Risk data disposal methods, devices, equipment, media and program products

Abstract

This disclosure provides a method, apparatus, device, medium, and program product for handling risk data, which can be applied to the field of artificial intelligence, specifically involving risk data in the financial sector. The risk data handling method includes: acquiring historical risk data and corresponding handling methods; inputting the historical risk data and corresponding handling methods into a semi-supervised convolutional generative adversarial network (GAN) for training to obtain a first GAN model; using the GAN model to process the current risk data to be processed, obtaining the handling method with the highest matching degree to the current risk data; training the first GAN model with the current risk data to be processed and the corresponding handling method with the highest matching degree to obtain a second GAN model; and using the second GAN model to process the next risk data to be processed, obtaining the handling method with the highest matching degree to the next risk data to be processed.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence, specifically to risk data in the financial field, and more specifically to a risk data processing method, device, equipment, medium and program product. Background Art

[0002] In the current risk data disposal module of the risk control system of banks and merchants, it is necessary to go through the stages of risk data access, task allocation, task disposal, etc., and finally the assigned teller will handle the task. The whole process is relatively complicated. The most important disposal stage is divided into manual disposal and intelligent disposal. Manual disposal requires business personnel to manually identify the characteristics of risk data and determine which disposal measures are needed. Intelligent disposal requires business personnel to configure business lists and risk strategies of different models by themselves. These risk strategies are often based on different feature values ​​set for different business scenarios. When transaction data hits these lists or model strategies (feature values), the system automatically takes corresponding disposal measures for these transactions.

[0003] Since the feature identification of risk data itself is relatively complex and involves many dimensions, traditional intelligent disposal requires the configuration of a large number of model strategies. Different business scenarios require the configuration of different rules, which is a large workload and difficult. At present, in order to ensure the accuracy of intelligent disposal, business rules are generally relatively simple. Generally, a few characteristic dimensions of risk data match the configured characteristic rules, which means a hit, and then corresponding disposal measures are taken. For example, for merchant acquisition risk data, if the merchant's corresponding acquisition area is overseas, the merchant will be "deactivated". Therefore, the existing intelligent disposal has defects such as low versatility, simple rules, and low coverage. Most scenarios still use manual disposal. The intelligent risk disposal stage urgently needs to be replaced by more efficient technologies. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a risk data handling method, apparatus, device, medium and program product.

[0005] According to the first aspect of the present disclosure, a risk data handling method is provided, including: obtaining historical risk data and corresponding handling methods; inputting the historical risk data and the corresponding handling methods into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model; using the generative adversarial network model to process the current risk data to be processed to obtain the handling method with the highest degree of match with the current risk data to be processed; training the first generative adversarial network model with the current risk data to be processed and the corresponding handling method with the highest degree of match to obtain a second generative adversarial network model; using the second generative adversarial network model to process the next risk data to be processed to obtain the handling method with the highest degree of match with the next risk data to be processed.

[0006] According to an embodiment of the present disclosure, historical risk data and corresponding disposal methods are input into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model, including: traversing historical risk data and corresponding disposal methods, dividing the training into 12 cycles according to 12 months per year, and dividing the cycles into several batches according to the card type dimension; training the generative adversarial network model according to cycles and batches; counting the accuracy of the training recognition results of each batch, sampling the training recognition results of the batches in the same cycle, running the optimizer to update the model parameters, and obtaining the first generative adversarial network model.

[0007] According to an embodiment of the present disclosure, the risk data to be processed is processed using a generative adversarial network model to obtain a treatment method that best matches the risk data to be processed, including: generating a set including risk data and treatment methods; smoothing the set; processing the risk data to be processed using a back propagation method, comparing the eigenvalues ​​generated by each iteration with the eigenvalues ​​of the original input risk data to be processed, and obtaining a treatment method that best matches the eigenvalues.

[0008] According to an embodiment of the present disclosure, generating a set including risk data and treatment methods includes: generating a set including risk data and treatment methods under the restriction condition of a Relu activation function.

[0009] According to the embodiment of the present disclosure, the back propagation method is used to process the risk data to be processed, and the treatment method with the highest matching degree includes: generating random noise G(z0); traversing the set, iterating and updating it to G(z0) under the constraint of the loss function through n iterations. n ), G(z n ) is the disposal method with the highest matching degree.

[0010] According to an embodiment of the present disclosure, the risk data handling method also includes constructing a semi-supervised convolutional generative adversarial network, including: using the first softmax classifier as the output layer of the generator; using the second softmax classifier as the output layer of the discriminator; wherein the second softmax classifier has at least one more output neuron than the first softmax classifier.

[0011] According to an embodiment of the present disclosure, the first generative adversarial network model adopts an ADAM optimizer.

[0012] According to an embodiment of the present disclosure, obtaining historical risk data and corresponding handling methods includes: using a kafka message queue to collect historical risk data and corresponding handling methods.

[0013] According to an embodiment of the present disclosure, obtaining historical risk data and corresponding disposal methods further includes: cleaning the historical risk data and corresponding disposal methods to obtain pre-processed risk data and corresponding disposal methods that meet preset standards.

[0014] A risk data handling device is provided, including: a collection module, used to obtain historical risk data and corresponding handling methods; a first training module, used to input the historical risk data and the corresponding handling methods into a semi-supervised convolutional generative adversarial network for training, to obtain a first generative adversarial network model; a first processing module, used to use the generative adversarial network model to process the current risk data to be processed, to obtain the handling method with the highest degree of match with the current risk data to be processed; a second training module, used to train the first generative adversarial network model with the current risk data to be processed and the corresponding handling method with the highest degree of match, to obtain a second generative adversarial network model; a second processing module, used to use the second generative adversarial network model to process the next risk data to be processed, to obtain the handling method with the highest degree of match with the next risk data to be processed.

[0015] The third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned risk data handling method.

[0016] The fourth aspect of the present disclosure also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned risk data handling method.

[0017] The fifth aspect of the present disclosure also provides a computer program product, including a computer program, which implements the above-mentioned risk data handling method when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above contents 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:

[0019] Figure 1 The application scenario diagram of the risk data handling method, apparatus, device, medium and program product according to the embodiment of the present disclosure is schematically shown;

[0020] Figure 2 A flowchart of a risk data handling method according to an embodiment of the present disclosure is schematically shown;

[0021] Figure 3The training flow chart of the first generative adversarial network model according to the embodiment of the present disclosure is schematically shown;

[0022] Figure 4 A flowchart of processing risk data to be processed currently according to an embodiment of the present disclosure is schematically shown;

[0023] Figure 5 The back propagation method processing flow chart according to the embodiment of the present disclosure is schematically shown;

[0024] Figure 6 A schematic diagram of a structure of a risk data handling device according to an embodiment of the present disclosure is shown; and

[0025] Figure 7 A block diagram of an electronic device suitable for implementing a risk data handling method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0026] 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.

[0027] 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.

[0028] 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.

[0029] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly 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.).

[0030] In the current risk data processing module of the risk control system of banks and merchants, it is necessary to go through the stages of risk data access, task allocation, task processing, etc., and finally the assigned teller will process the task. The whole process is relatively complicated. Since the feature identification of risk data itself is relatively complex and involves many dimensions, the existing intelligent processing has defects such as low versatility, simple rules, and low coverage. Most scenarios still use manual processing. The intelligent risk processing stage urgently needs to be replaced by more efficient technology.

[0031] In view of the above-mentioned problems, the embodiments of the present disclosure provide a risk data disposal method to solve the problems of traditional task disposal, such as complex processes, low versatility, and simple rules. The present disclosure is based on a deep convolutional generative adversarial network to perform intelligent identification and disposal of risk data. In the embodiments of the present disclosure, based on the historical risk data and disposal methods recorded in the risk monitoring system, as the input of the model, a semi-supervised convolutional generative adversarial network is used to extract features and train the model to obtain a generative adversarial network model. After the current risk data to be processed is transmitted from the upstream to the risk monitoring system, it is input into the generative adversarial network model for processing, and a back propagation method is used to iteratively generate a set of risk data and disposal methods, and the set is smoothed. In the iterative process, the characteristic value of the current risk data to be processed is compared with the characteristic value of the iteratively generated risk data, and the difference in the characteristic value is continuously reduced, and finally the disposal method with the highest matching degree is obtained, which is the final disposal measure. At the same time, the current risk data to be processed and the disposal method with the highest matching degree are used as the input of the generative adversarial network model, and the real-time generated data are used to train the generative adversarial network model in real time to obtain a real-time trained generative adversarial network model, and the real-time trained generative adversarial network model is used to dispose of the next risk data to be processed.

[0032] The present disclosure relates to the field of artificial intelligence, and specifically to risk data in the financial field. It should be noted that the risk data processing method and device provided by the present disclosure can be used for processing risk data in the financial field, and can also be used for processing risk data in any field other than the financial field. The application field of the risk data processing method and device provided by the present disclosure is not limited.

[0033] In the technical solution disclosed in the present invention, the acquisition, storage and application of the risk data 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.

[0034] Figure 1 The application scenario diagram of the risk data handling method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0035] like Figure 1As shown, the application scenario 100 according to this embodiment may include a bank risk data handling system. A network 102 is used to provide a medium for a communication link between a terminal device 101 and a server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0036] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send risk data or a disposal method corresponding to the risk data, etc. The terminal device 101 can be installed with a financial application for sending risk data.

[0037] The terminal device 101 may be any electronic device with interactive functions, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, and the like.

[0038] The server 103 may be a server that provides risk data processing, for example, inputting the risk data generated by the user using the terminal device 101 into the server 103 for risk data processing, and receiving the corresponding processing method sent back by the server 103. For example, the user uses the financial software of the terminal device 101 to generate corresponding risk data, which is sent to the server 103 through the network 102 for processing, and the processing method that best matches the risk data is obtained. Depending on the different processing methods, it is possible to choose to send or not send the information corresponding to the processing method to the terminal device 101.

[0039] It should be noted that the risk data handling method provided in the embodiment of the present disclosure can generally be executed by the server 103. Accordingly, the risk data handling device provided in the embodiment of the present disclosure can generally be set in the server 103. The risk data handling 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 103 and can communicate with the terminal device 101 and / or the server 103. Accordingly, the risk data handling 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 103 and can communicate with the terminal device 101 and / or the server 103.

[0040] 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.

[0041] The following will be based on Figure 1 The scene described by Figure 2 to Figure 5 And related introductions describe in detail the risk data handling method of the disclosed embodiment.

[0042] Figure 2The flowchart of the risk data handling method according to the embodiment of the present disclosure is schematically shown.

[0043] like Figure 2 As shown, the risk data handling method of this embodiment includes operations S201 to S205.

[0044] In operation S201, historical risk data and corresponding disposal methods are obtained.

[0045] In operation S202, the historical risk data and the corresponding treatment methods are input into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model.

[0046] In operation S203, the risk data to be processed is processed using the generative adversarial network model to obtain a treatment method with the highest matching degree with the risk data to be processed.

[0047] In operation S204, the first generative adversarial network model is trained with the current risk data to be processed and the corresponding treatment method with the highest matching degree to obtain a second generative adversarial network model.

[0048] In operation S205, the next risk data to be processed is processed using the second generative adversarial network model to obtain a disposal method with the highest matching degree with the next risk data to be processed.

[0049] In a specific embodiment of the present application, historical risk data including risk data and disposal methods of cloud-based Peony Card acquiring merchants is obtained, and the risk data and disposal methods of the cloud-based Peony Card acquiring merchants are collected using a Kafka message queue, and redundant fields such as duplications, carriage returns, line feeds, spaces, tabs, etc. are deleted, and dirty data such as missing fields and inconsistent case are processed. Outliers are filled with the mean to obtain pre-processed risk data and disposal methods of cloud-based Peony Card acquiring merchants.

[0050] The peony card data and the corresponding disposal method can form a one-to-one corresponding two-dimensional matrix. The two-dimensional matrix is ​​input into the semi-supervised convolutional generative adversarial network for training, which is divided into 12 cycles and recursively trained to obtain the trained generative adversarial network model, that is, the first generative adversarial network model.

[0051] A risk transaction data of the Peony quasi-credit card dimension is input to the first generative adversarial network model, and the first generative adversarial network model outputs the transaction disposal method. For example, if the risk transaction data of the Peony quasi-credit card dimension has the highest matching degree with the historical risk data and the historical risk data with the disposal method "", then "add card to blacklist" is the disposal method with the highest matching degree with the risk transaction data of the Peony quasi-credit card dimension. The disposal method is executed by the business personnel.

[0052] At the same time, the risk transaction data of the Peony quasi-credit card dimension and the corresponding "add to card blacklist" disposal method are used to train the first generative adversarial network model. By training the generative adversarial network model in real time with real-time processed data, a second generative adversarial network model updated in real time is obtained, which realizes the characteristics of versatility, high coverage, and high rule dimension of the generative adversarial network model of this application. The strategy based on the generative adversarial network model that can realize real-time training is not targeted at a single risk dimension. The training model extracts and learns feature variables by itself, without the need to configure for different business scenarios, and all disposal scenarios can be covered.

[0053] Figure 3 The training flowchart of the first generative adversarial network model according to an embodiment of the present disclosure is schematically shown.

[0054] like Figure 3 As shown, the training process of the first generative adversarial network model of this embodiment includes operations S301 to S303.

[0055] In operation S301, the historical risk data and the corresponding disposal methods are traversed, and the training is divided into several cycles, each of which contains several batches. Optionally, the training is divided into 12 cycles according to the 12 months of each year, and the cycles are divided into several batches according to the card type dimension; in fact, the card types include cards corresponding to different types of accounts such as Peony Quasi-Credit Card, Peony Special Card, and Peony Money Link Card. By training the generative adversarial network model by cycle and batch, feature clustering can be performed on risk transactions of different times and different cards, and the feature values ​​generated during the training process are more representative.

[0056] In operation S302 , training of a generative adversarial network model is performed in cycles and batches.

[0057] In operation S303, the accuracy of several batches of training recognition results is counted, several batches of training recognition results in the same cycle are sampled, and the optimizer is run to update the model parameters to obtain a first generative adversarial network model.

[0058] Figure 4 The flowchart of processing risk data to be processed currently according to an embodiment of the present disclosure is schematically shown.

[0059] like Figure 4 As shown, the processing flow of the risk data to be processed currently in this embodiment includes operations S401 to S403.

[0060] In operation S401 , a set including risk data and treatment methods is generated.

[0061] In operation S40, a smoothing process is performed on the set.

[0062] In operation S40, the back propagation method is used to process the risk data to be processed, and the characteristic values ​​generated in each iteration are compared with the characteristic values ​​of the original input risk data to be processed to obtain a disposal method with the highest characteristic value matching degree.

[0063] In an optional solution of the first embodiment, generating a set including risk data and disposal methods includes: using a Relu activation function to allow a generator of a generative adversarial network model to generate a set including risk data and disposal methods under the restriction of the Relu activation function. Using the Relu activation function can reduce the amount of computation of the generative adversarial network model, increase the computation speed, and enhance the user experience.

[0064] Figure 5 The flowchart of the back propagation method according to the embodiment of the present disclosure is schematically shown.

[0065] like Figure 5 As shown, the back propagation method processing flow of this embodiment includes operations S501-S502.

[0066] In operation S501, random noise G(z0) is generated.

[0067] In operation S502, the set is traversed and updated to G(z) through n iterations under the constraint of the loss function. n ), G(z n ) is the disposal method with the highest matching degree.

[0068] In this embodiment of the present disclosure, a random noise G(z0) is defined, and then back propagation is used to traverse the risk-treatment set that meets the requirements, and G(z0) is iteratively updated under the constraint of the loss function. n )) until a certain number of layers are reached, and the final G(z n ) is the treatment method we need. The number of iterations n is not the more the better. When iterating to a certain number of layers, the mean of the data will converge to a value that is highly similar to the eigenvalue of the original input data.

[0069] In another embodiment of the present disclosure, the risk data handling method also includes constructing a semi-supervised convolutional generative adversarial network. The generative adversarial network includes a generator and a discriminator. The generator inputs random noise, that is, generates "false data"; the discriminator identifies the "false data" and generates a discriminant structure; the two learn from each other during the training process until the discriminator cannot determine that the output of the generator is "false data", and an ideal training model is obtained.

[0070] Furthermore, constructing a semi-supervised convolutional generative adversarial network includes: using the first softmax classifier as the output layer of the generator; using the second softmax classifier as the output layer of the discriminator; wherein the first softmax classifier and the second softmax classifier are both softmax classifiers, but the second softmax classifier used for the discriminator output layer has at least one more output neuron than the first softmax classifier used for the generator output layer, and the extra output neuron is used to indicate the type of "fake data" generated by the generator. Based on the semi-supervised convolutional generative adversarial network constructed as above, both labeled training samples and unlabeled training samples can be used, thereby realizing the recognition of fixed treatment types.

[0071] In another embodiment of the present disclosure, the risk data processing method is used to intelligently process risk data, which is divided into three stages: risk data preprocessing, model training, and risk data intelligent processing.

[0072] In the risk data preprocessing stage, we first need to collect the record information of historical risk data. For different data sources, such as databases, redis, and cloud, we use the Kafka message queue method to collect historical risk data and corresponding disposal methods. This method can collect data in real time to ensure the timeliness of training samples. Secondly, we need to clean the record information of the collected historical risk data, use the finite state machine to configure regular rules, delete duplicate, space, extra characters, missing fields, inconsistent upper and lower case data, and use the mean to supplement abnormal values. Finally, we get the preprocessed risk data and corresponding disposal methods that meet the preset standards.

[0073] In the model training stage, the pre-processed risk data and the corresponding disposal methods that meet the preset standards are input into the semi-supervised generative adversarial network model to train the generative adversarial network model. Among them, the optimizer used is ADAM, an adaptive non-convex optimization method for optimizing the loss function, which can adaptively adjust the learning rate and update parameters. The generative adversarial network model is trained using the training method of the first generative adversarial network model in the above-mentioned embodiment.

[0074] In the stage of intelligent disposal of risk data, the risk data to be processed is input into the trained generative adversarial network model. First, a set containing risk data and disposal methods is generated. The Relu activation function is used to allow the generator of the generative adversarial network model to generate a set containing risk data and disposal methods under the restriction of the Relu activation function. The set is smoothed using Poisson fusion. Then, the back propagation method is used to perform gradient descent in the set for the input risk data to be processed, and the record of risk data with the highest eigenvalue matching and the corresponding disposal method is found. Finally, the disposal measure corresponding to the record is obtained as the final disposal measure result.

[0075] After completing the intelligent disposal of risk data, the risk data and the disposal method with the highest matching degree are used as samples for training the generative adversarial network model to train the generative adversarial network.

[0076] The risk data handling method disclosed in the present invention has a simpler handling process. It does not require a task allocation process or business personnel to configure model strategies. Instead, it conducts model training based on a large amount of historical risk handling data and risk data handling data processed by the model. This can achieve intelligent handling of all risk data, improve the accuracy of intelligent risk handling, and greatly reduce the workload of business personnel in writing and configuring risk models.

[0077] The risk data disposal method provided by the present disclosure is versatile, has high coverage, and has high rule dimensions. The strategy based on the generative adversarial network model that can achieve real-time training does not target a single risk dimension, and the training model extracts and learns feature variables by itself. There is no need to configure for different business scenarios, and all disposal scenarios can be covered.

[0078] Based on the above risk data processing method, the present disclosure also provides a risk data processing device. Figure 6 The device is described in detail.

[0079] Figure 6 The structure block diagram of the risk data handling device according to the embodiment of the present disclosure is schematically shown.

[0080] like Figure 6 As shown, the risk data handling device 600 of this embodiment includes a collection module 601 , a first training module 602 , a first processing module 603 , a second training module 604 , and a second processing module 605 .

[0081] The collection module 601 is used to obtain historical risk data and corresponding treatment methods. In one embodiment, the collection module 601 can be used to perform the operation S201 described above, which will not be described in detail here.

[0082] The first training module 602 is used to input the historical risk data and the corresponding treatment methods into the semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model. In one embodiment, the first training module 602 can be used to perform the operation S202 described above, which will not be repeated here.

[0083] The first processing module 603 is used to process the risk data to be processed by using the generative adversarial network model to obtain the treatment method with the highest matching degree with the risk data to be processed. In one embodiment, the first processing module 603 can be used to perform the operation S203 described above, which will not be repeated here.

[0084] The second training module 604 is used to train the first generative adversarial network model with the current risk data to be processed and the corresponding treatment method with the highest matching degree to obtain a second generative adversarial network model. In one embodiment, the second training module 604 can be used to perform the operation S204 described above, which will not be repeated here.

[0085] The second processing module 605 is used to process the next risk data to be processed using the second generative adversarial network model to obtain a treatment method with the highest matching degree with the next risk data to be processed. In one embodiment, the second processing module 605 can be used to perform the operation S205 described above, which will not be repeated here.

[0086] According to an embodiment of the present disclosure, any multiple modules of the collection module 601, the first training module 602, the first processing module 603, the second training module 604, and the second processing module 605 can be combined in one module, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the collection module 601, the first training module 602, the first processing module 603, the second training module 604, and the second processing module 605 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 any appropriate combination of any of them. Alternatively, at least one of the collection module 601, the first training module 602, the first processing module 603, the second training module 604, and the second processing module 605 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0087] Figure 7 A block diagram of an electronic device suitable for implementing a risk data handling method according to an embodiment of the present disclosure is schematically shown.

[0088] like Figure 7 As shown, the electronic device 700 according to an embodiment of the present disclosure includes a processor 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage part 708 into a random access memory (RAM) 703. The processor 701 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 701 may also include an onboard memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0089] In RAM 703, various programs and data required for the operation of electronic device 700 are stored. Processor 701, ROM 702 and RAM 703 are connected to each other via bus 704. Processor 701 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 702 and / or RAM 703. It should be noted that the program can also be stored in one or more memories other than ROM 702 and RAM 703. Processor 701 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.

[0090] According to an embodiment of the present disclosure, the electronic device 700 may further include an input / output (I / O) interface 705, which is also connected to the bus 704. The electronic device 700 may further include one or more of the following components connected to the I / O interface 705: an input portion 706 including a keyboard, a mouse, etc.; an output portion 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 708 including a hard disk, etc.; and a communication portion 709 including a network interface card such as a LAN card, a modem, etc. The communication portion 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage portion 708 as needed.

[0091] 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.

[0092] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, 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 of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 702 and / or RAM 703 described above and / or one or more memories other than ROM 702 and RAM 703.

[0093] The 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 shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the item recommendation method provided by the embodiment of the present disclosure.

[0094] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 701. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0095] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 709, and / or installed from the removable medium 711. 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.

[0096] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the processor 701, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0097] 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).

[0098] The flow charts 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 flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order 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 flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0099] It will be appreciated by those skilled in the art that 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, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, 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 without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0100] 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 risk data handling method, comprising: Acquire historical risk data and corresponding disposal methods, wherein the risk data represents the risk of transactions conducted by the user, and the disposal method represents the disposal of the account corresponding to the transaction; Inputting the historical risk data and the corresponding disposal methods into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model; Using the generative adversarial network model to process the risk data to be processed, and obtaining a treatment method with the highest matching degree with the risk data to be processed; The first generative adversarial network model is trained using the current risk data to be processed and the corresponding treatment method with the highest matching degree to obtain a second generative adversarial network model; Using the second generative adversarial network model to process the next risk data to be processed, and obtaining a disposal method with the highest matching degree with the next risk data to be processed; The step of inputting the historical risk data and the corresponding disposal methods into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model includes: Traversing the historical risk data and the corresponding disposal methods, the training is divided into 12 cycles according to 12 months per year, and the cycles are divided into several batches according to the card type dimension; Training the generative adversarial network model according to the cycle and the batch; The accuracy of the training recognition results of each batch is counted, the training recognition results of the batches within the same cycle are sampled, and the optimizer is run to update the model parameters to obtain the first generative adversarial network model.

2. According to the method of claim 1, the risk data to be processed is processed by using the generative adversarial network model to obtain a treatment method with the highest matching degree with the risk data to be processed, including: Generate a collection of risk data and treatment methods; performing smoothing processing on the set; The back-propagation method is used to iteratively process the risk data to be processed, and the eigenvalues ​​generated by each iteration are compared with the eigenvalues ​​of the original input risk data to be processed to obtain the disposal method with the highest eigenvalue matching degree.

3. The method according to claim 2, wherein generating a set of risk data and treatment methods comprises: Generate a set of risk data and treatment methods under the constraints of the Relu activation function.

4. According to the method of claim 2, the method of obtaining the highest matching degree comprises: Generate random noise G(z0); Traverse the set, through n iterations, iteratively update to G(z n ), G(z n ) is the disposal method with the highest matching degree.

5. The method according to claim 1, further comprising constructing the semi-supervised convolutional generative adversarial network, comprising: Use the first softmax classifier as the output layer of the generator; Use the second softmax classifier as the output layer of the discriminator; The second softmax classifier has at least one more output neuron than the first softmax classifier.

6. According to the method of claim 1, the first generative adversarial network model adopts ADAM optimizer.

7. According to the method of claim 1, the obtaining of historical risk data and corresponding disposal methods comprises: The Kafka message queue is used to collect the historical risk data and the corresponding disposal methods.

8. According to the method of claim 1, the obtaining of historical risk data and corresponding disposal methods further comprises: The historical risk data and the corresponding disposal methods are cleaned to obtain pre-processed risk data and the corresponding disposal methods that meet the preset standards.

9. A risk data processing device, comprising: A collection module, used to obtain historical risk data and corresponding disposal methods, wherein the risk data represents the risk of transactions conducted by the user, and the disposal method represents the disposal of the account corresponding to the transaction; A first training module, used for inputting the historical risk data and the corresponding disposal methods into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model; A first processing module, used to process the risk data to be processed currently using the generative adversarial network model to obtain a disposal method with the highest matching degree with the risk data to be processed currently; A second training module is used to train the first generative adversarial network model with the current risk data to be processed and the corresponding disposal method with the highest matching degree to obtain a second generative adversarial network model; A second processing module, configured to process the next risk data to be processed using the second generative adversarial network model to obtain a disposal method with the highest matching degree with the next risk data to be processed; The step of inputting the historical risk data and the corresponding disposal methods into a semi-supervised convolutional generative adversarial network for training to obtain a first generative adversarial network model includes: Traversing the historical risk data and the corresponding disposal methods, the training is divided into 12 cycles according to 12 months per year, and the cycles are divided into several batches according to the card type dimension; Training the generative adversarial network model according to the cycle and the batch; The accuracy of the training recognition results of each batch is counted, the training recognition results of the batches within the same cycle are sampled, and the optimizer is run to update the model parameters to obtain the first generative adversarial network model.

10. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 8.

12. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images