A business processing method, model training method, device and equipment

By receiving and integrating service application information with multiple modal features, the accuracy of user identity verification in online service processing is solved, the difficulty of identifying forged data is improved, and the security of service applications is enhanced.

CN111950641BActive Publication Date: 2024-12-06INDUSTRIAL AND COMMERCIAL BANK OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010821202.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-08-14
Publication Date
2024-12-06
Estimated Expiration
2040-08-14

AI Technical Summary

Technical Problem

In online business processing, user identity information in a single data form is easily forged, which makes it difficult to identify forged information and causes data or funds leakage.

Method used

By receiving service application information containing multiple modal features, extracting and fusing each modal feature data, determining whether the service processing request is an abnormal request, thereby improving the accuracy of identity verification.

Benefits of technology

By comprehensively considering data of multiple modal features, the difficulty of forging data is increased, the accuracy of user application verification is improved, the situation of single data types is avoided, and the security of business applications is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111950641B_ABST
    Figure CN111950641B_ABST
Patent Text Reader

Abstract

The embodiments of this specification provide a business processing method, a model training method, an apparatus and equipment, which can be used in the field of artificial intelligence. The method includes: receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features; respectively extracting modal feature data corresponding to the modal features in the business application information; fusing each modal feature data to obtain fused feature data; judging whether the business processing request is an abnormal request based on the fused feature data; and processing the business corresponding to the business processing request if the business processing request is not an abnormal request. Through the above method, when processing the user's business, the user's identity can be verified using data of multiple modal features, thereby improving the accuracy of identity authentication.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of this specification relate to the field of artificial intelligence technology, and in particular to a business processing method, a model training method, a device and equipment. Background Art

[0002] With the development of Internet technology, more and more institutions and enterprises use their own service systems to realize online business processing. When processing online business, the system verifies the identity of the user and processes the business application submitted by the user after the identity verification is passed.

[0003] However, currently when processing services, the received user identity information and service application information are in a single data format, such as describing the user only in text. Such application information in a single data format is often easier to forge. When criminals forge the application information, since the forged application information often cannot reflect more comprehensive feature information, it is generally difficult to identify the forged information, which in turn causes data or funds to be leaked. Therefore, there is an urgent need for a technical solution that can more comprehensively verify the identity of users when processing services. Summary of the invention

[0004] The purpose of the embodiments of this specification is to provide a business processing method, model training method, device and equipment to solve the technical problem of how to accurately verify the identity of a user when processing business.

[0005] In order to solve the above technical problems, the embodiment of this specification proposes a service processing method, including:

[0006] Receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features;

[0007] respectively extracting modal feature data corresponding to the modal features from the service application information;

[0008] Fusing feature data of each modality to obtain fused feature data;

[0009] Determining whether the service processing request is an abnormal request according to the fused feature data;

[0010] When the service processing request is not an abnormal request, the service corresponding to the service processing request is processed.

[0011] The embodiment of this specification also provides a service processing device, including:

[0012] A request receiving module, used to receive a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features;

[0013] A modal feature data extraction module, used to extract modal feature data corresponding to the modal features from the service application information;

[0014] A modal feature data fusion module is used to fuse the feature data of each modal to obtain fused feature data;

[0015] A judgment module, used for judging whether the service processing request is an abnormal request according to the fused feature data;

[0016] The business processing module is used to process the business corresponding to the business processing request when the business processing request is not an abnormal request.

[0017] The embodiment of this specification also proposes a business processing device, including a memory and a processor; the memory is used to store computer program instructions; the processor is used to execute the computer program instructions to implement the following steps: receiving a business processing request; the business processing request contains business application information; the business application information corresponds to at least two modal features; extracting modal feature data corresponding to the modal features from the business application information; fusing the various modal feature data to obtain fused feature data; judging whether the business processing request is an abnormal request based on the fused feature data; and if the business processing request is not an abnormal request, processing the business corresponding to the business processing request.

[0018] In order to solve the above technical problems, the embodiment of this specification also proposes a model training method, including:

[0019] Acquire business application sample data; the business application sample data corresponds to at least two modal features;

[0020] Extracting modal feature sample data corresponding to the modal features from the business application sample data respectively;

[0021] Fusing the modal feature sample data to obtain a fused sample vector;

[0022] A pre-built abnormal business identification model is trained according to the fused sample vector.

[0023] The embodiment of this specification also proposes a model training device, including:

[0024] A sample data acquisition module, used to acquire business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features;

[0025] A modal feature sample data extraction module, used to extract modal feature sample data corresponding to the modal features from the business application sample data;

[0026] A sample vector fusion module, used for fusing the modal feature sample data to obtain a fused sample vector;

[0027] The model training module is used to train an abnormal business identification model according to the fused sample vector and the corresponding abnormal condition.

[0028] The embodiment of this specification also proposes a model training device, including a memory and a processor;

[0029] The memory is used to store computer program instructions;

[0030] The processor is used to execute the computer program instructions to implement the following steps: obtaining business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features; modal feature sample data corresponding to the modal features are respectively extracted from the business application sample data; the modal feature sample data are fused to obtain a fused sample vector; and an abnormal business identification model is trained based on the fused sample vector and the corresponding abnormal conditions.

[0031] It can be seen from the technical solutions provided by the above embodiments of this specification that the embodiments of this specification obtain business application information corresponding to at least two modal features, extract the modal feature data corresponding to these modal features respectively, fuse these modal feature data, and use the fused fusion feature data to determine whether the business processing request is an abnormal request. Through the above method, when judging the business application information, multiple modal features can be comprehensively considered. By identifying the data after fusing multiple modal features, the situation of a relatively single data type can be avoided, so that the fused data has richer features, thereby increasing the difficulty of forging data and improving the accuracy of verifying user applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 This is a flow chart of a model training method according to an embodiment of this specification;

[0034] Figure 2 This is a flow chart of a business processing method according to an embodiment of this specification;

[0035] Figure 3 This is a module diagram of a model training device according to an embodiment of this specification;

[0036] Figure 4 A module diagram of a service processing device according to an embodiment of this specification;

[0037] Figure 5 This is a structural diagram of a model training device according to an embodiment of this specification;

[0038] Figure 6 This is a structural diagram of a business processing device according to an embodiment of this specification. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0040] In order to solve the above technical problems, firstly, Figure 1 An embodiment of a model training method in an embodiment of this specification is described. The execution subject of the method can be a model training device, and the model training device includes but is not limited to a server, an industrial computer, a PC, etc. The method includes the following specific execution steps.

[0041] S110: Acquire business application sample data; the business application sample data corresponds to at least two modal features.

[0042] The business application sample data is sample data obtained for training the abnormal business identification model. The business application sample data may be sample data provided for the business to be identified. For example, when the business is a credit card application business, the business application information submitted by the user in the historical records may be collected and used as the business application sample data.

[0043] The business application sample data may correspond to an abnormal condition annotation, and the abnormal condition annotation is used to mark whether the business application corresponding to the business application sample data is an abnormal application. Accordingly, based on the abnormal condition annotation and the business application sample data, a supervised learning method may be used for subsequent model training. The business application sample data may also not have corresponding annotations, and the preset classifier may be trained directly using an unsupervised learning method during subsequent training. The specific method of machine learning can be selected according to the actual needs, and will not be elaborated here.

[0044] The modal feature is used to indicate the type corresponding to the data contained in the business application sample data. In some embodiments, the modal feature may include at least two of image modal features, fingerprint modal features, audio modal features, and text modal features. Image modal features may correspond to image data, such as data such as a photo of the applicant, an ID image, etc.; fingerprint modal features may correspond to fingerprint data provided by the applicant, such as a fingerprint collected by a terminal device; audio modal features may correspond to audio sent by the applicant, such as a voice command pre-set by the applicant, or sound information emitted by the applicant himself; text modal features may correspond to text submitted by the applicant, such as application material information, personal identity information, and other text information. The above examples are only for the purpose of introducing the modal features. In actual applications, other information types may be selected as modal features, which will not be elaborated here.

[0045] S120: extracting modal feature sample data corresponding to the modal features from the business application sample data respectively.

[0046] The modal feature sample data is data corresponding to the modal feature. For example, when the modal feature is an image modal feature, the image in the modal feature sample data can be extracted as the modal feature sample data; when the modal feature is an audio modal feature, the audio file in the modal feature sample data can be extracted as the modal feature sample data. The specific method of extracting the modal feature sample data can be adjusted according to the needs of the actual application, and will not be described in detail.

[0047] In some embodiments, when extracting the modal feature sample data, data extraction can be performed on feature dimensions with strong business significance. For example, when extracting text modal feature sample data, text information corresponding to certain specific content can be selected as the text modal feature sample data; when selecting data corresponding to image modal features, pictures with strong business significance, such as certificate images, can be identified. By combining the business significance of modal feature sample data for screening, the effectiveness of the selected modal feature sample data is improved, and the recognition accuracy of the trained model is improved.

[0048] S130: Fusing the modal feature sample data to obtain a fused sample vector.

[0049] After obtaining the modal feature sample data corresponding to at least two modal features, the modal feature sample data can be fused. A specific fusion method can be to fuse these modal feature sample data into a feature vector, and then map the fused feature vector to a pre-set vector subspace to obtain a fused sample vector. The above method is only an exemplary introduction to the process of fusing modal feature sample data. In actual applications, other methods for obtaining fused sample vectors can be selected according to needs, which will not be described in detail here.

[0050] In some embodiments, before fusing the modal feature sample data, the modal feature sample data may be preprocessed to reduce noise in the modal feature sample data and improve the accuracy of the training result. The preprocessing may include denoising and standardization.

[0051] De-noising is to remove the noise sample data according to the modal feature sample data. Specifically, the noise sample data can be determined in the modal feature sample data according to a preset noise data template and then removed, or the noise sample data can be directly screened out in the modal feature sample data based on the continuity and other characteristics of the modal feature sample data. The specific method of determining the noise sample data can be adjusted according to the needs of the actual application, and is not limited to the above examples, and will not be repeated here.

[0052] Standardization processing can adjust the modal feature sample data to the same data distribution range, so that the sample data distribution is consistent, which is conducive to the processing of the modal feature sample data in the subsequent process. Standardization processing can be, for example, normalization processing, which divides the data into a distribution range of [0,1], so that the application of the data is more convenient.

[0053] S140: Training a pre-built abnormal business identification model according to the fused sample vector.

[0054] The abnormal business identification model may be a model for identifying abnormal business. The abnormal business identification model may be a mathematical model for identifying whether the corresponding business is an abnormal business in combination with different modal features of the business application data. The abnormal business identification model may be a Bayesian classification model, a support vector machine classification model (Support Vector Machine, SVM), or a convolutional neural network classification model (Convolutional Neural Networks, CNN), etc., without limitation.

[0055] After the fused sample vector is obtained, the abnormal business identification model can be trained according to the fused sample vector, so that the trained model can identify abnormal conditions of the business based on the training samples.

[0056] In some implementations, multiple abnormal business identification models can be pre-built, and the abnormal business identification models can be trained based on the same business application sample data. For these abnormal business identification models, the verification data is input into these abnormal business identification models to obtain verification results corresponding to each model. The accuracy of each abnormal business identification model is judged according to the verification results, and the final abnormal business identification model applied to business processing is determined according to the accuracy of different models.

[0057] The above model training method obtains the business application information corresponding to at least two modal features, extracts the modal feature data corresponding to these modal features respectively, fuses these modal feature data, and uses the fused fusion feature data to train the abnormal business identification model, so that the abnormal business identification model can identify the abnormal conditions of the business. Through the above method, when judging the business application information, multiple modal features can be comprehensively considered. By identifying the data after fusing multiple modal features, the situation of a relatively single data type can be avoided, so that the fused data has richer features, thereby increasing the difficulty of forging data and improving the accuracy of verifying user applications.

[0058] Based on the above-mentioned model training method embodiment, this specification also proposes a business processing method. The execution subject of the business processing method is a business processing device, which includes but is not limited to a server, an industrial computer, a PC, etc. Figure 2 As shown, the method includes the following specific execution steps.

[0059] S210: Receive a service processing request; the service processing request includes service application information; the service application information corresponds to at least two modal features.

[0060] The business processing request is a request provided by a user for processing a specific business. For example, when a user needs to apply for a credit card, the business processing request may be a request for applying for a credit card.

[0061] The service processing request includes service application information, and the service application information may be data provided by the service to be processed, for example, personal identity information, certificate information, biometric information, etc. of the user.

[0062] The modal feature is used to indicate the type corresponding to the data contained in the business application sample data. In some embodiments, the modal feature may include at least two of image modal features, fingerprint modal features, audio modal features, and text modal features. Image modal features may correspond to image data, such as data such as a photo of the applicant, an ID image, etc.; fingerprint modal features may correspond to fingerprint data provided by the applicant, such as a fingerprint collected by a terminal device; audio modal features may correspond to audio sent by the applicant, such as a voice command pre-set by the applicant, or sound information emitted by the applicant himself; text modal features may correspond to text submitted by the applicant, such as application material information, personal identity information, and other text information. The above examples are only for the purpose of introducing the modal features. In actual applications, other information types may be selected as modal features, which will not be elaborated here.

[0063] S220: extracting modal feature data corresponding to the modal features from the service application information respectively.

[0064] The modal feature data is the data corresponding to the modal feature. For example, when the modal feature is an image modal feature, the image in the modal feature data can be extracted as the modal feature data; when the modal feature is an audio modal feature, the audio file in the modal feature data can be extracted as the modal feature data. The specific method of extracting modal feature data can be adjusted according to the needs of the actual application, which will not be described in detail.

[0065] In some implementations, when extracting the modal feature data, data extraction can be performed on feature dimensions with strong business implications. For example, when extracting text modal feature data, text information corresponding to certain specific content can be selected as the text modal feature data; when selecting data corresponding to image modal features, pictures with strong business implications, such as certificate images, can be identified. By combining the business implications of modal feature data for screening, the effectiveness of the selected modal feature data is improved, and the recognition accuracy of the trained model is improved.

[0066] S230: Fusing feature data of each modality to obtain fused feature data.

[0067] After obtaining the modal feature sample data corresponding to at least two modal features, the modal feature sample data can be fused. A specific fusion method can be to fuse these modal feature sample data into a feature vector, and then map the fused feature vector to a pre-set vector subspace to obtain a fused sample vector. The above method is only an exemplary introduction to the process of fusing modal feature sample data. In actual applications, other methods for obtaining fused sample vectors can be selected according to needs, which will not be described in detail here.

[0068] In some embodiments, before fusing the modal feature sample data, the modal feature sample data may be preprocessed to reduce noise in the modal feature sample data and improve the accuracy of the training result. The preprocessing may include denoising and standardization.

[0069] De-noising is to remove the noise sample data according to the modal feature sample data. Specifically, the noise sample data can be determined in the modal feature sample data according to a preset noise data template and then removed, or the noise sample data can be directly screened out in the modal feature sample data based on the continuity and other characteristics of the modal feature sample data. The specific method of determining the noise sample data can be adjusted according to the needs of the actual application, and is not limited to the above examples, and will not be repeated here.

[0070] Standardization processing can adjust the modal feature sample data to the same data distribution range, so that the sample data distribution is consistent, which is conducive to the processing of the modal feature sample data in the subsequent process. Standardization processing can be, for example, normalization processing, which divides the data into a distribution range of [0,1], so that the application of the data is more convenient.

[0071] S240: Determine whether the business processing request is an abnormal request based on the fused feature data.

[0072] In some implementations, judging whether a service processing request is an abnormal request based on the fused feature data may be to use an abnormal service identification model to identify whether the service processing request corresponding to the fused feature data is an abnormal request. The abnormal service identification model may be based on Figure 1 The model obtained by training in the corresponding embodiment. For the introduction of the abnormal business identification model, please refer to Figure 1 The description in the corresponding embodiments will not be repeated here.

[0073] S250: When the service processing request is not an abnormal request, process the service corresponding to the service processing request.

[0074] If the service processing request is not an abnormal request based on the fused feature data, it indicates that the service processing request is a normal request, and there is no high risk in processing it, and the service can be processed based on the normal processing flow. The specific processing method can be combined with the characteristics of the service in actual application to process the service, which will not be described in detail here.

[0075] If the business processing request is identified as an abnormal request based on the fusion feature data, it indicates that there may be certain risks in processing the business corresponding to the business processing request, which may easily cause data or funds to be leaked. The business processing request can be rejected and the corresponding rejection information can be fed back. The specific processing method can also be set according to the needs of the actual application, which will not be repeated here.

[0076] The above-mentioned business processing method obtains business application information corresponding to at least two modal features, extracts modal feature data corresponding to these modal features respectively, fuses these modal feature data, and uses the fused fused feature data to determine whether the business processing request is an abnormal request. Through the above-mentioned method, when judging the business application information, multiple modal features can be comprehensively considered. By identifying the data after fusing multiple modal features, the situation of relatively single data type can be avoided, so that the fused data has richer features, thereby increasing the difficulty of forging data and improving the accuracy of verifying user applications.

[0077] Based on the above-mentioned embodiment of the model training method, the embodiment of this specification also provides a model training device, which can be integrated into the model training device, such as Figure 3 As shown, the model training device may include the following modules.

[0078] The sample data acquisition module 310 is used to acquire business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features;

[0079] A modal feature sample data extraction module 320, used to extract modal feature sample data corresponding to the modal features from the business application sample data;

[0080] A sample vector fusion module 330, used to fuse the modal feature sample data to obtain a fused sample vector;

[0081] The model training module 340 is used to train an abnormal business identification model according to the fused sample vector and the corresponding abnormal condition.

[0082] Based on the above-mentioned embodiment of the service processing method, the embodiment of this specification also provides a service processing device, which can be integrated into the service processing device, such as Figure 4 As shown, the business processing device may include the following modules.

[0083] The request receiving module 410 is used to receive a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features;

[0084] A modal feature data extraction module 420, used to extract modal feature data corresponding to the modal features from the service application information;

[0085] A modal feature data fusion module 430 is used to fuse the feature data of each modality to obtain fused feature data;

[0086] A judgment module 440 is used to judge whether the service processing request is an abnormal request according to the fused feature data;

[0087] The business processing module 450 is used to process the business corresponding to the business processing request when the business processing request is not an abnormal request.

[0088] Based on the above model training method, such as Figure 5 As shown, the embodiment of this specification also provides a model training device. The model training device may include a memory and a processor.

[0089] In this embodiment, the memory may be implemented in any appropriate manner. For example, the memory may be a read-only memory, a mechanical hard disk, a solid-state hard disk, or a USB flash drive, etc. The memory may be used to store computer program instructions.

[0090] In this embodiment, the processor may be implemented in any suitable manner. For example, the processor may take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc.

[0091] The processor can execute the computer program instructions to implement the following steps: obtaining business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features; respectively extracting modal feature sample data corresponding to the modal features from the business application sample data; fusing the modal feature sample data to obtain a fused sample vector; and training an abnormal business identification model based on the fused sample vector and the corresponding abnormal conditions.

[0092] Based on the above business processing method, such as Figure 6 As shown, the embodiment of this specification also provides a service processing device. The service processing device may include a memory and a processor.

[0093] In this embodiment, the memory may be implemented in any appropriate manner. For example, the memory may be a read-only memory, a mechanical hard disk, a solid-state hard disk, or a USB flash drive, etc. The memory may be used to store computer program instructions.

[0094] In this embodiment, the processor may be implemented in any suitable manner. For example, the processor may take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc.

[0095] The processor can execute the computer program instructions to implement the following steps: receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features; extracting modal feature data corresponding to the modal features in the business application information; fusing the various modal feature data to obtain fused feature data; judging whether the business processing request is an abnormal request based on the fused feature data; and if the business processing request is not an abnormal request, processing the business corresponding to the business processing request.

[0096] It should be noted that the business processing methods, model training methods, devices and equipment disclosed in the embodiments of this specification can be used in the field of artificial intelligence technology to realize the verification and processing of business applications. Of course, the business processing methods, model training methods, devices and equipment can also be applied to other fields without limitation.

[0097] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip2. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog2. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0098] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0099] It can be known from the above description of the implementation mode that the technicians in this field can clearly understand that the present specification can be implemented by means of software plus the necessary general hardware platform. Based on such an understanding, the technical solution of the present specification can be essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present specification or some parts of the embodiments.

[0100] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0101] This specification can be used in many general or special computer system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments that include any of the above systems or devices.

[0102] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0103] Although the present specification is described through embodiments, those skilled in the art will appreciate that there are many modifications and changes to the present specification without departing from the spirit of the present specification, and it is intended that the appended claims include these modifications and changes without departing from the spirit of the present specification.

Claims

1. A business processing method, characterized in that: include: Receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features; the modal features include at least two of image modal features, fingerprint modal features, audio modal features and text modal features corresponding to the applicant; the modal features are used to verify the user application; Extracting modal feature data corresponding to the modal features from the business application information respectively; wherein the steps include: extracting data from feature dimensions with strong business implications; The feature data of each modality are fused to obtain fused feature data; the fused feature data is a vector; Determining whether the service processing request is an abnormal request according to the fused feature data; When the service processing request is not an abnormal request, the service corresponding to the service processing request is processed.

2. The method according to claim 1, characterized in that The fusing of the feature data of each modality to obtain fused feature data includes: Fuse the feature data of each modality into a fused feature vector; The fused feature vector is mapped to a preset vector subspace to obtain fused feature data.

3. The method according to claim 1, characterized in that The determining whether the service processing request is an abnormal request according to the fused feature data includes: An abnormal business identification model is used to identify whether the business processing request corresponding to the fused feature data is an abnormal request.

4. A service processing device, characterized in that: include: A request receiving module, used for receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features; the modal features include at least two of the following: image modal features, fingerprint modal features, audio modal features, and text modal features corresponding to the applicant; the modal features are used for verifying the user application; A modal feature data extraction module is used to extract modal feature data corresponding to the modal features in the business application information; wherein, it includes: extracting data for feature dimensions with strong business significance; A modal feature data fusion module, used for fusing each modal feature data to obtain fused feature data; the fused feature data is a vector; A judgment module, used for judging whether the service processing request is an abnormal request according to the fused feature data; The business processing module is used to process the business corresponding to the business processing request when the business processing request is not an abnormal request.

5. A business processing device comprising a memory and a processor; The memory is used to store computer program instructions; The processor is used to execute the computer program instructions to implement the following steps: receiving a business processing request; the business processing request includes business application information; the business application information corresponds to at least two modal features; the modal features include at least two of the image modal features, fingerprint modal features, audio modal features and text modal features corresponding to the applicant; the modal features are used to verify the user application; modal feature data corresponding to the modal features are respectively extracted from the business application information; wherein, include: Extract data from feature dimensions with strong business significance; fuse feature data of each modality to obtain fused feature data; The fused feature data is a vector; judging whether the service processing request is an abnormal request according to the fused feature data; and processing the service corresponding to the service processing request if the service processing request is not an abnormal request.

6. A model training method, characterized in that: include: Acquire business application sample data; the business application sample data corresponds to at least two modal features; the modal features include at least two of image modal features, fingerprint modal features, audio modal features and text modal features corresponding to the applicant; the modal features are used to verify the user application; Extracting modal feature sample data corresponding to the modal features from the business application sample data respectively; wherein the steps include: extracting data from feature dimensions with strong business significance; The modal feature sample data is fused to obtain a fused sample vector; the fused sample vector is a vector; A pre-built abnormal business identification model is trained according to the fused sample vector.

7. The method according to claim 6, characterized in that Before fusing the feature data of each modality to obtain a fused sample vector, the method further includes: The modal feature data are preprocessed respectively; the preprocessing includes denoising processing and standardization processing.

8. A model training device, characterized in that: include: A sample data acquisition module, used to acquire business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features; the modal features include at least two of the image modal features, fingerprint modal features, audio modal features and text modal features corresponding to the applicant; the modal features are used to verify the user application; A modal feature sample data extraction module is used to extract modal feature sample data corresponding to the modal features from the business application sample data; wherein, it includes: extracting data for feature dimensions with strong business significance; A sample vector fusion module, used for fusing the modal feature sample data to obtain a fused sample vector; the fused sample vector is a vector; The model training module is used to train an abnormal business identification model according to the fused sample vector and the corresponding abnormal condition.

9. A model training device comprising a memory and a processor; The memory is used to store computer program instructions; The processor is used to execute the computer program instructions to implement the following steps: obtaining business application sample data; the business application sample data is marked with abnormal conditions; the business application sample data corresponds to at least two modal features; the modal features include at least two of the image modal features, fingerprint modal features, audio modal features and text modal features corresponding to the applicant; the modal features are used to verify the user application; modal feature sample data corresponding to the modal features are respectively extracted from the business application sample data; wherein, include: Extract data from feature dimensions with strong business significance; fuse the modal feature sample data to obtain a fused sample vector; The fused sample vector is a vector; the abnormal business identification model is trained according to the fused sample vector and the corresponding abnormal conditions.