure is schematically shown.

[0076] like Figure 2 As shown, the transaction risk factor prediction method includes operations S201 to S203.

[0077] In operation S201, in response to receiving a transaction risk factor prediction request, a transaction risk factor prediction parameter corresponding to the transaction data to be predicted in the transaction risk factor prediction request is obtained. The transaction risk factor prediction parameter includes a concentration risk factor, a risk weight, and an initial sensitivity.

[0078] In operation S202, the transaction risk factor prediction parameters are input into a support vector machine module in a trained transaction risk factor prediction model, and an initial transaction risk factor prediction value is output.

[0079] In operation S203, the initial transaction risk factor prediction value is input into the radial basis function neural network module in the trained transaction risk factor prediction model, and the transaction risk factor prediction value is output.

[0080] According to an embodiment of the present disclosure, the transaction risk factor prediction request may be a request message sent by a client to a server when a user initiates a transaction to be predicted. The transaction risk factor prediction request may include transaction data to be predicted corresponding to the transaction to be predicted.

[0081] According to an embodiment of the present disclosure, financial instruments may include financial assets that can be traded in financial markets, such as stocks, futures, gold, foreign exchange, and insurance policies. The transaction data to be predicted may include a written certificate used to prove the balance of money between the lender and the borrower of the financial instrument. For example, the transaction data to be predicted may include the transaction amount and transaction conditions.

[0082] According to an embodiment of the present disclosure, a transaction risk factor prediction parameter corresponding to the transaction data to be predicted can be calculated based on the transaction data to be predicted.

[0083] According to an embodiment of the present disclosure, the initial transaction risk factor prediction value can be obtained by calculating the transaction data to be predicted according to a formula, etc. The initial transaction risk factor prediction value can also be obtained by processing the transaction data to be predicted based on a machine learning method.

[0084] For example, the support vector machine module in the trained transaction risk factor prediction model can be used to process the concentration risk factor of the transaction data to be predicted, the risk weight of the transaction data to be predicted, and the initial sensitivity of the transaction data to be predicted to obtain the initial transaction risk factor prediction value.

[0085] According to an embodiment of the present disclosure, the radial basis function neural network module in the trained transaction risk factor prediction model can be used to process the initial transaction risk factor prediction value to obtain the transaction risk factor prediction value.

[0086] According to an embodiment of the present disclosure, the predicted value of the transaction risk factor is the fitting value of the weighted sensitivity K, and the fitting value can be substituted into the inter-currency aggregation formula of the risk type to obtain the Delta mode margin amount.

[0087] According to the embodiments of the present disclosure, by using a transaction risk factor prediction model including a support vector machine and a radial basis function neural network, the concentration risk factor, risk weight and initial sensitivity of the transaction data to be predicted are processed to obtain a transaction risk factor prediction value. Through the above technical means, the historical limitations of transaction risk factors caused by manual calculations in related technologies are at least partially overcome, and the technical problem that the accuracy of prediction cannot be guaranteed in the environment of rapidly changing financial markets is overcome. Based on artificial intelligence technology, the prediction method of transaction risk factors is optimized by using a trained transaction risk factor prediction model, thereby improving the prediction efficiency of transaction risk factors and ensuring the accuracy of transaction risk factor prediction results.

[0088] Reference below Figure 3 to Figure 6 , combined with specific embodiments Figure 2 The method shown is further explained.

[0089] According to an embodiment of the present disclosure, the method for predicting transaction risk factors may further include the following operations.

[0090] According to the predicted value of the transaction risk factor, the transaction to be predicted corresponding to the transaction data to be predicted is evaluated to obtain the margin amount of the transaction to be predicted. If the margin amount meets the preset warning conditions, a prompt message indicating that the transaction to be predicted is allowed to be executed is output. If the margin amount does not meet the preset warning conditions, a warning message indicating that the transaction to be predicted is prohibited is output.

[0091] According to an embodiment of the present disclosure, the predicted value of the transaction risk factor can be substituted into the inter-currency aggregation formula of the risk type to facilitate the evaluation of the transaction to be predicted corresponding to the transaction data to be predicted and obtain the margin amount of the Delta model.

[0092] According to the embodiments of the present disclosure, the specific content of the preset warning conditions can be flexibly set by those skilled in the art according to actual application conditions, and the embodiments of the present disclosure do not limit the specific content of the preset warning conditions.

[0093] For example, the preset warning condition may include being greater than a preset threshold. In this case, if the margin amount meets the preset warning condition, that is, the margin amount is greater than the preset threshold, a prompt message may be output, and the prompt message may be used to indicate that the predicted transaction is allowed to be executed. If the margin amount does not meet the preset warning condition, that is, the margin amount is less than or equal to the preset threshold, a warning message may be output, and the warning message may be used to indicate that the predicted transaction is prohibited to be executed.

[0094] According to the embodiments of the present disclosure, by using the transaction risk factor prediction value for the assessment of the margin amount, since the transaction risk factor prediction value is outputted via the trained transaction risk factor prediction model, the assessment accuracy of the margin amount is improved. In addition, since the output information is determined by the relationship between the margin amount of the transaction to be predicted and the preset warning condition, the output information can indicate to the relevant personnel whether the transaction to be predicted can be executed, thereby ensuring the security of the transaction.

[0095] Figure 3 A flowchart of a method for training a transaction risk factor prediction model according to an embodiment of the present disclosure is schematically shown.

[0096] like Figure 3 As shown, the training method of the prediction model of the transaction risk factor includes operations S301 to S303.

[0097] In operation S301, training sample data corresponding to sample transaction data is extracted from a source database, where the training sample data includes a sample concentration risk factor, a sample risk weight, a sample initial sensitivity, and a true value of a transaction risk factor.

[0098] According to an embodiment of the present disclosure, training sample data corresponding to sample transaction data can be stored in a source database in a persistent manner. When training a prediction model for transaction risk factors, sample data can be trained from the source database by calling a source database interface.

[0099] According to an embodiment of the present disclosure, the option margin model may include a traditional model, a Delta model, and a Span model. In the Delta model margin, the sensitivity S(i, r) of the financial instrument i to the term t of the risk-free curve r is t ) can be defined as:

[0100] S(i,r t )=V i (r t +1bp, CS t )-V i (r t , C.S. t ) (1)

[0101] Among them, S(i, r t ) The sensitivity of financial instrument i to risk factor rt; r t represents the risk-free interest rate at term t; CS t V represents the credit spread at term t; i Represents the market value of financial instrument i as a function of the risk-free rate and credit spread curve; 1bp represents 1 basis point and can include 0.0001 or 0.01%.

[0102] According to an embodiment of the present disclosure, each set of training sample data may include a sample concentration risk factor CR, a sample risk weight W K and the sample initial sensitivity S k，i .

[0103] In operation S302, the training sample data is input into the transaction risk factor prediction model, and the sample transaction risk factor prediction value is output.

[0104] According to the embodiments of the present disclosure, the sample transaction risk factor prediction value can be obtained by calculating the training sample data according to a formula, etc. The sample transaction risk factor prediction value can also be obtained by processing the training sample data based on a machine learning method.

[0105] According to an embodiment of the present disclosure, the transaction risk factor prediction model may be a model constructed based on different machine learning methods.

[0106] In operation S303, the model parameters of the transaction risk factor prediction model are adjusted using the sample transaction risk factor prediction values ​​and the transaction risk factor true values ​​to obtain a trained transaction risk factor prediction model.

[0107] According to the embodiments of the present disclosure, by processing the training sample data using the transaction risk factor prediction model, the sample transaction risk factor prediction value is output, and then the model parameters of the transaction risk factor prediction model are adjusted using the sample transaction risk factor prediction value and the corresponding transaction risk factor true value, a trained transaction risk factor prediction model is obtained. Through the above technical means, the adaptability, calculation efficiency and calculation accuracy of the transaction risk factor are improved.

[0108] Figure 4 A flowchart of a method for training a transaction risk factor prediction model according to another embodiment of the present disclosure is schematically shown.

[0109] According to an embodiment of the present disclosure, the transaction risk factor prediction model includes a support vector machine module and a radial basis function neural network module.

[0110] According to an embodiment of the present disclosure, inputting training sample data into a transaction risk factor prediction model and outputting a sample transaction risk factor prediction value may include the following operations.

[0111] The training sample data is input into the support vector machine module, and the initial sample transaction risk factor prediction value is output. The initial sample transaction risk factor prediction value is input into the radial basis function neural network module, and the sample transaction risk factor prediction value is output.

[0112] According to an embodiment of the present disclosure, using sample transaction risk factor prediction values ​​and transaction risk factor true values ​​to adjust model parameters of a transaction risk factor prediction model to obtain a trained transaction risk factor prediction model may include the following operations.

[0113] The model parameters of the support vector machine module are adjusted by using the initial sample transaction risk factor prediction value and the actual transaction risk factor value to obtain the support vector machine module in the trained transaction risk factor prediction model. While keeping the model parameters of the support vector machine module in the trained transaction risk factor prediction model unchanged, the model parameters of the radial basis function neural network module are adjusted by using the sample transaction risk factor prediction value and the actual transaction risk factor value to obtain the radial basis function neural network module in the trained transaction risk factor prediction model.

[0114] According to an embodiment of the present disclosure, a first output value can be obtained based on a first loss function using an initial sample transaction risk factor prediction value and a transaction risk factor true value. The model parameters of the support vector machine module are adjusted according to the first output value until a predetermined condition is met. The support vector machine module obtained when the predetermined condition is met is determined as the support vector machine module in the trained transaction risk factor prediction model. The specific form of the first loss function can be set by a person skilled in the art according to actual needs, and the embodiment of the present disclosure does not limit this.

[0115] According to an embodiment of the present disclosure, a second output value can be obtained based on a second loss function using a sample transaction risk factor prediction value and a transaction risk factor true value. The model parameters of the radial basis function neural network module are adjusted according to the second output value until a predetermined condition is met. The radial basis function neural network module obtained when the predetermined condition is met is determined as the radial basis function neural network module in the trained transaction risk factor prediction model. The specific form of the second loss function can be set by a person skilled in the art according to actual needs, and the embodiment of the present disclosure does not limit this.

[0116] According to the embodiments of the present disclosure, support vector machine (SVM) is a machine learning method developed on the basis of statistical learning theory, which can be used to solve the learning law of small samples. SVM avoids the traditional process from induction to deduction, realizes efficient conduction reasoning from training samples to prediction samples, simplifies conventional classification and regression problems, and better solves problems such as small samples, nonlinearity, overfitting and local minima. It has the advantages of complete mathematical theory, good global optimization performance, strong generalization ability, and algorithm complexity is independent of feature space dimension.

[0117] According to the embodiments of the present disclosure, the true value of the transaction risk factor can be obtained by calculating the training sample data according to a formula or the like.

[0118] According to an embodiment of the present disclosure, for each sample risk factor (k, i), the sample concentration risk factor CR and the sample risk weight RW of the sample risk factor k can be used. K Weighted to the sample initial sensitivity S k，i :

[0119] WS k，i =RW K S k，i CR (2)

[0120] Among them, WS k，i It represents the weighted value of the sample risk weight to the sample net sensitivity, which can be simply referred to as the sample weighted sensitivity.

[0121] According to an embodiment of the present disclosure, K may represent the k，i With further aggregation, the original business calculation formula of sample weighted sensitivity K can be:

[0122]

[0123] According to an embodiment of the present disclosure, the obtained WS k，i As the input sample of the support vector machine module The square of the sample weighted sensitivity K 2 As the dependent variable f(x) of the support vector machine module, the square of the sample weighted sensitivity is used to establish the support vector machine module:

[0124] K 2 =f(x)=wφ(x)+b (4)

[0125] Among them, w and b represent the coefficients of the maximum dividing plane; Represents the correlation φ used to simulate the sub-curve i，j and the term correlation parameter ρ k，l The weight vector of .

[0126] According to an embodiment of the present disclosure, an optimization function may be used to optimize the target value of the support vector machine module:

[0127]

[0128] Where C is a constant; ξ i * represents the original slack variable before a certain iteration; ξ i represents the slack variable after a certain iteration.

[0129] According to an embodiment of the present disclosure, the above optimization problem can be transformed into a convex quadratic optimization problem by introducing Lagrange multipliers:

[0130]

[0131] Among them, α * and γ * Is a constant, indicating the original value before a certain iteration, 0<α * , γ * <C; α and γ are variables, indicating the values ​​after a certain iteration, 0<α, γ<C.

[0132] According to the embodiment of the present disclosure, the fitting value of the weighted sensitivity can be obtained by solving the optimized convex quadratic optimization problem. It can be used as the initial sample transaction risk factor prediction value.

[0133] According to an embodiment of the present disclosure, by inputting training sample data into a support vector machine module in a risk factor prediction model, a series of initial transaction risk factor prediction values ​​that have errors with the true values ​​of the transaction risk factors can be obtained.

[0134] According to the embodiments of the present disclosure, the initial transaction risk factor prediction value can be used as an input sample, and the actual value of the transaction risk factor can be used as an output sample to train the neural network model, so as to obtain a series of node weights and thresholds for simulating the deviation relationship between the initial transaction risk factor prediction value and the actual value of the transaction risk factor, as well as the relationship between the sequences.

[0135] According to an embodiment of the present disclosure, the prediction value of the support vector machine module at the next moment or the prediction values ​​at multiple moments can be used as the input of the neural network model, and the final prediction value at the next moment or multiple moments can be used as the output of the neural network model.

[0136] According to an embodiment of the present disclosure, the neural network model may include a radial basis function neural network module. The local approximation performance of the radial basis function neural network enables the neural network to approximate any continuous function with arbitrary precision.

[0137] According to an embodiment of the present disclosure, the initial sample transaction risk factor prediction value obtained by the support vector machine module can be used as an input parameter of the radial basis function neural network module.

[0138] According to an embodiment of the present disclosure, a support vector machine module and a radial basis function neural network module may be used for combination and prediction to obtain a more optimized prediction result of the transaction risk factor, thereby optimizing the calculation of the margin.

[0139] like Figure 4 As shown, the transaction risk factor prediction model 402 may include a support vector machine module 402_1 and a radial basis function neural network module 402_2.

[0140] In the training process of the transaction risk factor prediction model, the training sample data 401 can be input into the support vector machine module 402_1 to output the initial sample transaction risk factor prediction value 403. The initial sample transaction risk factor prediction value 403 is input into the radial basis function neural network module 402_2 to output the sample transaction risk factor prediction value 404.

[0141] The model parameters of the support vector machine module 402_1 are adjusted by using the initial sample transaction risk factor prediction value 403 and the transaction risk factor true value 405 to obtain the support vector machine module in the trained transaction risk factor prediction model.

[0142] While keeping the model parameters of the support vector machine module in the trained transaction risk factor prediction model unchanged, the model parameters of the radial basis function neural network module 402_2 are adjusted using the sample transaction risk factor prediction value 404 and the transaction risk factor true value 405 to obtain the radial basis function neural network module in the trained transaction risk factor prediction model.

[0143] According to the embodiment of the present disclosure, by adjusting the model parameters of the radial basis function neural network module by using the sample transaction risk factor prediction value and the transaction risk factor true value while keeping the model parameters of the trained support vector machine module unchanged, the radial basis function neural network module in the trained transaction risk factor prediction model is obtained. Thus, the trained transaction risk factor prediction model is obtained. Since the combined model includes the trained support vector machine module and the trained radial basis function neural network module, the deviation between the transaction risk factor prediction value and the transaction risk factor true value can be reduced, and a better fitting result can be obtained, so as to improve the efficiency and accuracy of transaction risk factor prediction.

[0144] Figure 5 A flowchart of a method for determining a transaction risk factor prediction model according to an embodiment of the present disclosure is schematically shown.

[0145] like Figure 5 As shown, the method for determining a transaction risk factor prediction model includes operations S501 to S505.

[0146] In operation S501, according to the orthogonal least squares method, a central parameter of a kernel function is determined, wherein the central parameter includes the number of central nodes of a hidden layer and a position parameter of the central node.

[0147] According to an embodiment of the present disclosure, the radial basis function neural network module in the transaction risk factor prediction model includes an input layer, a hidden layer and an output layer, and the kernel function of the hidden layer includes a Gaussian function.

[0148] According to the embodiments of the present disclosure, since the Gaussian function decreases monotonically from the center to both sides, the response of the function is locally limited, and more realistic features can be obtained. Therefore, the Gaussian function can be selected as the radial basis function of the radial basis function neural network:

[0149]

[0150] According to an embodiment of the present disclosure, methods for determining the center parameter of a kernel function may include a fixed center method, a self-organizing selection method, a supervised center selection method, and the like.

[0151] According to the embodiments of the present disclosure, the fixed center method can randomly select a center from a training sample data set, which is suitable for situations where the distribution of training data is representative. However, the disadvantage is that it can easily cause the network to perform poorly or make the network volume too large.

[0152] According to an embodiment of the present disclosure, the self-organizing selection method may include a K-means clustering method, which clusters the training sample data set and uses the center of the cluster as the center of the network. The advantage of the K-means clustering method is that the center of the radial basis function neural network can be set as a more important data point, and the disadvantage is that if the center of the initial cluster is accurate, the final prediction result may be a local optimal solution.

[0153] According to the embodiments of the present disclosure, the supervised center selection method needs to determine the cost function during the network learning process, and the process of determining the center is the process of minimizing the cost function.

[0154] According to an embodiment of the present disclosure, the method for determining the central parameter of the kernel function may also include an orthogonal least squares method, and the accuracy of the output prediction value of the radial basis function neural network module may be improved through training.

[0155] In operation S502, a variance parameter of the kernel function is determined according to the center parameter.

[0156] According to the embodiments of the present disclosure, the number of nodes in the hidden layer of the radial basis function neural network can be determined first, and the central parameter c can be determined and the variance δ can be calculated by using the orthogonal least squares method. 2 .

[0157] In operation S503, a weight parameter between the hidden layer and the output layer is determined according to a least mean square algorithm and a gradient descent method.

[0158] According to the embodiments of the present disclosure, the weights can be calculated by the least mean square algorithm, and the network can be learned by the gradient descent method. The gradient descent method can gradually adjust the weights in the opposite direction of the objective function gradient to obtain the optimal value of the weights and optimize the network. The weight adjustment process can be expressed as:

[0159]

[0160] In operation S504, it is determined whether an error parameter between the actual value of the transaction risk factor and the predicted value of the sample transaction risk factor meets a preset condition according to at least one of the mean absolute error, the mean square error, the mean absolute percentage error and the root mean square error.

[0161] According to the embodiments of the present disclosure, a variety of error evaluation methods can be used to determine the error parameters. The error evaluation methods may include mean absolute error (MAE), mean square error (MSE), mean absolute percentage error (MAPE), root mean square error (RMSE), etc. The calculation methods of the above error evaluation methods are as follows:

[0162]

[0163]

[0164]

[0165]

[0166] Where N is the number of samples; x i represents the true value of the risk factor; Represents the predicted value of risk factor.

[0167] According to the embodiments of the present disclosure, by using a plurality of error evaluation methods in the fitting process, a better fitting result can be obtained, thereby preventing the data blind spot problem caused by using a single error evaluation method.

[0168] In operation S505, the model obtained when the error parameter satisfies the preset condition is determined as the radial basis function neural network module in the trained transaction risk factor prediction model.

[0169] According to the embodiments of the present disclosure, by using the orthogonal least squares method to determine the center parameter and variance parameter of the kernel function, using the least mean square algorithm and the gradient descent method to determine the weight parameters between the hidden layer and the output layer, and performing error parameter analysis on the true value of the transaction risk factor and the predicted value of the sample transaction risk factor, the optimal value of the weight parameter can be obtained, and a better fitting result can be obtained, thereby improving the prediction accuracy of the transaction risk factor.

[0170] Figure 6 A flowchart of a method for determining whether an error parameter meets a preset condition according to an embodiment of the present disclosure is schematically shown.

[0171] like Figure 6 As shown, the method for determining whether the error parameter meets the preset condition includes operations S601 to S603.

[0172] In operation S601, in response to the error parameter not satisfying a preset condition, a target learning rate of the radial basis function neural network module is updated.

[0173] In operation S602, the predicted value of the sample transaction risk factor is updated according to the target learning rate.

[0174] In operation S603, an updated error parameter is determined according to the actual value of the transaction risk factor and the updated predicted value of the sample transaction risk factor.

[0175] According to the embodiments of the present disclosure, in the use of radial basis function neural networks, the learning rate is usually artificially set to a fixed value, which remains unchanged during the entire network learning process. However, if the learning rate is set too large, although it can improve the convergence speed of the network, it may cause the network to be unstable or even unable to learn; and if the learning rate is set too small, it will cause the network to converge slowly, consume a lot of computing time, and fail to meet the timeliness requirements of practical applications.

[0176] According to an embodiment of the present disclosure, a "dynamic optimal learning rate" is provided, by which the target learning rate of the radial basis function neural network module can be updated, so that the target learning rate is applicable to each step of iterative learning of the network.

[0177] According to the embodiments of the present disclosure, the following model can be established:

[0178]

[0179] Among them, φ(||xc i ||) represents Gaussian function; ω i Represents the connection weight.

[0180] Based on the above model, the t-th cost function can be obtained, namely:

[0181]

[0182] Among them, e(t)=e(t-1)-η(t)ΦΦ T e(t-1) represents the output error; η(t) represents the learning rate; Φ represents the matrix constructed according to the input sample; Φ T represents the transposed matrix of Φ.

[0183] The above cost function can be regarded as a function of the learning rate, and the optimal value of the learning rate can be obtained by minimizing it. The second-order condition is as follows:

[0184]

[0185] Based on the above conditions, the dynamic optimal learning rate η can be obtained * (t) is:

[0186]

[0187] According to an embodiment of the present disclosure, the learning accuracy rmse can be set * and the maximum number of iterations of network learning maxt ; You can set the initial connection weight ω0 from the hidden layer to the output layer and calculate the network output value And calculate the root mean square error rmse between the actual output value and the network output value; the variance δ can be calculated based on the central parameter c 2 , find the matrix Φ.

[0188] According to the embodiment of the present disclosure, it can be judged that rmse<rmse * Is it true, or has the maximum number of iterations max been reached? t , if either of them holds true, the network learning can be completed; if neither of them holds true, the dynamic optimal learning rate can be calculated by formula (16) to obtain the target learning rate.

[0189] According to an embodiment of the present disclosure, the network output value can be updated according to the target learning rate And update the root mean square error rmse between the actual output value and the network output value, and then the final root mean square error rmse and network output value can be output And calculate the final number of iterations.

[0190] According to the embodiments of the present disclosure, when the error parameter does not meet the preset conditions, the target learning rate of the radial basis function neural network module is updated, and the predicted value of the sample transaction risk factor is updated according to the updated target learning rate. Through the above technical means, the learning rate can be optimized and calculated in each iteration step, and the dynamic optimal learning rate can be obtained in real time, so that the convergence speed of the network can be taken into account while ensuring the stable learning of the radial basis function neural network, thereby improving the operation efficiency of the network.

[0191] Figure 7 The structural block diagram of the transaction risk factor prediction device according to the embodiment of the present disclosure is schematically shown.

[0192] like Figure 7 As shown, the transaction risk factor prediction device 700 includes a first acquisition module 701 , a first processing module 702 and a second processing module 703 .

[0193] The first acquisition module 701 is used to obtain transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request in response to receiving the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include concentration risk factor, risk weight and initial sensitivity.

[0194] The first processing module 702 is used to input the transaction risk factor prediction parameters into the support vector machine module in the trained transaction risk factor prediction model, and output the initial transaction risk factor prediction value.

[0195] The second processing module 703 is used to input the initial transaction risk factor prediction value into the radial basis function neural network module in the trained transaction risk factor prediction model, and output the transaction risk factor prediction value.

[0196] According to an embodiment of the present disclosure, the transaction risk factor prediction device 700 may further include an evaluation module, a first output module and a second output module.

[0197] The evaluation module is used to evaluate the transaction to be predicted corresponding to the transaction data to be predicted according to the predicted value of the transaction risk factor, and obtain the margin amount of the transaction to be predicted.

[0198] The first output module is used to output prompt information indicating that the predicted transaction is allowed to be executed when the margin amount meets the preset warning condition.

[0199] The second output module is used to output warning information indicating that the execution of the transaction to be predicted is prohibited when the margin amount does not meet the preset warning condition.

[0200] Figure 8 The structural block diagram of the training device of the transaction risk factor prediction model according to the embodiment of the present disclosure is schematically shown.

[0201] like Figure 8 As shown, the training device 800 for the transaction risk factor prediction model includes a second acquisition module 801 , a third processing module 802 and a first training module 803 .

[0202] The second acquisition module 801 is used to extract training sample data corresponding to the sample transaction data from the source database, wherein the training sample data includes a sample concentration risk factor, a sample risk weight and a sample initial sensitivity.

[0203] The third processing module 802 inputs the training sample data into the transaction risk factor prediction model and outputs the sample transaction risk factor prediction value.

[0204] The training module 803 is used to adjust the model parameters of the transaction risk factor prediction model by using the sample transaction risk factor prediction values ​​and the transaction risk factor true values ​​to obtain a trained transaction risk factor prediction model.

[0205] According to an embodiment of the present disclosure, the transaction risk factor prediction model includes a support vector machine module and a radial basis function neural network module.

[0206] According to an embodiment of the present disclosure, the third processing module 802 includes a first processing unit and a second processing unit.

[0207] The first processing unit is used to input the training sample data into the support vector machine module and output the initial sample transaction risk factor prediction value.

[0208] The second processing unit is used to input the initial sample transaction risk factor prediction value into the radial basis function neural network module and output the sample transaction risk factor prediction value.

[0209] According to an embodiment of the present disclosure, the training module 803 includes a first adjustment unit and a second adjustment unit.

[0210] The first adjustment unit is used to adjust the model parameters of the support vector machine module by using the initial sample transaction risk factor prediction value and the transaction risk factor true value to obtain the support vector machine module in the trained transaction risk factor prediction model.

[0211] The second adjustment unit is used to adjust the model parameters of the radial basis function neural network module by using the sample transaction risk factor prediction value and the transaction risk factor true value while keeping the model parameters of the support vector machine module in the trained transaction risk factor prediction model unchanged, so as to obtain the radial basis function neural network module in the trained transaction risk factor prediction model.

[0212] According to an embodiment of the present disclosure, the radial basis function neural network module in the transaction risk factor prediction model includes an input layer, a hidden layer and an output layer, and the kernel function of the hidden layer includes a Gaussian function.

[0213] According to an embodiment of the present disclosure, the second adjustment unit includes a first determining subunit, a second determining subunit, a third determining subunit, a fourth determining subunit, and a fifth determining subunit.

[0214] The first determination subunit is used to determine the central parameter of the kernel function according to the orthogonal least squares method, wherein the central parameter includes the number of central nodes of the hidden layer and the position parameter of the central node.

[0215] The second determination subunit is used to determine the variance parameter of the kernel function according to the center parameter.

[0216] The third determination subunit is used to determine the weight parameters between the hidden layer and the output layer according to the least mean square algorithm and the gradient descent method.

[0217] The fourth determination subunit is used to determine whether the error parameter between the true value of the transaction risk factor and the predicted value of the sample transaction risk factor meets the preset conditions based on at least one of the mean absolute error, the mean square error, the mean absolute percentage error and the root mean square error.

[0218] The fifth determination subunit is used to determine the model obtained when the error parameter meets the preset conditions as the radial basis function neural network module in the trained transaction risk factor prediction model.

[0219] According to an embodiment of the present disclosure, the second adjusting unit may further include a first updating subunit, a second updating subunit and a first determining subunit.

[0220] The first updating subunit is used to update the target learning rate of the radial basis function neural network module in response to the error parameter not satisfying the preset condition.

[0221] The second updating subunit is used to update the predicted value of the sample transaction risk factor according to the target learning rate.

[0222] The sixth determination subunit is used to determine the updated error parameter according to the actual value of the transaction risk factor and the updated predicted value of the sample transaction risk factor.

[0223] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.

[0224] For example, any multiple of the first acquisition module 701, the first processing module 702, the second processing module 703, the second acquisition module 801, the third processing module 802, and the training module 803 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the first acquisition module 701, the first processing module 702, the second processing module 703, the second acquisition module 801, the third processing module 802 and the training module 803 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in an appropriate combination of any of them. Alternatively, at least one of the first acquisition module 701, the first processing module 702, the second processing module 703, the second acquisition module 801, the third processing module 802 and the training module 803 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.

[0225] It should be noted that the prediction device part of the transaction risk factor in the embodiment of the present disclosure corresponds to the prediction method part of the transaction risk factor in the embodiment of the present disclosure. The description of the prediction device part of the transaction risk factor specifically refers to the prediction method part of the transaction risk factor, which will not be repeated here. The training device part of the transaction risk factor prediction model in the embodiment of the present disclosure corresponds to the training method part of the transaction risk factor prediction model in the embodiment of the present disclosure. The description of the training device part of the transaction risk factor prediction model specifically refers to the training method part of the transaction risk factor prediction model, which will not be repeated here.

[0226] Fig. 9 A block diagram of an electronic device suitable for implementing a method for predicting a transaction risk factor according to an embodiment of the present disclosure is schematically shown. Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0227] like Fig. 9As shown, the computer electronic device 900 according to the embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 902 or the program loaded from the storage part 908 to the random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.

[0228] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0229] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the I / O interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0230] According to an embodiment of the present disclosure, the method flow according to an embodiment of the present disclosure can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed. According to an embodiment of the present disclosure, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0231] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0232] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0233] For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 902 and / or the RAM 903 described above and / or one or more memories other than the ROM 902 and the RAM 903 .

[0234] An embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method provided by the embodiment of the present disclosure. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the method for predicting transaction risk factors provided by the embodiment of the present disclosure.

[0235] When the computer program is executed by the processor 901, the above functions defined in the system / device of the embodiment of the present disclosure are executed. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0236] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0237] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0238] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments and / or claims of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0239] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for predicting a transaction risk factor, comprising: In response to receiving a transaction risk factor prediction request, obtaining transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include a concentration risk factor, the risk weight, and an initial sensitivity; Inputting the transaction risk factor prediction parameters into a support vector machine module in a trained transaction risk factor prediction model, and outputting an initial transaction risk factor prediction value; and Inputting the initial transaction risk factor prediction value into the radial basis function neural network module in the trained transaction risk factor prediction model, and outputting the transaction risk factor prediction value; The training method of the trained transaction risk factor prediction model includes: extracting training sample data corresponding to sample transaction data from a source database, wherein the training sample data includes a sample concentration risk factor, a sample risk weight, a sample initial sensitivity and a true value of a transaction risk factor; inputting the training sample data into the transaction risk factor prediction model, and outputting a sample transaction risk factor prediction value; and adjusting the model parameters of the transaction risk factor prediction model using the sample transaction risk factor prediction value and the true value of the transaction risk factor to obtain the trained transaction risk factor prediction model; The step of inputting the training sample data into the transaction risk factor prediction model and outputting the sample transaction risk factor prediction value comprises: inputting the training sample data into the support vector machine module and outputting the initial sample transaction risk factor prediction value; and inputting the initial sample transaction risk factor prediction value into the radial basis function neural network module and outputting the sample transaction risk factor prediction value. The method of using the sample transaction risk factor prediction value and the transaction risk factor true value to adjust the model parameters of the transaction risk factor prediction model to obtain the trained transaction risk factor prediction model includes: using the initial sample transaction risk factor prediction value and the transaction risk factor true value to adjust the model parameters of the support vector machine module to obtain the support vector machine module in the trained transaction risk factor prediction model; and, while keeping the model parameters of the support vector machine module in the trained transaction risk factor prediction model unchanged, using the sample transaction risk factor prediction value and the transaction risk factor true value to adjust the model parameters of the radial basis function neural network module to obtain the radial basis function neural network module in the trained transaction risk factor prediction model.

2. The method according to claim 1, further comprising: According to the predicted value of the transaction risk factor, the transaction to be predicted corresponding to the transaction data to be predicted is evaluated to obtain a margin amount of the transaction to be predicted; When the margin amount meets the preset warning condition, outputting prompt information indicating that the predicted transaction is allowed to be executed; as well as When the margin amount does not satisfy the preset warning condition, a warning message indicating that the execution of the transaction to be predicted is prohibited is output.

3. The method according to claim 1, wherein: The radial basis function neural network module in the transaction risk factor prediction model includes an input layer, a hidden layer and an output layer, and the kernel function of the hidden layer includes a Gaussian function; The step of adjusting the model parameters of the radial basis function neural network module by using the sample transaction risk factor prediction value and the transaction risk factor true value to obtain the radial basis function neural network module in the trained transaction risk factor prediction model comprises: Determine the central parameter of the kernel function according to the orthogonal least squares method, wherein the central parameter includes the number of central nodes of the hidden layer and the position parameter of the central node; Determining a variance parameter of the kernel function according to the center parameter; Determine the weight parameters between the hidden layer and the output layer according to the least mean square algorithm and the gradient descent method; Determining whether an error parameter between the true value of the transaction risk factor and the predicted value of the sample transaction risk factor meets a preset condition according to at least one of a mean absolute error, a mean square error, a mean absolute percentage error, and a root mean square error; and The model obtained when the error parameter meets the preset conditions is determined as the radial basis function neural network module in the trained transaction risk factor prediction model.

4. The method according to claim 3, further comprising: In response to the error parameter not satisfying the preset condition, updating the target learning rate of the radial basis function neural network module; According to the target learning rate, updating the predicted value of the sample transaction risk factor; as well as An updated error parameter is determined based on the actual value of the transaction risk factor and the updated predicted value of the sample transaction risk factor.

5. A device for predicting transaction risk factors, comprising: A first acquisition module is configured to, in response to receiving a transaction risk factor prediction request, acquire transaction risk factor prediction parameters corresponding to the transaction data to be predicted in the transaction risk factor prediction request, wherein the transaction risk factor prediction parameters include a concentration risk factor, a risk weight, and an initial sensitivity; a first processing module, configured to input the transaction risk factor prediction parameter into a support vector machine module in a trained transaction risk factor prediction model, and output an initial transaction risk factor prediction value; and A second processing module is used to input the initial empirical risk factor prediction value into the radial basis function neural network module in the trained transaction risk factor prediction model, and output the transaction risk factor prediction value; Wherein, the trained transaction risk factor prediction model is obtained by training using a transaction risk factor prediction model training device, and the transaction risk factor prediction model training device includes: a second acquisition module, used to extract training sample data corresponding to sample transaction data from a source database, wherein the training sample data includes a sample concentration risk factor, a sample risk weight, a sample initial sensitivity and a true value of a transaction risk factor; a third processing module, which inputs the training sample data into the transaction risk factor prediction model and outputs a sample transaction risk factor prediction value; and a training module, which uses the sample transaction risk factor prediction value and the transaction risk factor true value to adjust the model parameters of the transaction risk factor prediction model to obtain the trained transaction risk factor prediction model; The third processing module includes: a first processing unit, which is used to input the training sample data into the support vector machine module and output the initial sample transaction risk factor prediction value; and a second processing unit, which is used to input the initial sample transaction risk factor prediction value into the radial basis function neural network module and output the sample transaction risk factor prediction value; The training module includes: a first adjustment unit, which is used to adjust the model parameters of the support vector machine module by using the initial sample transaction risk factor prediction value and the transaction risk factor true value, so as to obtain the support vector machine module in the trained transaction risk factor prediction model; and a second adjustment unit, which is used to adjust the model parameters of the radial basis function neural network module by using the sample transaction risk factor prediction value and the transaction risk factor true value while keeping the model parameters of the support vector machine module in the trained transaction risk factor prediction model unchanged, so as to obtain the radial basis function neural network module in the trained transaction risk factor prediction model.

6. An electronic device comprising: one or more processors; A memory for storing one or more instructions, Wherein, when the one or more instructions are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having executable instructions stored thereon, wherein when the executable instructions are executed by a processor, the processor is enabled to implement the method according to any one of claims 1 to 4.

8. A computer program product, comprising computer executable instructions, wherein the computer executable instructions are used to implement the method according to any one of claims 1 to 4 when executed.

Images