Bank card anti-fraud method and device

Abstract

The present application provides a bank card anti-fraud method and device that can be used for bank card or credit card transactions in the financial field. The method includes: obtaining signature information for this transaction, the signature information including a first static signature handwriting and a first dynamic writing trajectory; performing a similarity comparison between the first static signature handwriting and a stored second static signature handwriting to obtain a first comparison result; performing a similarity comparison between the first dynamic writing trajectory and a second dynamic writing trajectory to obtain a second comparison result, the second dynamic writing trajectory being generated by a dynamic writing trajectory generation model based on the second static signature handwriting; determining transaction result information for this transaction based on the first comparison result and the second comparison result, the transaction result information indicating whether this transaction is a normal transaction, and returning the transaction result information. In this way, dynamic writing trajectories can be compared when a customer conducts an offline bank card transaction, thereby achieving the function of real-time anti-fraud monitoring for offline transactions.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a bank card anti-fraud method and device. Background Art

[0002] Bank cards are a financial service tool spawned in the internet age. They represent both advancements in banking services and a representative achievement of the development of the market economy. Using bank cards not only enables cashless payments but also reduces the circulation of checks, truly revolutionizing traditional banking services. While the emergence of bank cards has provided a new direction for the development of the financial industry, the risks they present should not be underestimated. Financial market research reveals a variety of bank card fraud methods.

[0003] Current bank card fraud methods are diverse, covert, and widespread. Therefore, the development of bank card services not only faces the risks of traditional financial industry development but also requires proper mitigation of fraud risks.

[0004] In recent years, with the rise of mobile Internet technology, customers have been provided with convenient financial services such as mobile payment and online wealth management. However, financial institutions are also faced with a series of fraud risks such as false channel traffic, false customer fission, and false credit risks, which have increased the anti-fraud costs of financial institutions.

[0005] External bank card fraud risk refers to the risk that criminals steal cardholder account funds through fraudulent tactics such as counterfeiting bank cards, stealing card information, and impersonating others, causing financial losses to issuing banks. Fraud types primarily include counterfeit card fraud, stolen card fraud, false applications, and account theft. In addition to causing financial losses for banks, bank card fraud can also lead to significant complaints and lawsuits, posing systemic reputational risks. Therefore, it is attracting increasing attention from issuing banks. Summary of the Invention

[0006] The inventors discovered that current anti-fraud measures primarily focus on pre-fraud measures, such as preventing counterfeit bank cards, theft of card information, and standardizing third-party payment management; during online fraud, such as improving the stability and security of payment systems and leveraging big data technology for real-time anti-fraud monitoring; and post-fraud measures, such as contacting the issuing bank immediately to report the loss and freezing the account. However, there is a crucial offline fraud scenario for which there are no effective monitoring and prevention methods, such as offline credit card fraud.

[0007] In addition, when a customer opens a bank card, the bank only collects the customer's static signature handwriting. Therefore, when performing handwriting similarity comparison, only the similarity comparison of the static signature handwriting is compared, and the similarity comparison of the customer's dynamic writing trajectory cannot be performed.

[0008] In order to solve at least one of the above problems, an embodiment of the present application provides a bank card anti-fraud method, which obtains the signature handwriting and writing trajectory information of the cardholder when making offline purchases, and uses a time-series generative adversarial network to generate a dynamic writing trajectory. Thus, by comparing the generated dynamic writing trajectory with the writing trajectory of the cardholder when making offline purchases, offline credit card fraud monitoring can be performed to prevent the occurrence of offline credit card fraud in real time.

[0009] According to a first aspect of an embodiment of the present application, a bank card anti-fraud method is provided, the method comprising:

[0010] Acquire signature information of this transaction, wherein the signature information includes a first static signature handwriting and a first dynamic writing track;

[0011] Comparing the first static signature handwriting with the stored second static signature handwriting for similarity to obtain a first comparison result;

[0012] Comparing the first dynamic writing trajectory with the second dynamic writing trajectory for similarity to obtain a second comparison result, wherein the second dynamic writing trajectory is generated by a dynamic writing trajectory generation model based on the second static signature handwriting;

[0013] Transaction result information of the current transaction is determined according to the first comparison result and the second comparison result, the transaction result information indicating whether the current transaction is a normal transaction, and the transaction result information is returned.

[0014] According to the second aspect of the embodiment of the present application, when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is greater than or equal to a first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is greater than or equal to a second threshold, it is determined that this transaction is a normal transaction; when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is less than the first threshold, or the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is less than the second threshold, it is determined that this transaction is an abnormal transaction.

[0015] The first threshold and the second threshold are predetermined values.

[0016] According to the third aspect of the embodiment of the present application, when it is determined that the transaction is a normal transaction, information indicating that the transaction is continued is returned; when it is determined that the transaction is an abnormal transaction, the transaction is closed and information indicating that the transaction is stopped is returned.

[0017] According to the fourth aspect of the embodiment of the present application, when it is determined that this transaction is a normal transaction, the first dynamic writing trajectory and the first static signature handwriting are stored, the first static signature handwriting and the first dynamic writing trajectory are used as training samples, and the dynamic writing trajectory generation model is updated.

[0018] According to a fifth aspect of the embodiment of the present application, when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is greater than or equal to the first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is less than the second threshold, determining whether a preset maximum number of comparisons has been reached;

[0019] When it is determined that the preset maximum value has not been reached, reacquiring the first dynamic writing trajectory and re-comparing it with the second dynamic writing trajectory;

[0020] When it is determined that the preset maximum value has been reached, the transaction is determined to be an abnormal transaction.

[0021] According to a sixth aspect of the embodiment of the present application, when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is less than the first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is greater than or equal to the second threshold, it is determined whether a preset maximum number of comparisons has been reached;

[0022] When it is determined that the preset maximum number of comparisons has not been reached, reacquiring the first static handwriting and re-comparing it with the second static handwriting;

[0023] When it is determined that the preset maximum number of comparisons has been reached, the transaction is determined to be an abnormal transaction.

[0024] According to a seventh aspect of an embodiment of the present application, the dynamic writing trajectory generation model is implemented by a time-series-based generative adversarial network, wherein the time-series-based generative adversarial network includes a generator and a discriminator, and training the dynamic writing trajectory generation model includes:

[0025] training the generator to generate data similar to actual data;

[0026] training the discriminator to discriminate data generated by the generator;

[0027] The training goal is for the generator to generate data that the discriminator cannot distinguish from actual data.

[0028] According to an eighth aspect of the embodiment of the present application, the generator and the discriminator are respectively composed of a deep long short-term memory network (LSTM) and a fully connected layer;

[0029] The input of each LSTM in the generator is a random vector and the output of the previous LSTM, and the output of each LSTM in the generator serves as the input of the fully connected layer in the generator;

[0030] The input of each LSTM in the discriminator is the output of the next and previous LSTMs, and the output of each LSTM in the discriminator serves as the input of the fully connected layer in the discriminator.

[0031] According to a ninth aspect of an embodiment of the present application, a bank card anti-fraud device is provided, the device comprising:

[0032] An acquiring unit, configured to acquire signature information of the current transaction, wherein the signature information includes a first static signature handwriting and a first dynamic writing track;

[0033] a comparing unit, configured to compare the first static signature handwriting with a stored second static signature handwriting for similarity to obtain a first comparison result, and to compare the first dynamic writing trajectory with a second dynamic writing trajectory for similarity to obtain a second comparison result, wherein the second dynamic writing trajectory is generated by a dynamic writing trajectory generation model based on the second static signature handwriting;

[0034] A determining unit determines transaction result information of the current transaction based on the first comparison result and the second comparison result, wherein the transaction result information indicates whether the current transaction is a normal transaction, and returns the transaction result information.

[0035] One of the beneficial effects of the embodiments of the present application is:

[0036] By training a dynamic handwriting trajectory generation model based on a time-series-based generative adversarial network, we can automatically generate dynamic handwriting trajectories based on a small number of static customer signatures. This allows for comparison of dynamic handwriting trajectories during offline bank card transactions. Compared to previous methods of anti-fraud before and after the transaction, as well as real-time online transactions, this system enables real-time anti-fraud monitoring for offline transactions, significantly reducing the risk of bank card fraud. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0038] Figure 1 This is a schematic diagram of the bank card anti-fraud method according to an embodiment of the present application.

[0039] Figure 2 This is another schematic diagram of the bank card anti-fraud method according to an embodiment of the present application.

[0040] Figure 3 This is another schematic diagram of the bank card anti-fraud method according to an embodiment of the present application.

[0041] Figure 4 2 is a schematic diagram of a time-series-based generative adversarial network used in an embodiment of the present application.

[0042] Figure 5 A schematic diagram of a bank card anti-fraud device according to an embodiment of the present application.

[0043] Figure 6 This is a schematic diagram of the processing process of the bank card anti-fraud device based on the time-series generative adversarial network in an embodiment of the present application. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0045] It should be noted that the bank card anti-fraud method and device disclosed in this application can be used in the field of financial technology, and can also be used in any field other than the field of financial technology. This application does not limit the application field of the bank card anti-fraud method and device.

[0046] In addition, in the embodiments of this application, the acquisition, storage, use, and processing of data are in compliance with the relevant provisions of national laws and regulations. In addition, the user information involved in the embodiments of this application is obtained through legal and compliant channels, and the acquisition, storage, use, and processing of user information are authorized and agreed by the customer.

[0047] Example 1

[0048] Figure 1Schematic diagram of the bank card anti-fraud method according to an embodiment of the present application. Figure 1 As shown, the bank card anti-fraud method provided by the embodiment of the present invention includes:

[0049] 101: Obtain signature information of this transaction, where the signature information includes a first static signature handwriting and a first dynamic writing track;

[0050] 102: Compare the first static signature handwriting with the stored second static signature handwriting for similarity to obtain a first comparison result;

[0051] 103: Comparing the first dynamic writing trajectory with the second dynamic writing trajectory for similarity to obtain a second comparison result, wherein the second dynamic writing trajectory is generated by the dynamic writing trajectory generation model based on the second static signature handwriting;

[0052] 104: Determine transaction result information of this transaction based on the first comparison result and the second comparison result. The transaction result information indicates whether this transaction is a normal transaction, and return the transaction result information.

[0053] In the above embodiment, the first static signature and the first dynamic writing trace are obtained from the signature information written in real time by the cardholder during the current transaction. This application does not limit the specific acquisition method. In addition, the second static signature is the customer's static signature pre-stored in the bank's signature information database, for example, the handwriting entered by the cardholder when the card was activated.

[0054] In the above embodiment, the dynamic writing trajectory generation model is trained using pre-stored static signature handwriting and its corresponding dynamic writing trajectory to obtain a dynamic writing trajectory generation model. The pre-stored static signature handwriting and its corresponding dynamic writing trajectory can be the signature information entered by the customer when the card is activated. That is, the signature information pre-entered by the customer when the card is activated is used as a training data sample to obtain the dynamic writing trajectory generation model. Thus, the dynamic trajectory generation model can be obtained using only a small amount of data samples. This dynamic trajectory generation model can then be used to generate a corresponding dynamic writing trajectory (second dynamic writing trajectory) based on the static signature trajectory (second static signature trajectory) in the signature information entered by the user when the card is activated, and then compared with the dynamic writing trajectory (first dynamic writing trajectory) in the signature information entered by the user during the transaction.

[0055] Through the above method, when the cardholder conducts an offline transaction at the POS machine, the signature information of the current transaction can be obtained. Not only the static signature handwriting is compared, but also the dynamic writing trajectory in the obtained signature information is compared with the dynamic writing trajectory generated by the dynamic writing trajectory generation model. Based on the two comparison results, it is determined whether the transaction is a normal transaction, which improves the accuracy of offline transaction security information verification and the security of offline transactions. In addition, offline transactions can be detected in real time to achieve real-time anti-fraud monitoring of offline transactions.

[0056] Figure 2 is another schematic diagram of the bank card anti-fraud method according to an embodiment of the present application. Figure 2 As shown, the bank card anti-fraud method of the embodiment of the present application may include a training process and a usage process.

[0057] During the training process, when activating a card, the customer enters a static signature handwriting (201), and the static signature handwriting is stored in the customer's static handwriting information database (202); in addition, a dynamic writing trajectory generation model is used to generate a corresponding dynamic writing trajectory (200) based on the static signature handwriting; wherein the dynamic writing trajectory generation model is obtained by training a small amount of dynamic signature handwriting (203), that is, the dynamic signature handwriting can be used as a training data sample to train the dynamic writing trajectory generation model.

[0058] During use, the static signature handwriting of the trajectory information entered by the user in this transaction is compared with the static signature handwriting recorded when the card is activated (204). At the same time, the dynamic writing trajectory in the trajectory information entered by the user in this transaction is compared with the dynamic writing trajectory generated according to the dynamic writing trajectory generation model (205). Based on the results of the two comparisons, it is monitored whether this transaction is a normal transaction or an abnormal transaction (206).

[0059] In some embodiments, when both the first comparison result and the second comparison result obtained in step 102 and step 103, respectively, show a similarity greater than or equal to a predetermined threshold, the transaction is determined to be a normal transaction; otherwise, the transaction is determined to be an abnormal transaction. Specifically, when the first comparison result shows a similarity between the first static signature handwriting and the second static signature handwriting greater than or equal to a first threshold, and the second comparison result shows a similarity between the first dynamic writing trajectory and the second dynamic writing trajectory greater than or equal to a second threshold, the transaction is determined to be a normal transaction; when the first comparison result shows a similarity between the first static signature handwriting and the second static signature handwriting less than the first threshold, or when the second comparison result shows a similarity between the first dynamic writing trajectory and the second dynamic writing trajectory less than the second threshold, the transaction is determined to be an abnormal transaction. That is, if either comparison result shows a similarity less than a predetermined threshold, the transaction is considered an abnormal transaction; only if both comparison results show a similarity greater than or equal to the predetermined threshold is the transaction considered a normal transaction.

[0060] In the above embodiment, the first and second comparison results are obtained using a similarity scoring method. By presetting first and second thresholds for evaluating similarity, when the scores are lower than the corresponding threshold scores, the results are determined to be dissimilar, and when the scores are higher than the corresponding threshold scores, the results are determined to be similar. However, the present application is not limited thereto. For example, the above comparison results can also be obtained using similarity calculation methods such as cosine similarity and Euclidean distance, and this is not limited in the present application.

[0061] Through the above method, security verification can be performed not only by comparing static signatures, but also by comparing the similarity of dynamic signature trajectories, thereby improving the accuracy of signature information verification and the security of offline transactions.

[0062] In some embodiments, when it is determined that the transaction is a normal transaction, information indicating that the transaction will continue is returned, and thus the transaction will proceed normally; on the other hand, when it is determined that the transaction is an abnormal transaction, that is, the transaction may be a fraudulent act such as card theft, the transaction will be closed and information indicating that the transaction has been stopped is returned. At this time, the staff can conduct telephone confirmation and other processing operations with the cardholder.

[0063] In the above embodiment, when the transaction is determined to be normal, the first dynamic writing trajectory and the first static signature handwriting may also be stored. For example, the first static signature handwriting and the first dynamic writing trajectory may be used as training samples to update the dynamic writing trajectory generation model. This can increase the amount of sample data for model training and improve the accuracy of the dynamic writing trajectory generation model.

[0064] In the above embodiment, when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is greater than or equal to the first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is less than the second threshold, it can also be determined whether a preset maximum number of comparisons has been reached;

[0065] When it is determined that the preset maximum number of comparisons has not been reached, the first dynamic writing trajectory is reacquired and re-compared with the second dynamic writing trajectory;

[0066] When it is determined that the maximum number of comparisons set in advance has been reached, the transaction is determined to be an abnormal transaction, and subsequent processing under the abnormal transaction situation is performed, including but not limited to closing the transaction and returning a message that the transaction has been stopped.

[0067] In the above embodiment, when the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is less than the first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is greater than or equal to the second threshold, it can also be determined whether a preset maximum number of comparisons has been reached;

[0068] When it is determined that the preset maximum number of comparisons has not been reached, reacquiring the first static handwriting and re-comparing it with the second static handwriting;

[0069] When it is determined that the maximum number of comparisons set in advance has been reached, the transaction is determined to be an abnormal transaction, and subsequent processing under the abnormal transaction situation is performed, including but not limited to closing the transaction and returning a message that the transaction has been stopped.

[0070] In the above embodiment, there is no limit on the maximum number of comparisons, which can be 3, 4, 5, etc. This can avoid misjudgment of the result due to typos when the cardholder signs, thereby improving the accuracy of offline transactions.

[0071] Figure 3 This is another schematic diagram of the bank card anti-fraud method of the embodiment of the present application. Figure 3 As shown, the method includes:

[0072] 301: Obtain the static signature information and dynamic writing track of this transaction;

[0073] 302: The dynamic writing trajectory generation model generates a dynamic writing trajectory according to the static signature handwriting;

[0074] 303: Compare the static signature information obtained in this transaction with the pre-stored static signature information to obtain a first comparison result; compare the static signature information obtained in this transaction with the pre-stored static signature information to obtain a second comparison result;

[0075] 304: Determine whether the first comparison result is greater than a first threshold and the second comparison result is greater than a second threshold. If so, execute step 305; otherwise, execute step 306.

[0076] 305: This transaction is normal and the message indicating that the transaction is continuing is returned.

[0077] 306: This transaction is an abnormal transaction;

[0078] 307: Determine whether the maximum number of comparisons has been reached. If so, proceed to step 308; otherwise, return to step 301 to reacquire the signature information for this transaction.

[0079] 308: Close this transaction and return the information that the transaction is stopped.

[0080] It is worth noting that the above Figure 1 、 2 3 are only schematic illustrations of the embodiments of the present application, but the present application is not limited thereto. For example, the execution order of the various operations can be appropriately adjusted, and other operations can be added or some operations can be reduced. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the above appended examples. Figure 1 、 2 and 3 records.

[0081] Through the above method, when the cardholder conducts an offline transaction at the POS machine, the signature information of the current transaction can be obtained. Not only the static signature handwriting is compared, but also the dynamic writing trajectory in the obtained signature information is compared with the dynamic writing trajectory generated by the dynamic writing trajectory generation model. The two comparison results are used to determine whether the current transaction is a normal transaction. Moreover, the current transaction is determined to be a normal transaction only when the comparison results of the static signature handwriting and the dynamic writing trajectory are both greater than the similarity threshold. This can avoid misjudgment of the result caused by the cardholder's typo when signing, further improve the accuracy of offline transaction security information verification, improve the security of offline transactions, and detect offline transactions in real time to achieve real-time anti-fraud monitoring of offline transactions.

[0082] In some embodiments, the dynamic writing trajectory generation model is implemented by a time-series-based generative adversarial network. However, the present application is not limited thereto, and the dynamic writing trajectory generation model can also be implemented by other neural networks.

[0083] Generative adversarial networks (GANs) are a class of neural network architectures designed to generate realistic data. The approach involves training two neural models, a generator (G) and a discriminator (D), with conflicting objectives, forcing each to improve. The generator attempts to generate realistic-looking samples, while the discriminator attempts to distinguish between generated samples and real data.

[0084] Figure 4 : is a schematic diagram of a time-series-based generative adversarial network used in the embodiment of the present application. The time-series-based generative adversarial network can be a recursive neural network. Figure 4 As shown, the time-series-based generative adversarial network includes a generator (G) 401 and a discriminator (D) 402. In this embodiment of the present application, the step of training the dynamic writing trajectory generation model may include: training the generator 401 to generate data similar to actual data; and training the discriminator 402 to identify the data generated by the generator 401. The training goal is to enable the generator 401 to generate data that the discriminator 402 cannot identify from the actual data.

[0085] In some embodiments, the generator 401 and the discriminator 402 are composed of a deep long short-term memory network (LSTM) and a fully connected layer, respectively; Figure 4 As shown, the input of each LSTM in the generator 401 is a random vector and the output of the previous LSTM, and the output of each LSTM serves as the input of the fully connected layer in the generator 401; the input of each LSTM in the discriminator 402 is the output of the next and previous LSTMs, and the output of each LSTM serves as the input of the fully connected layer in the discriminator 402.

[0086] The bank card anti-fraud method provided in this application uses a dynamic writing trajectory generation model derived from training a time-series-based generative adversarial network. This model can automatically generate dynamic writing trajectories based on a small number of static customer signatures, allowing for comparison of these trajectories during offline bank card transactions. Compared to previous methods of anti-fraud before and after the transaction, as well as real-time online transactions, this method enables real-time anti-fraud monitoring for offline transactions, significantly reducing the risk of bank card fraud.

[0087] Example 2

[0088] The embodiment of the present application also provides a bank card anti-fraud device, which corresponds to the bank card anti-fraud method in Example 1. Therefore, the implementation of the device can refer to the implementation of the bank card anti-fraud method in Example 1, and the repeated parts will not be repeated.

[0089] Figure 5 A schematic diagram of the bank card anti-fraud device according to an embodiment of the present application. Figure 5As shown, the bank card anti-fraud device 600 of the embodiment of the present application includes: an acquisition unit 601, a comparison unit 602, and a determination unit 603.

[0090] An acquisition unit 601 acquires signature information of this transaction, the signature information including a first static signature handwriting and a first dynamic writing track;

[0091] Comparison unit 602 compares the first static signature handwriting with the stored second static signature handwriting for similarity to obtain a first comparison result, and compares the first dynamic writing trajectory with the second dynamic writing trajectory generated based on the second static signature handwriting using the dynamic writing trajectory generation model for similarity to obtain a second comparison result;

[0092] The determining unit 603 determines the transaction result information of the current transaction according to the first comparison result and the second comparison result, wherein the transaction result information indicates whether the current transaction is a normal transaction, and returns the transaction result information.

[0093] Figure 6 FIG. 6 is a schematic diagram of the processing process of the bank card anti-fraud device 600 based on the temporal generation adversarial network according to an embodiment of the present application. Figure 6 As shown, the bank card anti-fraud device 600 first obtains signature information (including static signature information and dynamic writing trajectory) through the acquisition unit 601, and then performs static signature handwriting similarity comparison and dynamic writing trajectory similarity comparison through the comparison unit 602, and monitors whether the current transaction is a fraud such as stolen card based on the comparison result through the determination unit 603; then, subsequent processing is performed through the processing unit 605. For example, when it is determined that the current transaction is a fraud, the transaction is closed, and the staff or the system can confirm with the cardholder by phone; if it is determined that the current transaction is not a fraud, the subsequent processing is continued.

[0094] The bank card anti-fraud device based on a time-series generative adversarial network involved in the embodiment of the present application can monitor offline credit card fraud by acquiring the signature handwriting and writing trajectory information of the cardholder when making purchases at an offline POS machine, and prevent the occurrence of offline credit card fraud through an innovative bank card anti-fraud module based on a time-series generative adversarial network. In addition, the dynamic writing trajectory generation model in the device can automatically generate dynamic writing trajectories based on the static signature handwriting of existing customers, so as to compare the dynamic writing trajectories when the customer conducts offline bank card transactions. Compared with the previous pre- and post-process and online transaction real-time anti-fraud, this device realizes the role of real-time anti-fraud monitoring for offline transactions.

[0095] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0096] The user information in the embodiments of this application is obtained through legal and compliant channels, and the acquisition, storage, use, and processing of the user information are authorized and agreed upon by the customer.

[0097] Although this application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on routine or non-creative work. The order of steps listed in the embodiments is only one way of executing the steps among many, and does not represent the only execution order. When an actual device or client product executes the method steps shown in the embodiments or the figures, the steps may be executed sequentially or in parallel (for example, in a parallel processor or multi-threaded processing environment).

[0098] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, devices (systems), or computer program products. Therefore, the embodiments of this specification may take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware. Furthermore, this application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0099] The present application is described with reference to the flowcharts and / or block diagrams of the methods, apparatus (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0100] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0102] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between the various embodiments may be referred to in conjunction with each other. Each embodiment focuses on the differences from the other embodiments. In particular, the system embodiments, since they are generally similar to the method embodiments, are described more simply. For relevant details, refer to the description of the method embodiments. In this document, relational terms such as first and second, etc., are used solely to distinguish one entity or operation from another and do not necessarily require or imply any actual relationship or order between these entities or operations. Furthermore, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. Terms such as "upper" and "lower" indicate positions or locations based on those shown in the accompanying drawings and are intended solely to facilitate the description of this application and simplify the description. They are not intended to indicate or imply that the device or element referred to must have, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on this application. Unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments in this application can be combined with each other. This application is not limited to any single aspect, nor to any single embodiment, nor to any combination and / or permutation of these aspects and / or embodiments. Moreover, each aspect and / or embodiment of the present application can be used alone or in combination with one or more other aspects and / or embodiments thereof.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and they should all be included in the scope of the claims and description of the present application.

Claims

1. A bank card anti-fraud method, characterized in that: The method comprises: Acquire signature information of this transaction, wherein the signature information includes a first static signature handwriting and a first dynamic writing track; Comparing the first static signature handwriting with the stored second static signature handwriting for similarity to obtain a first comparison result; Comparing the first dynamic writing trajectory with the second dynamic writing trajectory for similarity to obtain a second comparison result, wherein the second dynamic writing trajectory is generated by a dynamic writing trajectory generation model based on the second static signature handwriting; determining transaction result information of the current transaction based on the first comparison result and the second comparison result, the transaction result information indicating whether the current transaction is a normal transaction, and returning the transaction result information; The dynamic writing trajectory generation model is implemented by a time-series-based generative adversarial network. The time-series-based generative adversarial network includes a generator and a discriminator, and training the dynamic writing trajectory generation model includes: training the generator to generate data similar to actual data; training the discriminator to discriminate data generated by the generator; The training goal is for the generator to generate data that the discriminator cannot distinguish from actual data.

2. The method according to claim 1, characterized in that When the first comparison result shows that the similarity between the first static signature handwriting and the second static signature handwriting is greater than or equal to a first threshold, and the second comparison result shows that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is greater than or equal to a second threshold, determining that the transaction is a normal transaction; When the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is less than the first threshold, or when the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is less than the second threshold, it is determined that the transaction is an abnormal transaction. The first threshold and the second threshold are predetermined values.

3. The method according to claim 1, characterized in that When the transaction is determined to be normal, the message of continuing the transaction is returned; When it is determined that this transaction is an abnormal transaction, the transaction is closed and the information that the transaction is stopped is returned.

4. The method according to claim 3, characterized in that When it is determined that this transaction is a normal transaction, the first dynamic writing trajectory and the first static signature handwriting are stored, the first static signature handwriting and the first dynamic writing trajectory are used as training samples, and the dynamic writing trajectory generation model is updated.

5. The method according to claim 3, characterized in that When the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is greater than or equal to a first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is less than a second threshold, determining whether a preset maximum number of comparisons has been reached; When it is determined that the preset maximum value has not been reached, reacquiring the first dynamic writing trajectory and re-comparing it with the second dynamic writing trajectory; When it is determined that the preset maximum value has been reached, the transaction is determined to be an abnormal transaction.

6. The method according to claim 3, characterized in that When the first comparison result is that the similarity between the first static signature handwriting and the second static signature handwriting is less than a first threshold, and the second comparison result is that the similarity between the first dynamic writing trajectory and the second dynamic writing trajectory is greater than or equal to a second threshold, determining whether a preset maximum number of comparisons has been reached; When it is determined that the preset maximum value has not been reached, reacquiring the first static signature handwriting and re-comparing it with the second static signature handwriting; When it is determined that the preset maximum value has been reached, the transaction is determined to be an abnormal transaction.

7. The method according to claim 4, characterized in that The generator and the discriminator are respectively composed of a deep long short-term memory network LSTM and a fully connected layer; The input of each LSTM in the generator is a random vector and the output of the previous LSTM, and the output of each LSTM in the generator serves as the input of the fully connected layer in the generator; The input of each LSTM in the discriminator is the output of the next and previous LSTMs, and the output of each LSTM in the discriminator serves as the input of the fully connected layer in the discriminator.

8. A bank card anti-fraud device, characterized in that: The device comprises: An acquiring unit, configured to acquire signature information of the current transaction, wherein the signature information includes a first static signature handwriting and a first dynamic writing track; a comparing unit, configured to compare the first static signature handwriting with a stored second static signature handwriting for similarity to obtain a first comparison result, and to compare the first dynamic writing trajectory with a second dynamic writing trajectory for similarity to obtain a second comparison result, wherein the second dynamic writing trajectory is generated by a dynamic writing trajectory generation model based on the second static signature handwriting; a determining unit, which determines transaction result information of the current transaction based on the first comparison result and the second comparison result, wherein the transaction result information indicates whether the current transaction is a normal transaction, and returns the transaction result information; The device further comprises: The dynamic writing trajectory generation model is implemented by a time-series-based generative adversarial network. The time-series-based generative adversarial network includes a generator and a discriminator, and training the dynamic writing trajectory generation model includes: training the generator to generate data similar to actual data; training the discriminator to discriminate data generated by the generator; The training goal is for the generator to generate data that the discriminator cannot distinguish from actual data.

Images