Description:FORM 2

THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003

COMPLETE SPECIFICATION
(See Section 10 and Rule 13)

Title of invention:
REAL-TIME CAPSULE NETWORK BASED SYSTEM FOR FRAUD DETECTION

Applicant
Tata Consultancy Services Limited
A company Incorporated in India under the Companies Act, 1956
Having address:
Nirmal Building, 9th floor,
Nariman point, Mumbai 400021,
Maharashtra, India

Preamble to the description:

The following specification particularly describes the invention and the manner in which it is to be performed.
TECHNICAL FIELD
The disclosure herein generally relates to the field of real-time capsule network-based system for fraud detection, and more particularly, to a real-time capsule network-based Visual Question Answering (VQA) system for fraud detection.

BACKGROUND
Banking transactions can be vulnerable to fraudulent activities, which can cause financial harm to individuals and institutions. Traditional methods of fraud detection may not be able to keep up with the pace and complexity of modern banking transactions. The use of machine learning and artificial intelligence has shown promise in improving fraud detection in banking transactions. Capsule Networks are a new type of neural network architecture that have shown promise in improving the accuracy and efficiency of image analysis tasks. However, their potential for fraud detection in banking transactions has not been fully explored.

SUMMARY
Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, in one embodiment, a method for real-time capsule network-based Visual Question Answering (VQA) system for fraud detection is provided. The processor-implemented method includes receiving, via an input/output interface, a plurality of transactional data associated with a banking transaction, wherein the plurality of transactional data comprises both visual and non-visual data points associated with the banking transaction.
Further, the processor-implemented method comprises applying, via one or more hardware processors, a dynamic routing between capsules of a Capsule Network-based Visual Question Answering (VQA) module to learn a spatial relationship between the non-visual data points, and generating, via the one or more hardware processors, a hierarchical relationship between the visual and non-visual data points associated with the banking transaction in real-time, wherein the non-visual data points include a transaction amount, and a transaction history.
Furthermore, the processor-implemented method comprises analyzing, via one or more hardware processors, the hierarchical relationships between the visual and non-visual data points using a Capsule Network-based Visual Question Answering (VQA) module to identify one or more potential instances of fraud, wherein the hierarchical relationships identify complex patterns and anomalies that may indicate fraud. Finally, the processor-implemented method comprises determining, via the one or more hardware processors, a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the Capsule Network-based VQA module.
In another aspect, a real-time capsule network-based Visual Question Answering (VQA) system for fraud detection is provided. The system comprises a memory storing a plurality of instructions and one or more Input/Output (I/O) interfaces to receive a plurality of transactional data associated with a banking transaction, wherein the plurality of transactional data comprises both visual and non-visual data points associated with the banking transaction. Further, the system comprises one or more hardware processors coupled to the memory via the one or more I/O interfaces, wherein the one or more hardware processors are configured by the instructions to execute a plurality of modules of the system. A capsule network based Visual Question Answering (VQA) module of the system is configured to analyze the plurality of transactional data associated with the banking transaction in real-time to identify one or more potential instances of fraud and a fraud detection module of the system is configured to determine a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the capsule network based VQA model.
In yet another aspect, there are provided one or more non-transitory machine-readable information storage mediums comprising one or more instructions, which when executed by one or more hardware processors causes a method for real-time capsule network-based Visual Question Answering (VQA) system for fraud detection is provided. The processor-implemented method includes receiving, via an input/output interface, a plurality of transactional data associated with a banking transaction, wherein the plurality of transactional data comprises both visual and non-visual data points associated with the banking transaction.
Further, the processor-implemented method comprises applying, via one or more hardware processors, a dynamic routing between capsules of a Capsule Network-based Visual Question Answering (VQA) module to learn a spatial relationship between the non-visual data points, and generating, via the one or more hardware processors, a hierarchical relationship between the visual and non-visual data points associated with the banking transaction in real-time, wherein the non-visual data points include a transaction amount, and a transaction history.
Furthermore, the processor-implemented method comprises analyzing, via one or more hardware processors, the hierarchical relationships between the visual and non-visual data points using a Capsule Network-based Visual Question Answering (VQA) module to identify one or more potential instances of fraud, wherein the hierarchical relationships identify complex patterns and anomalies that may indicate fraud. Finally, the processor-implemented method comprises determining, via the one or more hardware processors, a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the Capsule Network-based VQA module.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS
The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles:
FIG. 1 illustrates a real-time capsule network based VQA system for fraud detection, according to some embodiments of the present disclosure.
FIG. 2 is a functional block diagram to illustrate a real-time capsule network based VQA system for fraud detection, according to some embodiments of the present disclosure.
FIG. 3 is an exemplary flow diagram illustrating a processor-implemented method for real-time capsule network-based VQA system for fraud detection, according to some embodiments of the present disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS
Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments.
Conventional fraud detection systems may rely on rule-based methods or traditional machine learning techniques, which may not be able to effectively detect complex instances of fraud in real-time. Traditional machine learning methods often struggle to identify new forms of fraud different from those encountered during training. This limitation can result in financial losses, breaches of trust, and increased security risks for businesses and consumers. The problem of fraud detection in banking transactions is generally faced by financial institutions, such as banks and other organizations that process financial transactions.
Embodiments herein provide a real-time capsule network based Visual Question Answering (VQA) system for fraud detection. The system is leveraging the power of the capsule networks to model the hierarchical relationships between different features of the transaction data, allowing for more accurate and efficient fraud detection in real-time. Further, the system is able to provide a comprehensive analysis of the transaction and improve the accuracy of its fraud detection capabilities, while also incorporating a real-time component for enhanced fraud prevention.
Potential applications of the invention include real-time fraud detection in banking and other financial institutions, as well as in other industries where transactional data is collected and analyzed. The system can also be adapted for use in other types of data analysis, such as identifying patterns in customer behavior or predicting future trends in sales or other business metrics.
One limitation of the current implementation is the reliance on non-visual data for real-time fraud detection. While this approach can be effective in identifying potential instances of fraud, there may be cases where visual data could provide additional insights and improve accuracy.
Referring now to the drawings, and more particularly to FIG. 1 through FIG. 3, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments and these embodiments are described in the context of the following exemplary system and/or method.
FIG. 1 illustrates a block diagram of a real-time capsule network based VQA system for fraud detection, in accordance with an example embodiment. Although the present disclosure is explained considering that the system 100 is implemented on a server, it may be understood that the system 100 may comprise one or more computing devices 102, such as a laptop computer, a desktop computer, a notebook, a workstation, a cloud-based computing environment and the like. It will be understood that the system 100 may be accessed through one or more input/output interfaces 104-1, 104-2... 104-N, collectively referred to as I/O interface 104. Examples of the I/O interface 104 may include, but are not limited to, a user interface, a portable computer, a personal digital assistant, a handheld device, a smartphone, a tablet computer, a workstation, and the like. The I/O interface 104 are communicatively coupled to the system 100 through a network 106.
In an embodiment, the network 106 may be a wireless or a wired network, or a combination thereof. In an example, the network 106 can be implemented as a computer network, as one of the different types of networks, such as virtual private network (VPN), intranet, local area network (LAN), wide area network (WAN), the internet, and such. The network 106 may either be a dedicated network or a shared network, which represents an association of the different types of networks that use a variety of protocols, for example, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol/Internet Protocol (TCP/IP), and Wireless Application Protocol (WAP), to communicate with each other. Further, the network 106 may include a variety of network devices, including routers, bridges, servers, computing devices, storage devices. The network devices within the network 106 may interact with the system 100 through communication links.
The system 100 supports various connectivity options such as BLUETOOTH®, USB, ZigBee, and other cellular services. The network environment enables connection of various components of the system 100 using any communication link including Internet, WAN, MAN, and so on. In an exemplary embodiment, the system 100 is implemented to operate as a stand-alone device. In another embodiment, the system 100 may be implemented to work as a loosely coupled device to a smart computing environment. Further, the system 100 comprises at least one memory 110 with a plurality of instructions, one or more databases 112, and one or more hardware processors 108 which are communicatively coupled with the at least one memory to execute a plurality of modules 114 therein. The plurality of modules 114, for example, includes a capsule network based Visual Question Answering (VQA) module 116, and a fraud detection module 118. The components and functionalities of the system 100 are described further in detail.
The input/output interface 104 of the system 100 is configured to receive transactional data associated with a banking transaction, which may include non-visual data such as transaction amount, transaction history, and text or numerical data. In order to analyze the non-visual data associated with the banking transaction in real-time, the capsule network based VQA module 116 is configured to identify patterns and anomalies in the data that may indicate potential instances of fraud, such as unusual transaction amounts or transactions from unknown locations as shown in FIG. 2.
Capsule Networks (CapsNets) are a type of neural network that model hierarchical relationships between simple and complex objects. In our system, the CapsNet-based VQA module leverages this property to analyze and understand transactional data, learning to identify potential instances of fraud based on patterns and anomalies detected in real-time. Further, the capsule network based VQA module 116 is also configured to model hierarchical relationships between different features of the data in a more structured way compared to traditional machine learning methods. This modeling approach allows the system 100 to better understand the relationships between different features of the data and improve its accuracy in identifying potential instances of fraud.
Furthermore, the capsule network based VQA module 116 is responsible for analyzing the non-visual data such as text and numerical data associated with banking transactions in real-time to identify potential instances of fraud. The capsule network based VQA module 116 uses a modified Capsule Network (CapsNet), a type of neural network architecture that models hierarchical relationships between features of the data, to identify complex patterns and anomalies that may indicate fraud.
The capsule network based VQA module 116 comprises several sub-modules, including a pre-processing module, a primary capsule layer, a routing-by-agreement module, and a classification module. The pre-processing module is responsible for preparing the data for analysis, including normalizing, and encoding the data. The primary capsule layer is composed of several capsules, which are groups of neurons that represent different features of the data. Each capsule calculates a probability distribution over its inputs and outputs a vector representing the presence of the corresponding feature in the input data.
The routing-by-agreement module uses a modified dynamic routing technique with a signature agreement penalty term to iteratively refine the primary capsule layer's output and generate a more accurate representation of the input data. This encourages the CapsNet to explore diverse fraud signatures, improving generalization capabilities and enabling the detection of now fraud patterns. The modified routing technique includes the introduction of a signature agreement penalty, which encourages the network to explore diverse and novel fraud signatures. This penalty promotes adaptability and the ability to learn new fraud patterns on the fly.
The routing-by-agreement module calculates the agreement between capsules in the primary layer and capsules in the higher-level layers and updates the routing weights accordingly. A signature agreement coefficient of the modified routing technique calculates the cosine similarity between two vectors, which indicates the degree to which they agree.
S_ij=(v_i.v_j)/?v_i ??v_j ? (1)
Further, a squashing function of the modified routing technique is used to squash output vector, limiting its magnitude while preserving its direction. It ensures the output vector’s magnitude lies between 0 and 1.
v_j=?s_j ?^2/(1+?s_j ?^2 ) s_j/?s_j ? (2)
Furthermore, a modified softmax function with signature agreement penalty function applies the softmax function while incorporating a penalty term based on the signature agreement. The penalty term encourages diversity in the output by penalizing similarity among the output vectors.
c_i=(exp?(b_i-?_j¦? s_ij))/(?_k¦?exp?(b_k-?_j¦? s_kj)?) (3)
Finally, the classification module uses the output of the routing-by-agreement module to determine whether the transaction is fraudulent based on a threshold value. If the transaction is determined to be fraudulent, the system can initiate appropriate actions such as flagging the transaction for review or rejecting the transaction outright.
Overall, the capsule network based VQA module 116 enabling the system 100 to identify potential instances of fraud more accurately and efficiently. In one example, consider a banking transaction involving a large transfer of funds from a small business account to an overseas account. The system 100 would analyze this data in real-time as it is being processed. The CapsNet-based VQA module 116 identifies potential instances of fraud by recognizing patterns and anomalies in the data, such as an unusually large transaction amount or an unexpected overseas transfer. If a potential fraud is detected, the transaction could be halted or flagged for further investigation.
The fraud detection module 118 of the system 100 is configured to receive the results of the capsule network based VQA module 116 and determines whether the banking transaction is fraudulent based on a threshold value. If the transaction is determined to be fraudulent, the system 100 may initiate appropriate actions such as flagging the transaction for review or rejecting the transaction outright.
The system 100 could be modified to incorporate a real-time fraud detection component. This would involve analyzing the transactional data as it is being processed, rather than waiting until the entire transaction is complete. This would allow the system 100 to detect and prevent fraudulent transactions in real-time, before they are completed. Additionally, the system 100 is modular and scalable, enabling it to be easily adapted to different banking systems and environments. It can be deployed on-premises or in the cloud and can be integrated with existing banking systems using standard APIs.
Therefore, the combination of real-time processing, the use of Capsule Networks for fraud detection, and the incorporation of visual and non-visual data analysis in a hierarchical and structured way differentiates the real-time capsule network based VQA system allows for more accurate and efficient fraud detection during banking transactions. The system is able to provide a comprehensive analysis of the transaction and improve the accuracy of its fraud detection capabilities, while also incorporating a real-time component for enhanced fraud prevention
FIG. 3 is a flow diagram illustrating a processor-implemented method 300 for a real-time capsule network based VQA system for fraud detection implemented by the system 100 of FIG. 1. Functions of the components of the system 100 are now explained through steps of flow diagram in FIG. 3, according to some embodiments of the present disclosure.
Initially, at step 302 of the processor-implemented method 300, the one or more hardware processors 108 are configured by the programmed instructions to receive, via an input/output interface, a plurality of transactional data associated with a banking transaction. The plurality of transactional data comprises both visual and non-visual data points associated with the banking transaction.
At the next step 304 of the processor-implemented method 300, the one or more hardware processors 108 are configured by the programmed instructions to applying, via one or more hardware processors, a dynamic routing between capsules of a capsule network based Visual Question Answering (VQA) module to learn a spatial relationship between the non-visual data points.
At the next step 306 of the processor-implemented method 300, the one or more hardware processors 108 are configured by the programmed instructions to generate a hierarchical relationship between the visual and non-visual data points associated with the banking transaction in real-time, wherein the non-visual data points include a transaction amount, and a transaction history.
At the next step 308 of the processor-implemented method 300, the one or more hardware processors 108 are configured by the programmed instructions to analyze the hierarchical relationships between the visual and non-visual data points using a capsule network based VQA module to identify one or more potential instances of fraud, wherein the hierarchical relationships identify complex patterns and anomalies that may indicate fraud.
Finally, at the last step 310 of the processor-implemented method 300, the one or more hardware processors 108 are configured by the programmed instructions to determine a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the Capsule Network-based VQA module.
The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims and may include other modifications that occur to those skilled in the art. Such other modifications are intended to be within the scope of the claims if they have similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insubstantial differences from the literal language of the claims.
The embodiments of present disclosure herein address unresolved problem of real-time fraud detection in banking transactions. Traditional fraud detection systems may only identify potential instances of fraud after the transaction has already been completed, leading to losses for the bank and potentially compromised customer information. The real-time capsule network based VQA system for fraud detection in banking addresses this problem by providing a more comprehensive and real-time analysis of the transaction data, allowing potential instances of fraud to be identified and prevented before the transaction is completed.
The problem of fraud detection in banking transactions is generally faced by financial institutions, such as banks and other organizations that process financial transactions. It should be solved to protect financial institutions and their customers from financial losses and reputational damage. Fraudulent activities can result in significant financial losses for banks and their customers, as well as erode customer trust and damage the reputation of the financial institution. By detecting and preventing fraudulent activities in real-time, banks can reduce their financial losses and maintain the trust of their customers. Additionally, the use of advanced fraud detection technologies can also help banks comply with regulatory requirements related to financial crime prevention.
The problem of fraud in banking can be considered solved when the real-time capsule network based VQA system for fraud detection is successfully implemented and deployed in banking systems, and it consistently detects and prevents instances of fraud in real-time. This can be evaluated through metrics such as the number of fraudulent transactions prevented or the reduction in financial losses due to fraud. Additionally, feedback from banking institutions and their customers can be collected to determine the effectiveness and usability of the system
It is to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g., any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g., hardware means like e.g., an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g., an ASIC and an FPGA, or at least one microprocessor and at least one memory with software processing components located therein. Thus, the means can include both hardware means, and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, e.g., using a plurality of CPUs.
The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various components described herein may be implemented in other components or combinations of other components. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope of the disclosed embodiments. Also, the words “comprising,” “having,” “containing,” and “including,” and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
It is intended that the disclosure and examples be considered as exemplary only, with a true scope of disclosed embodiments being indicated by the following claims.
, Claims:WE CLAIM:
1. A system (100) comprising:
an input/output interface (104) for receiving a plurality of transactional data associated with a banking transaction, wherein the plurality of transactional data comprises visual and non-visual data points associated with the banking transaction;
a memory (110) in communication with the one or more hardware processors (108), wherein the one or more hardware processors are configured to execute one or more programmed instructions stored in the memory;
a capsule network based Visual Question Answering (VQA) module (116) for analyzing the plurality of transactional data associated with the banking transaction in real-time to identify one or more potential instances of fraud, wherein the capsule network based VQA module is trained on a predefined dataset of prior fraud instances; and
a fraud detection module (118) for determining a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the capsule network based VQA module based on a predefined threshold value.

2. The system (100) as claimed in claim 1, further comprising a feedback module (120) for receiving one or more feedbacks on detected one or more instances of fraud to update the capsule network based VQA module (116) in real-time.

3. The system (100) as claimed in claim 1, wherein the capsule network-based VQA module (116) is configured to apply a dynamic routing between capsules to learn a spatial relationship between the non-visual data points.

4. The system (100) as claimed in claim 3, wherein the dynamic routing is used to refine output of primary capsule layer to generate accurate representation of the input data.

5. The system (100) as claimed in claim 1, wherein the capsule network-based VQA module (116) is configured to generate a hierarchical relationship between the visual and non-visual data points associated with the banking transaction in real-time to identify potential instances of fraud.

6. The system (100) as claimed in claim 5, wherein the generated hierarchical relationship between visual and non-visual data points enables to identify complex pattern and anomalies to indicate fraud.

7. The system (100) as claimed in claim 1, wherein the real-time analysis is performed on a piece-wise basis as the transactional data is received.

8. A processor-implemented method (300), comprising:
receiving, via an input/output interface, a plurality of transactional data associated with a banking transaction, wherein the plurality of transactional data comprises both visual and non-visual data points associated with the banking transaction (302);
applying, via one or more hardware processors, a dynamic routing between capsules of a Capsule Network-based Visual Question Answering (VQA) module to learn a spatial relationship between the non-visual data points (304);
generating, via the one or more hardware processors, a hierarchical relationship between the visual and non-visual data points associated with the banking transaction in real-time, wherein the non-visual data points include a transaction amount, and a transaction history (306);
analyzing, via one or more hardware processors, the hierarchical relationships between the visual and non-visual data points using a Capsule Network-based Visual Question Answering (VQA) module to identify one or more potential instances of fraud, wherein the hierarchical relationships identify complex patterns and anomalies that may indicate fraud (308); and
determining, via the one or more hardware processors, a fraudulent and a non-fraudulent banking transaction by applying a classification layer to the output of the Capsule Network-based VQA module (310).

9. The processor-implemented method (300) as claimed in claim 8, further comprising receiving one or more feedbacks on detected one or more instances of fraud to update the capsule network based VQA module in real-time.

10. The processor-implemented method (300) as claimed in claim 8, wherein the real-time analysis is performed on a piece-wise basis as the plurality of transactional data is received.

11. The processor-implemented method (300) as claimed in claim 8, wherein a dynamic routing is applied to capsules of the capsule network based VQA module to learn a spatial relationship between the non-visual data points.

12. The processor-implemented method (300) as claimed in claim 11, wherein the dynamic routing is used to refine output of primary capsule layer to generate an accurate representation of the input data.

Dated this 31st Day of August 2023
Tata Consultancy Services Limited
By their Agent & Attorney

(Adheesh Nargolkar)
of Khaitan & Co
Reg No IN-PA-1086