CN111401908A - Transaction behavior type determination method, device and equipment - Google Patents
Transaction behavior type determination method, device and equipment Download PDFInfo
- Publication number
- CN111401908A CN111401908A CN202010167428.0A CN202010167428A CN111401908A CN 111401908 A CN111401908 A CN 111401908A CN 202010167428 A CN202010167428 A CN 202010167428A CN 111401908 A CN111401908 A CN 111401908A
- Authority
- CN
- China
- Prior art keywords
- transaction
- transaction data
- data
- user
- layer
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
- G06Q20/401—Transaction verification
- G06Q20/4016—Transaction verification involving fraud or risk level assessment in transaction processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Finance (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Computer Security & Cryptography (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The transaction behaviors of a user in a preset time period are depicted, the change of the behaviors of the user in different time dimensions is captured, a hidden vector generation model obtained through automatic encoder training is used, a hidden vector expressing a user behavior mode is learned, the hidden vector is input into a transaction behavior recognition model, and the transaction behavior type of the user can be recognized according to the behavior change of the user in different time dimensions.
Description
Technical Field
The present disclosure relates to the field of data processing technologies, and in particular, to a method, an apparatus, and a device for determining a transaction behavior type.
Background
With the gradual expansion of the internet technology to the financial field, on the basis of providing a relatively convenient payment channel for the masses, favorable conditions are provided for the implementation of various illegal behaviors, and one of the network money laundering is provided. In order to timely discover and process such illegal transactions, certain anti-money laundering measures are required.
Disclosure of Invention
In view of this, one or more embodiments of the present disclosure provide a method, an apparatus, and a device for determining a transaction behavior type, so as to solve the problem that local features of a transaction behavior of a user cannot be effectively obtained when the transaction behavior of the user is identified through a neural network model in the related art.
One or more embodiments of the present specification provide a transaction behavior type determination method, including:
acquiring transaction data of at least one user in a first preset time period;
inputting the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, wherein the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
inputting the hidden vector into a transaction behavior type identification model, and outputting the transaction behavior type of the at least one user in the first preset time period.
One or more embodiments of the present specification further provide a transaction behavior type determination device, including:
the system comprises a first acquisition module, a second acquisition module and a processing module, wherein the first acquisition module is used for acquiring transaction data of at least one user in a first preset time period;
the first input module is used for inputting the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, wherein the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
and the second input module is used for inputting the hidden vector into a transaction behavior type identification model and outputting the transaction behavior type of the at least one user in the first preset time period.
One or more embodiments of the present specification also provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement any one of the above-described transaction behavior type determination methods.
As can be seen from the above, the transaction behavior type determining method according to one or more embodiments of the present disclosure describes the transaction behavior of the user in a preset time period, can capture the behavior changes of the user in different time dimensions, uses the hidden vector generation model obtained through the automatic encoder training, can learn the hidden vector expressing the user behavior pattern without marking the user, inputs the hidden vector into the transaction behavior recognition model, can recognize the user transaction behavior type according to the behavior changes of the user in different time dimensions, simplifies the user transaction behavior recognition process, and improves the user transaction behavior recognition efficiency.
Drawings
In order to more clearly illustrate one or more embodiments or prior art solutions of the present specification, the drawings that are needed in the description of the embodiments or prior art will be briefly described below, and it is obvious that the drawings in the following description are only one or more embodiments of the present specification, and that other drawings may be obtained by those skilled in the art without inventive effort from these drawings.
FIG. 1 is a schematic diagram illustrating a convolutional neural network model in accordance with an exemplary embodiment;
FIG. 2 is a flow diagram illustrating a transaction behavior type determination method in accordance with an exemplary embodiment;
FIG. 3 is a block diagram illustrating a transaction behavior type determination device in accordance with an exemplary embodiment;
fig. 4 is a schematic diagram illustrating a more specific hardware architecture of an electronic device according to one or more embodiments of the present disclosure.
Detailed Description
For the purpose of promoting a better understanding of the objects, aspects and advantages of the present disclosure, reference is made to the following detailed description taken in conjunction with the accompanying drawings.
It should be noted that all expressions using "first" and "second" in one or more embodiments of the present specification are intended to distinguish two entities with the same name but different names or different parameters, and it is to be understood that "first" and "second" are merely for convenience of expression and should not be construed as limitations on one or more embodiments of the present specification, and they are not described in any more detail in the following embodiments.
In order to identify illegal transaction behaviors, a neural network model can be used for learning the expression of a user time behavior sequence, L STM (L ong Short-Term Memory network) or an encoder or decoder of a self-attention mechanism is used, wherein the L STM emphasizes long-dependent learning in the sequence, a special gate structure can selectively retain or discard historical information, the self-attention mechanism can obtain the relation between characteristics of a time position and other characteristics of the time position, but only can retain important information on the time sequence and cannot capture local information, namely information of the behavior sequence of the user in a certain period of time, in the scene of identifying the transaction behavior types of the user, some local information possibly can be suspicious transaction behaviors and is important for the transaction behavior patterns of the user, and in the scene of identifying the transaction behavior types of the user, a convolutional neural network (algorithm for classifying texts) model is used for directly classifying the user time sequence, intermediate vectors are used as the expression mode of the time sequence, a plurality of marking modes of the user are required to be used for learning the time sequence of the user, and a plurality of the relevant behavior identification modes of the user are effectively used for providing a plurality of the relevant transaction behavior recognition cases.
In one or more of the descriptions, a convolutional neural network model is adopted to obtain the local transaction behavior characteristics of the user, so that the change of the transaction behavior pattern of the user on a time axis is captured; the method comprises the steps that an AutoEncoder (self-encoding) is used for expressing a time sequence of user behaviors into a hidden vector (hereinafter referred to as hidden vector), the hidden vector not only contains internal characteristics in time and removes noise, but also can be used in a user risk classification scene, for example, the hidden vector is input into a user risk classification model, and the risk classification of the model output to a user can be obtained; meanwhile, the AutoEncoder is used for unsupervised learning, marking on a user is not needed, and the application range is wide. As shown in fig. 1, the convolutional neural network model mainly comprises an input layer, an encoder, a decoder and an output layer, where the input of the model is a behavior description of a user for a period of time, the behavior description can be formed into a user behavior map, in the user behavior map shown in fig. 1, T-1 to T-90 represent 90 days, and features F1 to F128 represent 128 user transaction behavior features, the behavior description is input into the encoder, for example, through convolution and pooling operations, to form hidden vectors, the decoder reduces the hidden vectors into the same size of the behavior map as the input by using deconvolution and inverse pooling, the convolutional neural network model aims to minimize root mean square error of the input and output, thereby obtaining hidden vectors expressing user behavior information by using a successfully trained model, and applies the hidden vectors to the recognition of the user behavior, transaction behaviors such as money laundering and normal transaction can be effectively identified.
Fig. 2 is a flow diagram illustrating a transaction behavior type determination method according to one or more embodiments of the present description, as shown in fig. 2, the method including:
step 202: acquiring transaction data of at least one user in a first preset time period;
for example, when a user who is not normally transacted is identified from a large number of users, transaction data of all users to be identified in a first preset time period may be acquired, the transaction data may include, for example, transaction data generated by the user at each sampling time point in the first preset time period, and the type of the transaction data may include, for example, at least one of the number of funds inflows, the number of funds outflows, the amount of funds inflows, and the amount of funds outflows.
Step 204: inputting the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, wherein the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
the hidden vector can be a one-dimensional vector, can represent and accurately depict the density of the user transaction data sample, and is a depiction of the user sequence behavior, and can reflect the preferred/customary behavior mode of each user.
For example, a matrix converted by user transaction data is input into a hidden vector generation model, and a hidden vector related to user transaction behaviors is output. Examples of the automatic Encoder include an AutoEncoder or a VAE (variant automatic Encoder).
Step 206: inputting the hidden vector into a transaction behavior type identification model, and outputting the transaction behavior type of the at least one user in the first preset time period.
The behavior type recognition model can be obtained by training based on a plurality of transaction data which are marked as different transaction behavior types in advance, for example, the transaction data of known transaction behavior types and marked with transaction behavior type labels are input into a neural network model in advance, the neural network model is continuously trained to obtain the transaction behavior type recognition model, the transaction data of unknown transaction behavior types are input into the model, and the transaction behavior types to which the transaction data belong can be output. The transaction behavior types may be classified into, for example, a normal transaction behavior and an abnormal transaction behavior, or may be classified into a money laundering behavior and a normal transaction behavior, or may be further classified into a plurality of types classified according to the level of the transaction risk of the user, for example, a high-risk transaction behavior, a medium-risk transaction behavior and a low-risk transaction behavior.
In an example, assuming that the at least one user is 1000 users, after the hidden vector obtained in step 206 is input into the transaction behavior type identification model, the transaction behavior types of the 1000 users within the first preset time period output by the model may be obtained, for example, the users may be classified according to the transaction behavior types of the users, so that the users performing illegal transactions may be determined according to the transaction behavior types of the users, or the preference of the users may be determined according to the transaction behavior types of the users, so that messages may be pushed to the users according to the preference of the users.
The transaction behavior type determining method provided by one or more embodiments of the present specification describes a transaction behavior of a user in a preset time period, can capture a change of a behavior of the user in different time dimensions, uses a hidden vector generation model obtained through automatic encoder training, can learn a hidden vector expressing a user behavior pattern without marking the user, inputs the hidden vector into a transaction behavior recognition model, can recognize a user transaction behavior type according to the behavior change of the user in different time dimensions, simplifies a user transaction behavior recognition process, and improves user transaction behavior recognition efficiency.
In one implementation, still taking the convolutional neural network model shown in fig. 1 as an example, the automatic encoder may include: the system comprises a time sequence characteristic layer, an encoder, a first full connection layer, a decoder, a second full connection layer and an output layer; the time sequence feature layer is used for converting the first transaction data into user time sequence data according to the type of the first transaction data and the time sequence of the generation of the first transaction data, for example, converting the transaction data of the user into a matrix according to the type of the transaction data of the user and the time sequence of the generation of the transaction data; the time series data of the user can be represented by time series, for example, the time series is represented by vectors of behaviors of the user for a period of time, for example, in a period of 7 days, the amount of money, the number of strokes, the number of opponents, the time and the like of each day are formed into vectors, and 7 such vectors are obtained, namely, the time series of 7 days. For example, the transaction data may be formed into a user behavior graph according to the type of the transaction data and the time sequence of the transaction data, wherein the horizontal axis of the graph is used for representing the characteristics (N dimension) of each day, such as the type of the transaction data, and the vertical axis is used for representing the time (T dimension), and each user may correspond to a graph (T x N dimension), i.e., a matrix corresponding to T x N dimensions, i.e., a user transaction data sample. The encoder is used for acquiring the characteristics of the user time sequence data under different time windows and pooling the characteristics to obtain a pooling result; for example, local information of transaction data is extracted by TextCNN: setting different sizes of the convolution kernels, for example, setting the convolution kernel to be 2 can capture the change of local transaction behaviors of the user in adjacent 2 days, setting the convolution kernel to be 6 can capture the change of local behavior changes of the user in adjacent 6 days, and so on, so that the learning of the long-term and short-term behavior sequences of the user can be realized through the combination of different sizes of the convolution kernels. The first full-connection layer is used for connecting the pooling results to obtain the hidden vector; the decoder is used for converting the implicit vector into second transaction data which is consistent with the first transaction data type; the decoder may include, for example, a deconvolution layer that transforms a one-dimensional hidden vector into a matrix, for example, by padding the hidden vector (e.g., with 0) into a matrix, and then summing the matrix with a convolution kernel to obtain a matrix: changing a certain value in the matrix into several values, for example, four values, which are the same, so that the size of the matrix becomes larger further, and a new matrix is obtained, and the number of rows of the matrix is the same as the number of rows of the input. The second full connection layer is used for converting the second transaction data into the size consistent with the first transaction data; for example, the second fully-connected layer may transform the matrix obtained by the decoder to conform to the input matrix size; the output layer is used for outputting the converted second transaction data. The first transaction data is input data of the automatic encoder, and the second transaction data is output data of the automatic encoder.
In one implementation, the transaction behavior type determination method may further include: training the autoencoder includes: and sequentially inputting a plurality of groups of different transaction data samples into the automatic encoder, training the automatic encoder until the error between the output data and the input data of the automatic encoder is less than a preset value, and obtaining the trained automatic encoder. For example, the root mean square error between the output data and the input data of the automatic encoder may be trained to be smaller than a preset value, or the average playback error between the output data and the input data of the automatic encoder may be trained to be smaller than a preset value, and the training is ended to obtain the network parameters. And each group of user time sequence sample data in the multiple groups of user time sequence data samples is data used for training the model once.
In one implementation, the hidden vector generation model may include: the trained sequential feature layer, the encoder, and the first fully-connected layer in the auto-encoder.
In one implementation, the encoder may include a convolutional layer and a pooling layer; the convolution layer is configured to obtain characteristics of the user time series data in a second preset time period according to user time series data corresponding to the second time period (for example, the user time series data obtained by converting transaction data generated in the second time period by the time series characteristic layer), for example, the fund inflow number, the fund outflow number, the fund inflow amount and the fund outflow amount of each user may be respectively obtained, the second preset time period is smaller than the first preset time period, for example, the first preset time period may include a plurality of second preset time periods; the Pooling layer is used to maximize Pooling (Max-Pooling) and average Pooling (Avg-Pooling) of features obtained from the convolutional layer, resulting in a maximum Pooling result and an average Pooling result. Wherein, Max-Pooling can be used for reserving the main state of the change of the user transaction behavior, and Avg-Pooling can be used for reserving the average state of the user transaction behavior characteristics.
In one implementation, the width of the convolution kernel of the convolutional layer is consistent with the dimension of the type of the transaction data, so that the feature of the transaction behavior needing to be extracted can be ensured not to be omitted. For example, the types of transaction data include: the number of funds in-flowed, the number of funds out-flowed, the amount of funds in-flowed, and the amount of funds out-flowed, the width of the convolution kernel of the convolutional layer may be set to 4 accordingly.
In one implementation, the transaction action types may include money laundering actions and non-money laundering actions; based on this, after transaction data of unknown transaction behavior types are input into the transaction behavior type identification model, the model can identify the input transaction data as money laundering behavior or non-money laundering behavior according to the characteristics of user behavior embodied in the transaction data. The types of transaction data include at least: one of a number of funds in-flows, a number of funds out-flows, an amount of funds in-flows, and an amount of funds out-flows. Based on the above, the transaction behavior type identification model can identify the user transaction behavior type corresponding to the input transaction data according to the behavior characteristics of the user embodied in the transaction data.
Fig. 3 is a block diagram illustrating a transaction behavior type determination apparatus according to an exemplary embodiment, and as shown in fig. 3, the apparatus 30 includes:
a first obtaining module 32, configured to obtain transaction data of at least one user within a first preset time period;
a first input module 34, configured to input the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, where the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
and a second input module 36, configured to input the hidden vector into a transaction behavior type identification model, and output a transaction behavior type of the at least one user within the first preset time period.
In one implementation, the auto-encoder may include: the system comprises a time sequence characteristic layer, an encoder, a first full connection layer, a decoder, a second full connection layer and an output layer; the time sequence characteristic layer is used for converting first transaction data into user time sequence data according to the type of the first transaction data and the time sequence generated by the first transaction data; the encoder is used for acquiring the characteristics of the user time sequence data under different time windows and pooling the characteristics to obtain a pooling result; the first full-connection layer is used for connecting the pooling results to obtain the hidden vector; the decoder is used for converting the implicit vector into second transaction data which is consistent with the first transaction data type; the second full connection layer is used for converting the second transaction data into the size consistent with the first transaction data; the output layer is used for outputting the converted second transaction data. The first transaction data is input data of the automatic encoder, and the second transaction data is output data of the automatic encoder.
In one implementation, the transaction behavior type determining apparatus may further include: and the first training module is used for sequentially inputting a plurality of groups of different transaction data samples into the automatic encoder and training the automatic encoder until the error between the output data and the input data of the automatic encoder is smaller than a preset value, so that the trained automatic encoder is obtained.
In one implementation, the hidden vector generation model may include: the trained sequential feature layer, the encoder, and the first fully-connected layer in the auto-encoder.
In one implementation, the encoder includes a convolutional layer and a pooling layer; the convolutional layer is used for acquiring the characteristics of each type of user time sequence data in a second preset time period according to the user time sequence data corresponding to the second time period, wherein the second preset time period is smaller than the first preset time period; the pooling layer is configured to maximally pool and average pool the features obtained from the convolutional layer to obtain a maximum pooling result and an average pooling result.
In one implementation, the width of the convolution kernel of the convolutional layer is consistent with the dimension of the type of the transaction data.
In one implementation, the transaction action types include money laundering actions and non-money laundering actions; the types of transaction data include at least: one of a number of funds in-flows, a number of funds out-flows, an amount of funds in-flows, and an amount of funds out-flows.
It should be noted that the method of one or more embodiments of the present disclosure may be performed by a single device, such as a computer or server. The method of the embodiment can also be applied to a distributed scene and completed by the mutual cooperation of a plurality of devices. In such a distributed scenario, one of the devices may perform only one or more steps of the method of one or more embodiments of the present disclosure, and the devices may interact with each other to complete the method.
The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
Fig. 4 is a schematic diagram illustrating a more specific hardware structure of an electronic device according to one or more embodiments of the present disclosure, where, as shown in fig. 4, the device may include: a processor 1010, a memory 1020, an input/output interface 1030, a communication interface 1040, and a bus 1050. Wherein the processor 1010, memory 1020, input/output interface 1030, and communication interface 1040 are communicatively coupled to each other within the device via bus 1050.
The processor 1010 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute related programs to implement the technical solutions provided in the embodiments of the present disclosure.
The Memory 1020 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random access Memory), a static storage device, a dynamic storage device, or the like. The memory 1020 may store an operating system and other application programs, and when the technical solution provided by the embodiments of the present specification is implemented by software or firmware, the relevant program codes are stored in the memory 1020 and called to be executed by the processor 1010.
The input/output interface 1030 is used for connecting an input/output module to input and output information. The i/o module may be configured as a component in a device (not shown) or may be external to the device to provide a corresponding function. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.
The communication interface 1040 is used for connecting a communication module (not shown in the drawings) to implement communication interaction between the present apparatus and other apparatuses. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, Bluetooth and the like).
It should be noted that although the above-mentioned device only shows the processor 1010, the memory 1020, the input/output interface 1030, the communication interface 1040 and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will appreciate that the above-described apparatus may also include only those components necessary to implement the embodiments of the present description, and not necessarily all of the components shown in the figures.
Computer-readable media of the present embodiments, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.
Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, is limited to these examples; within the idea of the present disclosure, features in the above embodiments or in different embodiments may also be combined, steps may be implemented in any order, and there are many other variations of the different aspects of the present disclosure as described above, which are not provided in detail for the sake of brevity.
In addition, well known power/ground connections to Integrated Circuit (IC) chips and other components may or may not be shown in the provided figures for simplicity of illustration and discussion, and so as not to obscure the disclosure. Furthermore, devices may be shown in block diagram form in order to avoid obscuring the disclosure, and also in view of the fact that specifics with respect to implementation of such block diagram devices are highly dependent upon the platform within which the present disclosure is to be implemented (i.e., specifics should be well within purview of one skilled in the art). Where specific details (e.g., circuits) are set forth in order to describe example embodiments of the disclosure, it should be apparent to one skilled in the art that the disclosure can be practiced without, or with variation of, these specific details. Accordingly, the description is to be regarded as illustrative instead of restrictive.
While the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those of ordinary skill in the art in light of the foregoing description. For example, other memory architectures (e.g., dynamic ram (dram)) may use the discussed embodiments.
The embodiments of the present disclosure are intended to embrace all such alternatives, modifications and variances that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalents, improvements, and the like that may be made within the spirit and principles of the disclosure are intended to be included within the scope of the disclosure.
Claims (15)
1. A transaction behavior type determination method, comprising:
acquiring transaction data of at least one user in a first preset time period;
inputting the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, wherein the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
inputting the hidden vector into a transaction behavior type identification model, and outputting the transaction behavior type of the at least one user in the first preset time period.
2. The method of claim 1, further comprising:
training the autoencoder includes:
and sequentially inputting a plurality of groups of different transaction data samples into the automatic encoder, training the automatic encoder until the error between the output data and the input data of the automatic encoder is less than a preset value, and obtaining the trained automatic encoder.
3. The method of claim 2, the auto-encoder, comprising:
the system comprises a time sequence characteristic layer, an encoder, a first full connection layer, a decoder, a second full connection layer and an output layer;
the time sequence characteristic layer is used for converting first transaction data into user time sequence data according to the type of the first transaction data and the time sequence generated by the first transaction data;
the encoder is used for acquiring the characteristics of the user time sequence data under different time windows and pooling the characteristics to obtain a pooling result;
the first full-connection layer is used for connecting the pooling results to obtain the hidden vector;
the decoder is used for converting the implicit vector into second transaction data which is consistent with the first transaction data type;
the second full connection layer is used for converting the second transaction data into the size consistent with the first transaction data;
the output layer is used for outputting the converted second transaction data, wherein the first transaction data is input data of the automatic encoder, and the second transaction data is output data of the automatic encoder.
4. The method of claim 3, the hidden vector generation model comprising:
the trained sequential feature layer, the encoder, and the first fully-connected layer in the auto-encoder.
5. The method of claim 3, the encoder comprising a convolutional layer and a pooling layer;
the convolutional layer is used for acquiring the characteristics of each type of user time sequence data in a second preset time period according to the user time sequence data corresponding to the second time period, wherein the second preset time period is smaller than the first preset time period;
the pooling layer is configured to maximally pool and average pool the features obtained from the convolutional layer to obtain a maximum pooling result and an average pooling result.
6. The method of claim 5, a width of a convolution kernel of the convolutional layer is consistent with a dimension of a type of the transactional data.
7. The method of any of claims 1 to 6, the transaction activity types including money laundering activities and non-money laundering activities;
the types of transaction data include at least: one of a number of funds in-flows, a number of funds out-flows, an amount of funds in-flows, and an amount of funds out-flows.
8. A transaction behavior type determination device, comprising:
the system comprises a first acquisition module, a second acquisition module and a processing module, wherein the first acquisition module is used for acquiring transaction data of at least one user in a first preset time period;
the first input module is used for inputting the transaction data into a hidden vector generation model to obtain a hidden vector related to the user transaction behavior, wherein the hidden vector generation model is obtained by training an automatic encoder based on a transaction data sample;
and the second input module is used for inputting the hidden vector into a transaction behavior type identification model and outputting the transaction behavior type of the at least one user in the first preset time period.
9. The apparatus of claim 8, the apparatus further comprising:
and the training module is used for sequentially inputting a plurality of groups of different transaction data samples into the automatic encoder and training the automatic encoder until the error between the output data and the input data of the automatic encoder is less than a preset value, so as to obtain the trained automatic encoder.
10. The apparatus of claim 9, the auto-encoder, comprising:
the system comprises a time sequence characteristic layer, an encoder, a first full connection layer, a decoder, a second full connection layer and an output layer;
the time sequence characteristic layer is used for converting first transaction data into user time sequence data according to the type of the first transaction data and the time sequence generated by the first transaction data;
the encoder is used for acquiring the characteristics of the user time sequence data under different time windows and pooling the characteristics to obtain a pooling result;
the first full-connection layer is used for connecting the pooling results to obtain the hidden vector;
the decoder is used for converting the implicit vector into second transaction data which is consistent with the first transaction data type;
the second full connection layer is used for converting the second transaction data into the size consistent with the first transaction data;
the output layer is used for outputting the converted second transaction data, wherein the first transaction data is input data of the automatic encoder, and the second transaction data is output data of the automatic encoder.
11. The apparatus of claim 10, the hidden vector generation model comprising:
the trained sequential feature layer, the encoder, and the first fully-connected layer in the auto-encoder.
12. The apparatus of claim 10, the encoder comprising a convolutional layer and a pooling layer;
the convolutional layer is used for acquiring the characteristics of each type of user time sequence data in a second preset time period according to the user time sequence data corresponding to the second time period, wherein the second preset time period is smaller than the first preset time period;
the pooling layer is configured to maximally pool and average pool the features obtained from the convolutional layer to obtain a maximum pooling result and an average pooling result.
13. The apparatus of claim 12, a width of a convolution kernel of the convolutional layer is consistent with a dimension of a type of the transaction data.
14. The apparatus of any of claims 8 to 13, the transaction activity types comprising money laundering activities and non-money laundering activities;
the types of transaction data include at least: one of a number of funds in-flows, a number of funds out-flows, an amount of funds in-flows, and an amount of funds out-flows.
15. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of determining a transaction behaviour type according to any one of claims 1 to 7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010167428.0A CN111401908A (en) | 2020-03-11 | 2020-03-11 | Transaction behavior type determination method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010167428.0A CN111401908A (en) | 2020-03-11 | 2020-03-11 | Transaction behavior type determination method, device and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111401908A true CN111401908A (en) | 2020-07-10 |
Family
ID=71432359
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010167428.0A Pending CN111401908A (en) | 2020-03-11 | 2020-03-11 | Transaction behavior type determination method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111401908A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116340632A (en) * | 2023-03-28 | 2023-06-27 | 北京有竹居网络技术有限公司 | Object recommendation method, device, medium and electronic equipment |
| CN120047244A (en) * | 2025-02-05 | 2025-05-27 | 同济大学 | Transaction behavior generation method based on model cascading technology |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108734338A (en) * | 2018-04-24 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Credit risk forecast method and device based on LSTM models |
| CN108921566A (en) * | 2018-05-03 | 2018-11-30 | 阿里巴巴集团控股有限公司 | A kind of wash sale recognition methods and device based on graph structure model |
| CN108920641A (en) * | 2018-07-02 | 2018-11-30 | 北京理工大学 | A kind of information fusion personalized recommendation method |
| CN110084609A (en) * | 2019-04-23 | 2019-08-02 | 东华大学 | A kind of transaction swindling behavior depth detection method based on representative learning |
| CN110276679A (en) * | 2019-05-23 | 2019-09-24 | 武汉大学 | A deep learning-oriented network personal credit fraud detection method |
-
2020
- 2020-03-11 CN CN202010167428.0A patent/CN111401908A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108734338A (en) * | 2018-04-24 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Credit risk forecast method and device based on LSTM models |
| CN108921566A (en) * | 2018-05-03 | 2018-11-30 | 阿里巴巴集团控股有限公司 | A kind of wash sale recognition methods and device based on graph structure model |
| CN108920641A (en) * | 2018-07-02 | 2018-11-30 | 北京理工大学 | A kind of information fusion personalized recommendation method |
| CN110084609A (en) * | 2019-04-23 | 2019-08-02 | 东华大学 | A kind of transaction swindling behavior depth detection method based on representative learning |
| CN110276679A (en) * | 2019-05-23 | 2019-09-24 | 武汉大学 | A deep learning-oriented network personal credit fraud detection method |
Non-Patent Citations (1)
| Title |
|---|
| R. SILIPO: "Fraud Detection Using a Neural Autoencoder", 《SMART DATA NEWS, ARTICLES, & EDUCATION》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116340632A (en) * | 2023-03-28 | 2023-06-27 | 北京有竹居网络技术有限公司 | Object recommendation method, device, medium and electronic equipment |
| CN120047244A (en) * | 2025-02-05 | 2025-05-27 | 同济大学 | Transaction behavior generation method based on model cascading technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | HPRN: Holistic prior-embedded relation network for spectral super-resolution | |
| CN114677556B (en) | Adversarial sample generation method and related equipment for neural network model | |
| CN112883227B (en) | Video abstract generation method and device based on multi-scale time sequence characteristics | |
| CN110532950B (en) | Video feature extraction method and micro-expression recognition method based on micro-expression video | |
| CN118799343B (en) | Plant segmentation method, device and equipment based on depth information | |
| CN113345463B (en) | Speech enhancement method, device, equipment and medium based on convolutional neural network | |
| CN113157987A (en) | Data preprocessing method for machine learning algorithm and related equipment | |
| CN114331901A (en) | Model training method and model training device | |
| CN112801960A (en) | Image processing method and device, storage medium and electronic equipment | |
| CN116542673B (en) | Fraud identification method and system applied to machine learning | |
| CN111401908A (en) | Transaction behavior type determination method, device and equipment | |
| CN111507461B (en) | Interpretability information determination method and device | |
| CN115760670B (en) | Unsupervised hyperspectral fusion method and device based on network implicit priori | |
| CN117009873A (en) | Payment risk identification model generation method, payment risk identification method and device | |
| CN112910890A (en) | Anonymous network flow fingerprint identification method and device based on time convolution network | |
| CN112232948A (en) | A kind of abnormal detection method and device of traffic data | |
| CN110401488A (en) | A demodulation method and device | |
| CN113792764A (en) | A sample expansion method, system, storage medium and electronic device | |
| CN119988860A (en) | Data center computing resources and refrigeration equipment energy consumption correlation analysis method and device | |
| CN118656695A (en) | Model training method, device, medium, electronic equipment and program product | |
| CN118968142A (en) | A method, device, terminal equipment and storage medium for extracting photovoltaic areas based on roofs | |
| CN118052582A (en) | Customer churn probability prediction method, device, computer equipment and storage medium | |
| CN114448661B (en) | Method for detecting slow denial of service attack and related equipment | |
| CN117351297A (en) | Noise interference image recognition method, device, processor and electronic equipment | |
| CN113419967B (en) | Automatic test method and device based on multi-user learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200710 |
|
| RJ01 | Rejection of invention patent application after publication |