CN117475231A - Supervision method and device for customer financial assets and electronic equipment - Google Patents

Supervision method and device for customer financial assets and electronic equipment Download PDF

Info

Publication number
CN117475231A
CN117475231A CN202311491663.3A CN202311491663A CN117475231A CN 117475231 A CN117475231 A CN 117475231A CN 202311491663 A CN202311491663 A CN 202311491663A CN 117475231 A CN117475231 A CN 117475231A
Authority
CN
China
Prior art keywords
fund
client
financial
running water
running
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311491663.3A
Other languages
Chinese (zh)
Inventor
文弘扬
梁婷
贾小茹
任国飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Industrial and Commercial Bank of China Ltd ICBC
Original Assignee
Industrial and Commercial Bank of China Ltd ICBC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Industrial and Commercial Bank of China Ltd ICBC filed Critical Industrial and Commercial Bank of China Ltd ICBC
Priority to CN202311491663.3A priority Critical patent/CN117475231A/en
Publication of CN117475231A publication Critical patent/CN117475231A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/09—Supervised learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

The invention discloses a supervision method and device for customer financial assets and electronic equipment, and relates to the field of financial science and technology or other related technical fields, wherein the method comprises the following steps: n customer class attributes of financial service are obtained, for each customer class attribute, a fund flow extraction strategy associated with the customer class attribute is determined based on a preset class mapping table, a fund flow set of all the customer class attributes is synthesized, a flow index feature set is generated, feature sequence reordering is carried out on the flow index feature set, an admission feature image is generated based on the ordering result, the admission feature image is input into a pre-trained convolutional neural network, and an asset state evaluation result of each target customer is output. The invention solves the technical problems that in the related technology, when the supervision of the financial assets of the clients is carried out, the information quantity is only single specific attribute, and the data information quantity is lower.

Description

Supervision method and device for customer financial assets and electronic equipment
Technical Field
The invention relates to the field of financial science and technology or other related fields, in particular to a supervision method and device for customer financial assets and electronic equipment.
Background
Under the background of continuous expansion and iterative updating of financial data, the demands of financial institutions on asset supervision are increasingly raised, but the difficulty of asset clearing is relatively high, especially the client main body is a legal client, and the asset clearing period is usually counted in units of years, so that the safety control of funds after lending is very unfavorable. Therefore, the method is particularly important for the regular observation of the asset scale of the customer, but too frequent clearing application can lead to customer loss, and the management period with too long period is unfavorable for the fund supervision of the financial institution. Therefore, the asset assessment can be performed in a targeted manner while the reasonable regular supervision is needed, the default early warning is performed in time, the asset condition is ensured to be checked when necessary, and unnecessary fund loss is reduced.
In the related technology, the evaluation of the condition of the financial asset is generally carried out by adopting a regression model, a rule model, a scoring model, an Xgboost model and the like, and in the model evaluation process, the asset condition prediction is completed by adopting time point data and regional statistics data within a period of time by using the characteristics in a plurality of modes and adopting data fitting of different models.
However, the above-mentioned financial asset assessment method has obvious drawbacks, the collected funds aggregate data are relatively isolated aggregate transaction amount, time point balance and other data, the time domain data are the same attribute feature aggregate data in the same time region, such as the number of the funds in continuous 7 days, transaction amount and the like, the information quantity often has specific attributes, and the data information quantity is lower. Meanwhile, a large number of features with the same attribute type are molded, a high requirement is also put on calculation force, especially when related statistics of the running water type occurs in the data statistics process, the data scale is exponentially increased, batch data prediction is not facilitated, and the problem of number generation after batch running occurs in serious conditions.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the invention provides a supervision method and device for customer financial assets and electronic equipment, which at least solve the technical problems that in the related technology, when customer financial assets are supervised, the information quantity is only single specific attribute, and the data information quantity is lower.
According to one aspect of an embodiment of the present invention, there is provided a method of supervising a customer financial asset, comprising: for each financial service in a financial service set, acquiring N client class attributes related to the financial service, wherein the financial service set comprises M financial services, and N and M are positive integers; for each client category attribute, determining a fund running water extraction strategy associated with the client category attribute based on a preset category mapping table, wherein the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2; integrating the fund running water sets of all the client category attributes to generate a running water index feature set; performing feature sequence reordering on the running water index feature set, and generating an admittance feature image based on the ordering result; inputting the access characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the target client is compared with the asset accounting data of the previous time point and the client asset state generated based on the compared data.
Optionally, after determining the funds pipeline extraction policy associated with the customer category attribute based on the preset category mapping table, further comprising: under the condition that the client category attribute is a legal category attribute, inquiring the preset category mapping table to obtain a first fund running water extraction strategy corresponding to the legal category attribute; and based on the first fund running water extraction strategy, extracting fund running water of financial accounts related to legal customers in U transaction dimensions to obtain a first fund running water set, wherein the transaction dimensions comprise: for public transactions, online transactions and banking transactions, the first fund running set comprises a fund flow direction, a transaction time point and a financial account number, and U is smaller than or equal to T.
Optionally, after determining the funds pipeline extraction policy associated with the customer category attribute based on the preset category mapping table, further comprising: querying the preset category mapping table under the condition that the client category attribute is a personal category attribute to obtain a second funds running extraction strategy corresponding to the personal category attribute, wherein the second funds running extraction strategy corresponds to a plurality of financial transaction channels; and extracting fund flowing of the individual client in P transaction dimensions corresponding to a plurality of financial transaction channels based on the second fund flowing extraction strategy to obtain a second fund flowing set, wherein the second fund flowing set comprises: financial transaction channel, fund flow direction and transaction category, P is less than or equal to T.
Optionally, the step of integrating the funds running water sets of all the client category attributes to generate a running water index feature set includes: acquiring a first fund running set corresponding to a legal category attribute, extracting fund running related to a specified financial product and/or a specified financial transaction channel from the first fund running set, and acquiring first running index characteristics based on the fund running statistics and accounting amount data, accumulated accounting amount data and ring ratio accounting increment data related to a target legal client in a first preset running period; acquiring a second fund running set corresponding to the personal category attribute, extracting fund running of each financial transaction channel in the second fund running set, and acquiring second running index characteristics based on the fund running statistics and internet banking payment amount data, credit card payment amount data and ring ratio payment amount increment data of target personal clients in a second preset running period; the set of running water index features is generated based on the first running water index feature and the first running water index feature.
Optionally, the step of reordering the feature sequence of the running water index feature set and generating an admission feature image based on the ordering result includes: according to the time dimension, the running water index features of the running water index feature set with the same transaction dimension and different client category attributes are reordered in parallel; sequentially reordering the running water index features with different transaction dimensions and the same client category attribute in the running water index feature set according to the time sequence; and generating an admission characteristic image based on the sequencing result.
Optionally, the network structure of the convolutional neural network includes: an input layer for inputting the admittance feature image; the convolution layer slides a convolution kernel in the input admittance feature image through convolution operation, and the convolution kernel extracts running water index features; a pooling layer for reducing the spatial size of the access feature image; and the full connection layer is connected with the convolution layer and the pooling layer and outputs the asset state evaluation result of each target client.
Optionally, the convolution layer includes: the first sub-convolution layer comprises two groups of convolution kernels, wherein the convolution dimension of the convolution kernels is R transverse E longitudinal, R is the total number of client category attributes, and E is the time dimension; and the convolution dimension selection format of the Q+1 layers is R horizontal and vertical, and Q is an integer greater than 0.
According to another aspect of the embodiments of the present invention, there is also provided a supervision apparatus for a customer financial asset, including: the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring N client class attributes related to each financial business in a financial business set, wherein the financial business set comprises M financial businesses, and N and M are positive integers; the determining unit is used for determining a fund running water extraction strategy associated with the client category attribute based on a preset category mapping table for each client category attribute, wherein the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2; the generation unit is used for integrating the fund running water sets of all the client category attributes to generate a running water index feature set; the reordering unit is used for reordering the feature sequences of the flowing water index feature sets and generating an admission feature image based on the ordering result; the evaluation unit is used for inputting the admission characteristic image into a pre-trained convolutional neural network and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises the following components: the target client is compared with the asset accounting data of the previous time point and the client asset state generated based on the compared data.
Optionally, the supervision apparatus for the customer financial asset further comprises: the first query unit is used for querying the preset category mapping table to obtain a first funds running extraction strategy corresponding to the legal category attribute under the condition that the client category attribute is the legal category attribute after determining the funds running extraction strategy related to the client category attribute based on the preset category mapping table; the first extraction unit is configured to extract a running line of funds of the financial account related to the legal client in U transaction dimensions based on the first running line of funds extraction policy, so as to obtain a first running line set of funds, where the transaction dimensions include: for public transactions, online transactions and banking transactions, the first fund running set comprises a fund flow direction, a transaction time point and a financial account number, and U is smaller than or equal to T.
Optionally, the supervision apparatus for the customer financial asset further comprises: a second query unit, configured to query, after determining a funds running extraction policy associated with the client category attribute based on a preset category mapping table, the preset category mapping table to obtain a second funds running extraction policy corresponding to the personal category attribute if the client category attribute is a personal category attribute, where the second funds running extraction policy corresponds to a plurality of financial transaction channels; the second extraction unit is configured to extract a running line of funds of the individual customer in P transaction dimensions corresponding to the plurality of financial transaction channels based on the second running line of funds extraction policy, so as to obtain a second running line set of funds, where the second running line set of funds includes: financial transaction channel, fund flow direction and transaction category, P is less than or equal to T.
Optionally, the generating unit includes: the first acquisition module is used for acquiring a first fund running water set corresponding to the legal category attribute, extracting fund running water related to a specified financial product and/or a specified financial transaction channel from the first fund running water set, and acquiring first running water index characteristics based on the fund running water statistics, accounting amount data, accumulated accounting amount data and ring ratio accounting increment data related to a target legal customer in a first preset running water period; the second acquisition module is used for acquiring a second fund running water set corresponding to the personal category attribute, extracting fund running water related to each financial transaction channel in the second fund running water set, and acquiring second running water index characteristics based on the fund running water statistics and internet banking payment amount data, credit card payment amount data and ring ratio payment amount increment data related to the target personal client in a second preset running water period; the first generation module is used for generating the running water index feature set based on the first running water index feature and the first running water index feature.
Optionally, the reordering unit includes: the first reordering module is used for reordering the running water index features of the running water index feature set, which have the same transaction dimension and different client category attributes in parallel according to the time dimension; the second reordering module is used for sequentially reordering the flowing water index features with different transaction dimensions and the same client category attribute in the flowing water index feature set according to the time sequence; and the second generation module is used for generating an admission characteristic image based on the sequencing result.
Optionally, the network structure of the convolutional neural network includes: an input layer for inputting the admittance feature image; the convolution layer slides a convolution kernel in the input admittance feature image through convolution operation, and the convolution kernel extracts running water index features; a pooling layer for reducing the spatial size of the access feature image; and the full connection layer is connected with the convolution layer and the pooling layer and outputs the asset state evaluation result of each target client.
Optionally, the convolution layer includes: the first sub-convolution layer comprises two groups of convolution kernels, wherein the convolution dimension of the convolution kernels is R transverse E longitudinal, R is the total number of client category attributes, and E is the time dimension; and the convolution dimension selection format of the Q+1 layers is R horizontal and vertical, and Q is an integer greater than 0.
According to another aspect of the embodiment of the present invention, there is also provided a computer readable storage medium, including a stored computer program, where the computer program when executed controls a device in which the computer readable storage medium is located to perform any one of the above methods for supervising a customer's financial asset.
According to another aspect of embodiments of the present invention, there is also provided an electronic device including one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of supervising a customer financial asset of any of the above.
In the method, for each financial service in a financial service set, N client class attributes related to the financial service are acquired, for each client class attribute, a fund running water extraction strategy related to the client class attribute is determined based on a preset class mapping table, the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client class attribute, fund running water sets of all client class attributes are synthesized to generate a running water index feature set, feature sequence reordering is carried out on the running water index feature set, and an admission feature image is generated based on a sequencing result; inputting the admittance characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data.
In the method, a trend magnitude derivative strategy of the multidimensional attribute is introduced, the fund running water information of the multidimensional transaction dimension is extracted (the data information is high), the fund running water sets of the multidimensional client category attribute are synthesized to generate a running water index feature set, the feature set is sequentially reordered, and then the asset state evaluation result of each client is analyzed through a convolutional neural network, so that whether the client asset data is possibly violated or not can be effectively monitored, the default rate is reduced, and the asset quality rate of a financial institution is improved, and the technical problems that in the related technology, when the client financial asset supervision is carried out, the information quantity is only a single specific attribute, and the data information quantity is lower are solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation on the invention. In the drawings:
FIG. 1 is a flow chart of an alternative method of policing a customer's financial assets in accordance with an embodiment of the present invention;
FIG. 2 is an alternative customer asset state assessment schematic based on multi-attribute trend magnitude derivation in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of an alternative convolutional neural network in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of an alternative monitoring device for a customer's financial assets in accordance with an embodiment of the invention;
fig. 5 is a block diagram of a hardware architecture of an electronic device (or mobile device) for a method of policing a customer's financial assets, in accordance with an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
To facilitate an understanding of the invention by those skilled in the art, some terms or nouns involved in the various embodiments of the invention are explained below:
trend magnitude derivation refers to features or variables derived from trend magnitudes (e.g., trend direction and magnitude in time series data). For example, in time series analysis, trend magnitude derivation may be used to capture trend properties in time series data and provide information about trends.
The convolutional neural network Convolutional Neural Network, CNN for short, is a deep learning algorithm, which is mainly used for image recognition and processing tasks, extracts the characteristics of images through multi-layer convolution and pooling operations, and classifies or regresses through a full connection layer.
It should be noted that, the method and the device for supervising the customer financial asset in the present disclosure may be used in the technical field of financial science and technology to integrate the multi-dimensional attribute category features by introducing trend magnitude derivative features, and may also be used in any field other than the technical field of financial science and technology to integrate the multi-dimensional attribute category features by introducing trend magnitude derivative features, so that the application field of the method and the device for supervising the customer financial asset in the present disclosure is not limited.
It should be noted that, related information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present disclosure are information and data authorized by a user or sufficiently authorized by each party, and the collection, use and processing of related data need to comply with related laws and regulations and standards of each region, and provide corresponding operation entries for the user to select authorization or rejection. For example, an interface is provided between the system and the relevant user or institution, before acquiring the relevant information, the system needs to send an acquisition request to the user or institution through the interface, and acquire the relevant information after receiving the consent information fed back by the user or institution.
The following embodiments of the present invention are applicable to various systems/applications/devices for verifying encrypted data based on financial technology. According to the trend magnitude derivative special generation method of the multidimensional attribute, the original characteristics are reordered in time dimension sequence, then asset flow data compression is achieved through the convolutional neural network, and then the asset state evaluation results of all clients are prepared to be analyzed, particularly for credit business of a financial institution, whether the client asset data are likely to be violated or not can be effectively monitored, the violation rate is reduced, and the asset quality rate of the financial institution is improved.
In the invention, after the time dimension sequence reordering is carried out on the original characteristics, different flows of the same time node can be used as a minimum dimension convolution unit to construct the convolution neural network, the prediction target is the same-ratio condition of a group of periodic asset accounting data closest to the current time point, and the proportional division classification is carried out according to the service requirement, so that the convolution neural network is constructed by taking the same-dimension convolution unit as the model output.
The present invention will be described in detail with reference to the following examples.
Example 1
In accordance with an embodiment of the present invention, there is provided an embodiment of a method of supervision of a customer's financial assets, it being noted that the steps illustrated in the flowchart of the figures may be performed in a computer system such as a set of computer executable instructions, and, although a logical order is illustrated in the flowchart, in some cases, the steps illustrated or described may be performed in an order other than that illustrated herein.
FIG. 1 is a flow chart of an alternative method of policing a customer's financial assets, as shown in FIG. 1, according to an embodiment of the invention, the method comprising the steps of:
step S101, for each financial business in a financial business set, N customer category attributes related to the financial business are obtained, wherein the financial business set comprises M financial businesses, and N and M are positive integers;
step S102, for each client category attribute, determining a fund flowing water extraction strategy associated with the client category attribute based on a preset category mapping table, wherein the fund flowing water extraction strategy is used for extracting fund flowing water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2;
step S103, integrating the fund running water sets of all the client category attributes to generate a running water index feature set;
step S104, carrying out feature sequence reordering on the water flow index feature set, and generating an admission feature image based on the ordering result;
step S105, inputting the admission characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data.
Through the steps, N client class attributes related to the financial service can be acquired for each financial service in a financial service set, for each client class attribute, a fund running water extraction strategy related to the client class attribute is determined based on a preset class mapping table, the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client class attribute, fund running water sets of all client class attributes are synthesized, a running water index feature set is generated, feature sequence reordering is carried out on the running water index feature set, and an admittance feature image is generated based on a sequencing result; inputting the admittance characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data. In the embodiment, a trend magnitude derivative strategy of the multidimensional attribute is introduced, the fund running water information of the multidimensional transaction dimension is extracted (the data information is high), the fund running water sets of the multidimensional client category attribute are synthesized to generate a running water index feature set, the feature set is sequentially reordered, and then the asset state evaluation result of each client is analyzed through a convolutional neural network, so that whether the client asset data is possibly violated or not can be effectively monitored, the default rate is reduced, and the asset quality rate of a financial institution is improved, and the technical problems that in the related technology, when the client financial asset supervision is carried out, the information quantity is only a single specific attribute, and the data information quantity is lower are solved.
The embodiment can be applied to a supervision system for supervising the financial assets of a customer, in particular to a credit business supervision system. Based on a trend magnitude derivation strategy of multiple attributes, the convolution calculation is carried out on the stream data of different attributes, the convolution kernel calculation of the same dimension and different depths is carried out, the stream data of the multiple attributes are integrated in a quantified mode, derived features are generated, the large-amplitude information compression is completed in the data dimension, the data quality is improved, the magnitude of the modeling parameters is reduced, and the possibility of running delay number possibly occurring in the batch prediction process is reduced.
Embodiments of the present invention will be described in detail with reference to the following steps.
Step S101, for each financial transaction in the financial transaction set, N customer category attributes related to the financial transaction are acquired.
In this embodiment, the financial service set includes M financial services, where N and M are positive integers. Prior to conducting the financial asset monitoring, financial transactions requiring monitoring are identified, and product information for all financial products involved in each financial transaction is determined, wherein the financial transactions include, but are not limited to: credit businesses, savings businesses, transfer businesses, etc., take credit businesses as examples, and the financial products may include products related to house credits, forensic financing products, etc., and in this embodiment, all financial businesses to be supervised are integrated to generate a financial business set.
Step S102, for each client category attribute, determining a fund flowing water extraction strategy associated with the client category attribute based on a preset category mapping table, wherein the fund flowing water extraction strategy is used for extracting fund flowing water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2.
It should be noted that, the preset category mapping table in this embodiment stores mapping relationships between all client category attributes and the funds flow extraction policy. Different service attribute flow index features are screened, different flow features are needed to be adopted according to clients with different category attributes, in this embodiment, the client category attributes are generally determined according to specific service scenes, and are generally as follows: corporate and personal guest groups are described below for the two customer category attributes, respectively.
Optionally, after determining the funds pipeline extraction policy associated with the customer category attribute based on the preset category mapping table, further comprising: under the condition that the client category attribute is a legal category attribute, inquiring a preset category mapping table to obtain a first fund running water extraction strategy corresponding to the legal category attribute; and based on a first fund running water extraction strategy, extracting fund running water of financial accounts related to legal customers in U transaction dimensions to obtain a first fund running water set, wherein the transaction dimensions comprise: for public transactions, online transactions and banking transactions, the first fund running set comprises a fund flow direction, a transaction time point and a financial account number, and U is smaller than or equal to T.
For legal customers, as the running water mainly carries out financial services such as fund settlement, transaction and the like through one or more than one near-day account number/financial card, when the account number is taken, the time point data of dimensions such as public transit easiness, internet connection transaction, silver connection transaction and the like can be extracted.
It should be noted that, in the transaction dimension related to legal clients, public transportation is easy to refer to transactions performed between enterprises, institutions or organizations, and these transactions generally involve transfer of large funds, such as payment settlement between enterprises, supply chain finance, international trade payment, and the like. The online transaction refers to an electronic payment transaction performed through an online payment platform, and funds transfer between different bank accounts can be realized through the transaction performed through the platform, including a personal account and an enterprise account. The Unionpay transaction refers to a transaction performed by Chinese Unionpay, and can comprise various payment modes such as card swiping payment, mobile terminal payment, graphic code payment (such as one-dimensional code and two-dimensional code) and the like.
Optionally, after determining the funds pipeline extraction policy associated with the customer category attribute based on the preset category mapping table, further comprising: under the condition that the client category attribute is the personal category attribute, inquiring a preset category mapping table to obtain a second funds flow extraction strategy corresponding to the personal category attribute, wherein the second funds flow extraction strategy corresponds to a plurality of financial transaction channels; and extracting fund flowing water of P transaction dimensions corresponding to the plurality of financial transaction channels by the individual clients based on a second fund flowing water extraction strategy to obtain a second fund flowing water set, wherein the second fund flowing water set comprises: financial transaction channel, fund flow direction and transaction category, P is less than or equal to T.
For personal clients/general small clients, public transportation easy data are not available, and the clients need to extract specific channel business data, such as quick transaction channels (e.g. micro-channel transfer, branch-precious channel transfer, etc.) and consumption funds settlement data, so that the data can be quantitatively analyzed, meanwhile, no intersection overlap exists, account entry and account exit need to be respectively counted and input into a model, and the statistics of funds flows is ensured to have no redundant noise.
It should be noted that, the financial transaction channel in this embodiment refers to a way or platform for performing transactions in a financial market, for example, a fast transaction channel, a financial institution application APP, a financial institution APP, an ATM, a financial institution business hall, and the like. The money flow direction refers to the direction of flow of money in a financial market, and money may flow from an investor, institution or financial institution to various assets, such as stocks, bonds, foreign exchange, etc. Transaction categories refer to different transaction types in the financial market, common transaction categories including transfer transactions, deposit transactions, payroll transactions, stock transactions, bond transactions, futures transactions, foreign exchange transactions, and the like.
Step S103, integrating the fund running water sets of all the client category attributes to generate a running water index feature set.
Optionally, step S103 includes: acquiring a first fund running set corresponding to the legal category attribute, extracting fund running related to a specified financial product and/or a specified financial transaction channel from the first fund running set, and acquiring first running index characteristics based on the fund running statistics and account amount data, accumulated account amount data and ring ratio account increment data related to a target legal client in a first preset running period; acquiring a second fund running set corresponding to the personal category attribute, extracting fund running of each financial transaction channel in the second fund running set, and acquiring second running index characteristics based on the fund running statistics and the online banking payment amount data, the credit card payment amount data and the ring ratio payment amount increment data of target personal clients in a second preset running period; a set of flow index features is generated based on the first flow index feature and the first flow index feature.
In this embodiment, for legal clients and individual clients, the running water index features counted by the legal clients are different, and the legal clients refer to enterprises, institutions or organizations as clients to conduct financial transactions, where the running water index features of funds mainly include: 1. funds inflow: the running of funds for legal customers will typically include the inflow of funds in the form of business revenues, investment benefits, loans, or bond issues. 2. Funds outflow: the running of funds for legal customers will typically include the outflow of funds in the form of corporate payouts, loan repayment, equity or interest payments, etc. 3. Fund transfer frequency: legal customers' running water of funds typically shows a high frequency and large amount of money variation, reflecting the activity of the business operations. 4. Diversification of the fund flow direction: the running of funds by legal customers may involve multiple accounts, multiple financial products, and transaction channels, such as bank deposits, investment markets, bond markets, etc. In the embodiment, when the running index characteristics of legal clients are counted, the account entry amount, the account entry increment amount, the account exit amount and the account exit increment amount are counted mainly according to the fund movement of the legal clients in each time period. For example, account amounts accumulated by juriders of recent 1 month, accumulated account amounts by juriders of recent 3 months, increased account amounts by a period of recent 3 months, and the like are counted.
The individual customer means that the individual performs financial transactions as a customer, and the fund running index features mainly include: 1. payroll income: the running of funds for individual customers will typically include payroll revenues in the form of payouts, prizes, benefits, and the like. 2. Consumption expenditure: the funds flow of an individual customer will typically include funds flow in the form of daily life consumption, house rentals, educational expenses, medical expenses, and the like. 3. Deposit and withdrawal: the running of funds for individual customers typically reveals the movement of funds in the form of regular deposits, demand deposits, withdrawals, and the like. 4. Investment income: the individual customer's funds flow may display a trade of the investment market, such as stock exchanges, funds exchanges, bond exchanges, etc. In the embodiment, when the running index characteristics of the individual customers are counted, the running index characteristics of the individual customers are mainly counted according to the fund directions of the individual customers in different financial transaction channels in each time period, for example, the account paying amount of the individual customers for nearly 1 month on the internet, the account paying amount of the credit card for nearly 1 month, the increment of the payment amount for nearly 1 month on the internet, the accumulated running net amount of funds in the current month and the like are counted.
Step S104, the feature sequence of the water flowing index feature set is reordered, and an admittance feature image is generated based on the ordering result.
In this embodiment, the convolutional neural network is constructed by reordering the time dimension sequence of the original features, and taking different flows of the same time node as the minimum dimension convolutional unit. Wherein: at the same time node, if two groups of features including the credit amount of the credit card on the day and the credit amount of the debit card on the day are selected, the minimum dimension convolution unit is (2×1), 2 is corresponding to the two groups of features, 1 is corresponding to the selected time frame, namely 1 unit feature, and the two dimensions are one dimension. For example, 3 is selected as a corresponding time window, for example, a credit amount of 7 months 13 days, a credit amount of 7 months 14 days, and a credit amount of 7 months 15 days; a debit card credit amount for 7 months 13, a debit card credit amount for 7 months 14, a debit card credit amount for 7 months 15; such two sets of dimensional features constitute a new minimum convolution unit (2*3).
Optionally, step S104 includes: according to the time dimension, the running water index features of the running water index feature set with the same transaction dimension and different client category attributes are reordered in parallel; sequentially reordering the running water index features of different transaction dimensions and the same client category attribute in the running water index feature set according to the time sequence; and generating an admission characteristic image based on the sequencing result.
In the embodiment, the admittance feature images are constructed through feature sequence reordering, the attribute features of the same dimension and different categories are arranged in parallel according to the time dimension, and no sequence is required; the attribute features of the same category with different dimensions are required to be sequentially ordered according to time sequence.
Step S105, inputting the admission characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data.
Optionally, the network structure of the convolutional neural network includes: an input layer for inputting an admission characteristic image; the convolution layer slides a convolution kernel in the input admittance feature image through convolution operation, and the convolution kernel extracts the running water index feature; a pooling layer for reducing the spatial size of the access feature image; and the full connection layer is connected with the convolution layer and the pooling layer and outputs the asset state evaluation result of each target client.
Optionally, the convolution layer includes: the first sub-convolution layer comprises two groups of convolution kernels, wherein the convolution dimension of the convolution kernels is R transverse E longitudinal, R is the total number of client category attributes, and E is the time dimension; and the convolution dimension selection format of the Q+1 layers is R horizontal and vertical, and Q is an integer greater than 0.
In this embodiment, a convolutional neural network is built and model training is performed to obtain a convolutional neural network capable of evaluating the asset state of a client. The first group of convolution kernels of the convolution neural network needs to meet (N x), wherein N is the number of accumulation categories selected by the current model, the value is invariable, the value of x can be [1, the number of accumulation time points ], and when the number of time dimensions selected by the model is large and the convolution kernels are small, a pooling layer can be added to reduce the data operation dimension.
The time dimension can be set according to different client types, for example, the time dimension can be the week for legal clients, and the time dimension can be the day for personal clients.
The following detailed description is directed to alternative embodiments.
FIG. 2 is an alternative customer asset status assessment schematic based on multi-attribute trend magnitude derivation, according to an embodiment of the present invention, as shown in FIG. 2, comprising:
step 201, filtering out the flow index features of different service attributes.
According to clients with different types of attributes, different flow characteristics are adopted, such as financial services such as fund settlement, transaction and the like are mainly carried out through specific one or more cards in flow for legal clients, when the account numbers are taken, time point data of dimensions such as public transit easiness, internet connection transaction, silver connection transaction and the like can be extracted, and for general individual clients, public transit easiness data are not needed, specific channel service data and consumption fund settlement data are needed to be extracted, so that quantitative analysis of the data is ensured, meanwhile, intersection overlapping does not exist, account entering and account exiting are respectively counted and input, and the statistic of fund flows is ensured without redundant noise.
And 202, reordering the feature sequences to construct an admission feature image.
When the feature sequence reordering is carried out and an admission feature image is constructed, the attribute features of different categories with the same dimension are arranged in parallel according to the time dimension, and no sequence is required; the attribute features of the same category with different dimensions are required to be sequentially ordered according to time sequence.
And 203, setting corresponding super parameters, and constructing a convolutional neural network to perform model training.
Fig. 3 is a schematic diagram of an alternative convolutional neural network according to an embodiment of the present invention, where, as shown in fig. 3, the input is the aforementioned generated admission feature image (fig. 3 includes 7*N dimensions, and abcdefg represents class dimensions respectively), model training is performed by building the convolutional neural network, where, the first set of convolutional kernels need to satisfy (n×x), where n is the number of accumulated classes selected by the current model, the value of n is invariable, the value of x may be [1, the number of accumulated points ], and when the model is selected for a relatively long time and the convolutional kernels are smaller, a pooling layer may be added to reduce the data operation dimension, and finally, multi-attribute trend magnitude prediction is implemented, so as to obtain the asset status of each target client.
The specific structure of the convolutional neural network illustrated in fig. 3 includes:
(1) The first convolution calculation is performed with multiple sets of convolution kernels of different dimensions, the two sets of convolution kernels referred to in fig. 3 being (7 x 1) and (7 x 2).
(2) The subsequent convolution layer can use convolution kernels with different dimensions to process, and the subsequent convolution dimension selection format is (m multiplied by 1) due to the processing of the first layer convolution, and when more small-dimension convolutions are used, the pooling layer can be used for data dimension reduction.
(3) The last layer of the model is a full-connection layer, and the data trend magnitude is predicted by integrating all the groups of convolution post-processing data.
And 204, evaluating the asset states of all target clients by using a convolutional neural network.
According to the embodiment, the trend magnitude derivative features are introduced, so that multidimensional similar attribute category features can be effectively integrated, feature information is effectively compressed, the quality of the model is improved, data integration of data at different time points and similar attribute features is completed through compressing the information, the derivative features have higher information quantity, and the quality of the model is improved; meanwhile, the data usage amount of the model on the flow characteristics can be reduced, the number of the model is reduced, the complexity of the model is reduced, and the evaluation accuracy and the evaluation efficiency of the client asset are improved.
The following describes in detail another embodiment.
Example two
The supervision apparatus for customer financial assets provided in this embodiment includes a plurality of implementation units, each implementation unit corresponding to each implementation step in the above-mentioned embodiment.
FIG. 4 is a schematic diagram of an alternative monitoring device for a customer financial asset, as shown in FIG. 4, according to an embodiment of the invention, which may include: an acquisition unit 41, a determination unit 42, a generation unit 43, a reordering unit 44, an evaluation unit 45.
The acquiring unit 41 is configured to acquire, for each financial service in the financial service set, N client class attributes related to the financial service, where the financial service set includes M financial services, and N and M are both positive integers;
a determining unit 42, configured to determine, for each client category attribute, a fund flowing extraction policy associated with the client category attribute based on a preset category mapping table, where the fund flowing extraction policy is used to extract a fund flowing of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2;
a generating unit 43, configured to integrate the fund running water sets of all the client category attributes to generate a running water index feature set;
A reordering unit 44, configured to reorder the feature sequences of the running water index feature sets, and generate an admission feature image based on the ordering result;
an evaluation unit 45, configured to input the admission feature image into a pre-trained convolutional neural network, and output an asset status evaluation result of each target client, where the asset status evaluation result includes: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data.
In the above supervision apparatus for customer financial assets, N customer category attributes related to a financial service may be acquired for each financial service in a financial service set by the acquisition unit 41, a fund running water extraction policy associated with the customer category attribute is determined for each customer category attribute by the determination unit 42 based on a preset category mapping table, the fund running water extraction policy is used for extracting fund running water of each target customer in T transaction dimensions under the customer category attribute, a running water index feature set is generated by integrating the fund running water sets of all the customer category attributes by the generation unit 43, feature sequence reordering is performed on the running water index feature set by the reordering unit 44, and an admission feature image is generated based on the ordering result; inputting the admission characteristic image into a pre-trained convolutional neural network through an evaluation unit 45, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data. In the embodiment, a trend magnitude derivative strategy of the multidimensional attribute is introduced, the fund running water information of the multidimensional transaction dimension is extracted (the data information is high), the fund running water sets of the multidimensional client category attribute are synthesized to generate a running water index feature set, the feature set is sequentially reordered, and then the asset state evaluation result of each client is analyzed through a convolutional neural network, so that whether the client asset data is possibly violated or not can be effectively monitored, the default rate is reduced, and the asset quality rate of a financial institution is improved, and the technical problems that in the related technology, when the client financial asset supervision is carried out, the information quantity is only a single specific attribute, and the data information quantity is lower are solved.
Optionally, the supervision apparatus for the customer financial asset further comprises: the first query unit is used for querying the preset category mapping table to obtain a first funds running extraction strategy corresponding to the legal category attribute under the condition that the client category attribute is the legal category attribute after determining the funds running extraction strategy related to the client category attribute based on the preset category mapping table; the first extraction unit is configured to extract a running line of funds in U transaction dimensions of a financial account related to a legal person client based on a first running line of funds extraction policy, so as to obtain a first running line set of funds, where the transaction dimensions include: for public transactions, online transactions and banking transactions, the first fund running set comprises a fund flow direction, a transaction time point and a financial account number, and U is smaller than or equal to T.
Optionally, the supervision apparatus for the customer financial asset further comprises: the second query unit is used for querying the preset category mapping table to obtain a second funds running extraction strategy corresponding to the personal category attribute under the condition that the client category attribute is the personal category attribute after determining the funds running extraction strategy related to the client category attribute based on the preset category mapping table, wherein the second funds running extraction strategy corresponds to a plurality of financial transaction channels; the second extraction unit is configured to extract a running line of funds of the individual customer in P transaction dimensions corresponding to the plurality of financial transaction channels based on a second running line of funds extraction policy, to obtain a second running line set of funds, where the second running line set of funds includes: financial transaction channel, fund flow direction and transaction category, P is less than or equal to T.
Optionally, the generating unit includes: the first acquisition module is used for acquiring a first fund running water set corresponding to the legal category attribute, extracting fund running water related to appointed financial products and/or appointed financial transaction channels from the first fund running water set, and acquiring first running water index characteristics based on the fund running water statistics, accounting amount data related to target legal customers, accumulated accounting amount data and loop ratio accounting increment data in a first preset running water period; the second acquisition module is used for acquiring a second fund running water set corresponding to the personal category attribute, extracting fund running water related to each financial transaction channel in the second fund running water set, and acquiring second running water index characteristics based on the fund running water statistics and the online banking payment amount data, the credit card payment amount data and the ring ratio payment amount increment data related to the target personal client in a second preset running water period; the first generation module is used for generating a running water index feature set based on the first running water index feature and the first running water index feature.
Optionally, the reordering unit includes: the first reordering module is used for reordering the running water index features of the running water index feature set in the same transaction dimension and different client category attributes in parallel according to the time dimension; the second reordering module is used for sequentially reordering the running water index features with different transaction dimensions and the same client category attribute in the running water index feature set according to the time sequence; and the second generation module is used for generating an admission characteristic image based on the sequencing result.
Optionally, the network structure of the convolutional neural network includes: an input layer for inputting an admission characteristic image; the convolution layer slides a convolution kernel in the input admittance feature image through convolution operation, and the convolution kernel extracts the running water index feature; a pooling layer for reducing the spatial size of the access feature image; and the full connection layer is connected with the convolution layer and the pooling layer and outputs the asset state evaluation result of each target client.
Optionally, the convolution layer includes: the first sub-convolution layer comprises two groups of convolution kernels, wherein the convolution dimension of the convolution kernels is R transverse E longitudinal, R is the total number of client category attributes, and E is the time dimension; and the convolution dimension selection format of the Q+1 layers is R horizontal and vertical, and Q is an integer greater than 0.
The above-mentioned supervision apparatus for customer financial assets may further include a processor and a memory, where the above-mentioned acquisition unit 41, determination unit 42, generation unit 43, reordering unit 44, evaluation unit 45, etc. are stored as program units, and the processor executes the above-mentioned program units stored in the memory to implement the corresponding functions.
The processor includes a kernel, and the kernel fetches a corresponding program unit from the memory. The kernel can set one or more than one, financial fund flowing water of the clients is extracted by adjusting kernel parameters, and the received characteristic images associated with the fund flowing water are input into a pre-trained convolutional neural network to obtain asset state evaluation results of all target clients.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), which includes at least one memory chip.
The present application also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with the method steps of: for each financial service in a financial service set, acquiring N client class attributes related to the financial service, wherein the financial service set comprises M financial services, and N and M are positive integers; for each client category attribute, determining a fund running water extraction strategy associated with the client category attribute based on a preset category mapping table, wherein the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2; integrating the fund running water sets of all the client category attributes to generate a running water index feature set; feature sequence reordering is carried out on the running water index feature set, and an access feature image is generated based on the ordering result; inputting the admittance characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the comparison data of the asset accounting data of the target client at the current time point and the historical asset accounting data of the last time point, and the client asset state generated based on the comparison data.
According to another aspect of the embodiments of the present invention, there is also provided a computer readable storage medium, the computer readable storage medium including a stored computer program, wherein the computer readable storage medium is controlled to execute the method of supervising a customer financial asset according to any one of the above embodiments when the computer program is run.
According to another aspect of embodiments of the present invention, there is also provided an electronic device comprising one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of supervising a customer financial asset according to any of the above embodiments.
Fig. 5 is a block diagram of a hardware architecture of an electronic device (or mobile device) for a method of policing a customer's financial assets, in accordance with an embodiment of the present invention. As shown in fig. 5, the electronic device may include one or more (shown in fig. 5 as 502a, 502b, … …,502 n) processors 502 (the processors 502 may include, but are not limited to, a microprocessor MCU, a programmable logic device FPGA, etc. processing means), a memory 504 for storing data. In addition, the method may further include: a display, an input/output interface (I/O interface), a Universal Serial Bus (USB) port (which may be included as one of the ports of the I/O interface), a network interface, a keyboard, a power supply, and/or a camera. It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 5 is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, the electronic device may also include more or fewer components than shown in FIG. 5, or have a different configuration than shown in FIG. 5.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims (10)

1. A method of policing a customer's financial assets, comprising:
for each financial service in a financial service set, acquiring N client class attributes related to the financial service, wherein the financial service set comprises M financial services, and N and M are positive integers;
for each client category attribute, determining a fund running water extraction strategy associated with the client category attribute based on a preset category mapping table, wherein the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2;
integrating the fund running water sets of all the client category attributes to generate a running water index feature set;
performing feature sequence reordering on the running water index feature set, and generating an admittance feature image based on the ordering result;
inputting the access characteristic image into a pre-trained convolutional neural network, and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises: the target client is compared with the asset accounting data of the previous time point and the client asset state generated based on the compared data.
2. The method of supervising of claim 1, further comprising, after determining a funds pipeline extraction policy associated with the customer category attribute based on a preset category mapping table:
under the condition that the client category attribute is a legal category attribute, inquiring the preset category mapping table to obtain a first fund running water extraction strategy corresponding to the legal category attribute;
and based on the first fund running water extraction strategy, extracting fund running water of financial accounts related to legal customers in U transaction dimensions to obtain a first fund running water set, wherein the transaction dimensions comprise: for public transactions, online transactions and banking transactions, the first fund running set comprises a fund flow direction, a transaction time point and a financial account number, and U is smaller than or equal to T.
3. The method of supervising of claim 1, further comprising, after determining a funds pipeline extraction policy associated with the customer category attribute based on a preset category mapping table:
querying the preset category mapping table under the condition that the client category attribute is a personal category attribute to obtain a second funds running extraction strategy corresponding to the personal category attribute, wherein the second funds running extraction strategy corresponds to a plurality of financial transaction channels;
And extracting fund flowing of the individual client in P transaction dimensions corresponding to a plurality of financial transaction channels based on the second fund flowing extraction strategy to obtain a second fund flowing set, wherein the second fund flowing set comprises: financial transaction channel, fund flow direction and transaction category, P is less than or equal to T.
4. The method of supervising according to claim 1, wherein the step of generating a running water index feature set by integrating the fund running water sets of all of the customer category attributes comprises:
acquiring a first fund running set corresponding to a legal category attribute, extracting fund running related to a specified financial product and/or a specified financial transaction channel from the first fund running set, and acquiring first running index characteristics based on the fund running statistics and accounting amount data, accumulated accounting amount data and ring ratio accounting increment data related to a target legal client in a first preset running period;
acquiring a second fund running set corresponding to the personal category attribute, extracting fund running of each financial transaction channel in the second fund running set, and acquiring second running index characteristics based on the fund running statistics and internet banking payment amount data, credit card payment amount data and ring ratio payment amount increment data of target personal clients in a second preset running period;
The set of running water index features is generated based on the first running water index feature and the first running water index feature.
5. The method of supervising according to claim 1, wherein the step of reordering the feature sequences of the set of flowing water index features and generating an admission feature image based on the ordering result comprises:
according to the time dimension, the running water index features of the running water index feature set with the same transaction dimension and different client category attributes are reordered in parallel;
sequentially reordering the running water index features with different transaction dimensions and the same client category attribute in the running water index feature set according to the time sequence;
and generating an admission characteristic image based on the sequencing result.
6. The policing method of claim 1 wherein the network structure of the convolutional neural network comprises:
an input layer for inputting the admittance feature image;
the convolution layer slides a convolution kernel in the input admittance feature image through convolution operation, and the convolution kernel extracts running water index features;
a pooling layer for reducing the spatial size of the access feature image;
and the full connection layer is connected with the convolution layer and the pooling layer and outputs the asset state evaluation result of each target client.
7. The policing method of claim 6, characterized in that the convolutional layer comprises:
the first sub-convolution layer comprises two groups of convolution kernels, wherein the convolution dimension of the convolution kernels is R transverse E longitudinal, R is the total number of client category attributes, and E is the time dimension;
and the convolution dimension selection format of the Q+1 layers is R horizontal and vertical, and Q is an integer greater than 0.
8. A supervision apparatus for a customer financial asset, comprising:
the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring N client class attributes related to each financial business in a financial business set, wherein the financial business set comprises M financial businesses, and N and M are positive integers;
the determining unit is used for determining a fund running water extraction strategy associated with the client category attribute based on a preset category mapping table for each client category attribute, wherein the fund running water extraction strategy is used for extracting fund running water of each target client in T transaction dimensions under the client category attribute, and T is a positive integer greater than 2;
the generation unit is used for integrating the fund running water sets of all the client category attributes to generate a running water index feature set;
the reordering unit is used for reordering the feature sequences of the flowing water index feature sets and generating an admission feature image based on the ordering result;
The evaluation unit is used for inputting the admission characteristic image into a pre-trained convolutional neural network and outputting an asset state evaluation result of each target client, wherein the asset state evaluation result comprises the following components: the target client is compared with the asset accounting data of the previous time point and the client asset state generated based on the compared data.
9. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored computer program, wherein the computer program, when run, controls a device in which the computer readable storage medium is located to perform the method of supervision of a customer financial asset according to any one of claims 1 to 7.
10. An electronic device comprising one or more processors and memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of policing customer financial assets of any one of claims 1 to 7.
CN202311491663.3A 2023-11-09 2023-11-09 Supervision method and device for customer financial assets and electronic equipment Pending CN117475231A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311491663.3A CN117475231A (en) 2023-11-09 2023-11-09 Supervision method and device for customer financial assets and electronic equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311491663.3A CN117475231A (en) 2023-11-09 2023-11-09 Supervision method and device for customer financial assets and electronic equipment

Publications (1)

Publication Number Publication Date
CN117475231A true CN117475231A (en) 2024-01-30

Family

ID=89634480

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311491663.3A Pending CN117475231A (en) 2023-11-09 2023-11-09 Supervision method and device for customer financial assets and electronic equipment

Country Status (1)

Country Link
CN (1) CN117475231A (en)

Similar Documents

Publication Publication Date Title
US10360575B2 (en) Consumer household spend capacity
Ericsson et al. A framework for valuing corporate securities
US8301530B2 (en) Automatic savings program
US7970676B2 (en) Method and system for modeling future action impact in credit scoring
CN108416664A (en) Methods of risk assessment and system based on consumptive credit scene are realized
US20080221990A1 (en) Estimating the spend capacity of consumer households
Okoro Impact of electronic banking instruments on the intermediation efficiency of the Nigerian economy
WO2017133456A1 (en) Method and device for determining risk evaluation parameter
MX2007012294A (en) Method and apparatus for rating asset-backed securities.
US20210398210A1 (en) Systems and methods of transaction tracking and analysis for near real-time individualized credit scoring
Bhat et al. The impact of risk modeling on the market perception of banks’ estimated fair value gains and losses for financial instruments
CN113421154B (en) Credit risk assessment method and system based on control chart
CN110458684A (en) A kind of anti-fraud detection method of finance based on two-way shot and long term Memory Neural Networks
Wattanawongwan et al. A mixture model for credit card exposure at default using the GAMLSS framework
CN109360085A (en) A kind of bank client responsible investigation method and system
CN117475231A (en) Supervision method and device for customer financial assets and electronic equipment
WO2022136692A1 (en) Method for calculating at least one score representative of a probable activity breakage of a merchant, system, apparatus and corresponding computer program
CN113807943A (en) Multi-factor valuation method, system, medium and equipment for bad assets
CN111563747B (en) Processing method and device for merchant credit
Budd Modelling credit card usage for individual card-holders
CN112116169B (en) User behavior determining method and device and electronic equipment
CN112037013A (en) Pedestrian credit variable derivation method and device
Чернецька et al. Bаnkruptcy of commerciаl bаnks: cаuses аnd wаys to prevent
Parkavi et al. Survey on Mortgage-Backed Securities Prepayment Risk Using Machine Learning Models
CN110135973A (en) A kind of intelligent credit method based on IM and intelligent credit device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination