Abstract: Provided is a method for fraud prevention using deep learning and survival models. The method may include receiving, with at least one processor, transaction data associated with a plurality of transactions of at least one payment account. At least one attempted attack may be detected based on the transaction data. A fraud risk score for each subperiod of a plurality of subperiods in a time period following the at least one attempted attack may be generated based on the transaction data using a deep learning model and a survival model. The fraud risk score for each respective subperiod may be associated with a probability that a fraudulent transaction will not occur by the respective subperiod. A system and computer program product are also disclosed.

Abstract: A method for detecting fraudulent interactions may include receiving interaction data, including a first plurality of interactions with (first) fraud labels and a second plurality of interactions (without fraud labels). Second fraud label data for each of the second plurality of interactions may be generated with a first neural network (e.g., classifying whether each interaction is fraudulent or not). Generated interaction data and generated fraud label data may be generated with a second neural network. Discrimination data for each of the second plurality of interactions and generated interactions may be generated with a third neural network (e.g., classifying whether the respective interaction is real or not). Error data may be determined based on the discrimination data (e.g., whether the respective interaction is correctly classified). At least one of the neural networks may be trained based on the error data. A system and computer program product are also disclosed.

Abstract: Provided is a computer-implemented method for monitoring and improving data quality of transaction data that may include receiving transaction data associated with a plurality of payment transactions from an acquirer system. The transaction data may include a transaction record associated with each payment transaction of the plurality of payment transactions. Each transaction record may include a plurality of data fields. Each respective data field of the plurality of data fields may be categorized into a respective type of a plurality of types. A data quality score for each respective data field of the plurality of data fields may be determined based on the respective type of the respective data field. A system and computer program product are also provided.

Abstract: Provided is a method for fraud prevention using deep learning and survival models. The method may include receiving, with at least one processor, transaction data associated with a plurality of transactions of at least one payment account. At least one attempted attack may be detected based on the transaction data. A fraud risk score for each subperiod of a plurality of subperiods in a time period following the at least one attempted attack may be generated based on the transaction data using a deep learning model and a survival model. The fraud risk score for each respective subperiod may be associated with a probability that a fraudulent transaction will not occur by the respective subperiod. A system and computer program product are also disclosed.

Abstract: Provided are a method, system, and computer program product for generating synthetic data. The method includes generating a correlation graph of a plurality of data types based on a plurality of correlations. The method also includes generating a directed acyclic graph of the plurality of data types based on the correlation graph. The method further includes traversing the directed acyclic graph to produce a hierarchical graph of the plurality of data types, wherein the hierarchical graph includes a plurality of nodes, and wherein each node of the plurality of nodes is associated with a data type of the plurality of data types. The method further includes generating synthetic training data including a plurality of records of data by repeatedly traversing the hierarchical graph and based on a plurality of sets of values and a plurality of sets of interdependencies.

Abstract: Provided is a computer-implemented method for monitoring and improving data quality of transaction data that may include receiving transaction data associated with a plurality of payment transactions from an acquirer system. The transaction data may include a transaction record associated with each payment transaction of the plurality of payment transactions. Each transaction record may include a plurality of data fields. Each respective data field of the plurality of data fields may be categorized into a respective type of a plurality of types. A data quality score for each respective data field of the plurality of data fields may be determined based on the respective type of the respective data field. A system and computer program product are also provided.

Abstract: Provided are a system, method, and computer program product for generating synthetic data. The method includes receiving a plurality of data types associated with an environment to be evaluated and receiving a plurality of correlations of one data type to another data type. The method also includes generating a correlation graph of the plurality of data types based on the plurality of correlations and generating a directed acyclic graph of the plurality of data types based on the correlation graph. The method further includes generating a hierarchical graph of the plurality of data types by applying a path traversal technique to the directed acyclic graph and generating a synthetic dataset by repeatedly traversing the hierarchical graph to generate a plurality of records of data.

Abstract: A method for detecting fraudulent interactions may include receiving interaction data, including a first plurality of interactions with (first) fraud labels and a second plurality of interactions (without fraud labels). Second fraud label data for each of the second plurality of interactions may be generated with a first neural network (e.g., classifying whether each interaction is fraudulent or not). Generated interaction data and generated fraud label data may be generated with a second neural network. Discrimination data for each of the second plurality of interactions and generated interactions may be generated with a third neural network (e.g., classifying whether the respective interaction is real or not). Error data may be determined based on the discrimination data (e.g., whether the respective interaction is correctly classified). At least one of the neural networks may be trained based on the error data. A system and computer program product are also disclosed.

Abstract: A method for detecting fraudulent interactions may include receiving interaction data, including a first plurality of interactions with (first) fraud labels and a second plurality of interactions (without fraud labels). Second fraud label data for each of the second plurality of interactions may be generated with a first neural network (e.g., classifying whether each interaction is fraudulent or not). Generated interaction data and generated fraud label data may be generated with a second neural network. Discrimination data for each of the second plurality of interactions and generated interactions may be generated with a third neural network (e.g., classifying whether the respective interaction is real or not). Error data may be determined based on the discrimination data (e.g., whether the respective interaction is correctly classified). At least one of the neural networks may be trained based on the error data. A system and computer program product are also disclosed.

Abstract: A system, method, and product for determining a reason for a deep learning model output that obtain training data associated with training samples and first labels for the training samples; train a first model using the training samples and the first labels, training the first model generating predictions for the training samples; train a second model using the training samples and the predictions as second labels for the training samples; extract one or more weights of the trained second model; process, using the first model, input data including features associated with at least one sample, to generate output data, the output data including at least one prediction for the at least one sample; and apply the one or more extracted weights to the features to determine one or more contributions of one or more features of the features to the at least one prediction for the at least one sample.

Abstract: Provided are a system, method, and computer program product for generating synthetic data. The method includes receiving a plurality of data types associated with an environment to be evaluated and receiving a plurality of correlations of one data type to another data type. The method also includes generating a correlation graph of the plurality of data types based on the plurality of correlations and generating a directed acyclic graph of the plurality of data types based on the correlation graph. The method further includes generating a hierarchical graph of the plurality of data types by applying a path traversal technique to the directed acyclic graph and generating a synthetic dataset by repeatedly traversing the hierarchical graph to generate a plurality of records of data.

Abstract: Provided is a method for fraud prevention using deep learning and survival models. The method may include receiving, with at least one processor, transaction data associated with a plurality of transactions of at least one payment account. At least one attempted attack may be detected based on the transaction data. A fraud risk score for each subperiod of a plurality of subperiods in a time period following the at least one attempted attack may be generated based on the transaction data using a deep learning model and a survival model. The fraud risk score for each respective subperiod may be associated with a probability that a fraudulent transaction will not occur by the respective subperiod. A system and computer program product are also disclosed.

Abstract: Described are a system, method, and computer program product for breach detection using convolutional neural networks. The method includes receiving transaction data associated with a plurality of transactions completed in a first time period. The method also includes identifying a set of suspected fraudulent transactions of the plurality of transactions based on inputting at least one parameter of the transaction data into a fraud evaluation model. The method further includes generating an image comprising a field of points, wherein each point is associated with at least one transaction of the set of suspected fraudulent transactions, and wherein an x-axis position in the image of each point in the field of points is associated with a time subperiod of the first time period in which the at least one transaction occurred. The method further includes detecting a breach event by processing the image with a convolutional neural network (CNN) model.

Abstract: Described are a system, method, and computer program product for merchant breach detection using convolutional neural networks. The method includes receiving transaction data associated with a plurality of transactions by a plurality of payment devices in a first time period subsequent to the plurality of payment devices transacting with a merchant. The method also includes identifying, based on inputting at least one parameter of the transaction data into a fraud evaluation model, a set of suspected fraudulent transactions of the plurality of transactions. The method further includes generating an image comprising a field of points, wherein each point of the field of points is associated with at least one transaction. The method further includes detecting breach of the merchant by processing the image with a convolutional neural network (CNN) model.

Abstract: Described are a system, method, and computer program product for early detection of and response to a merchant data breach through machine-learning analysis. The method includes receiving transaction data associated with a plurality of transactions and receiving fraudulent transaction data representative of at least one previously identified data-breach incident. The method also includes generating a first model input dataset associated with the at least one merchant and a second model input dataset associated with the at least one previously identified data-breach incident. The method also includes training at least one machine-learning prediction model to associate merchants with a likelihood of data breach and determining at least one breached merchant of the at least one merchant. The method further includes generating a communication configured to cause at least one action to be taken in response to the determination of the at least one breached merchant.

Abstract: A method for detecting fraudulent interactions may include receiving interaction data, including a first plurality of interactions with (first) fraud labels and a second plurality of interactions (without fraud labels). Second fraud label data for each of the second plurality of interactions may be generated with a first neural network (e.g., classifying whether each interaction is fraudulent or not). Generated interaction data and generated fraud label data may be generated with a second neural network. Discrimination data for each of the second plurality of interactions and generated interactions may be generated with a third neural network (e.g., classifying whether the respective interaction is real or not). Error data may be determined based on the discrimination data (e.g., whether the respective interaction is correctly classified). At least one of the neural networks may be trained based on the error data. A system and computer program product are also disclosed.

Abstract: Provided is a system that includes at least one processor programmed or configured to: determine an average payment transaction vector based on a first payment transaction vector associated with a first payment transaction involving an account and a second payment transaction vector associated with a second payment transaction involving the account; determine an account embedding vector associated with the account based on the first payment transaction vector associated with the first payment transaction and the second payment transaction vector associated with the second payment transaction; determine a predicted transaction aggregate vector associated with the account based on the account embedding vector and a plurality of embedding payment transaction vectors associated with a plurality of payment transactions; and store the predicted transaction aggregate vector in a data structure based on an account identifier of the account. A computer-implemented method and computer program product are also provided.

Abstract: A system, method, and product for determining a reason for a deep learning model output that obtain training data associated with training samples and first labels for the training samples; train a first model using the training samples and the first labels, training the first model generating predictions for the training samples; train a second model using the training samples and the predictions as second labels for the training samples; extract one or more weights of the trained second model; process, using the first model, input data including features associated with at least one sample, to generate output data, the output data including at least one prediction for the at least one sample; and apply the one or more extracted weights to the features to determine one or more contributions of one or more features of the features to the at least one prediction for the at least one sample.

Abstract: Provided is a computer-implemented method for monitoring and improving data quality of transaction data that may include receiving transaction data associated with a plurality of payment transactions from an acquirer system. The transaction data may include a transaction record associated with each payment transaction of the plurality of payment transactions. Each transaction record may include a plurality of data fields. Each respective data field of the plurality of data fields may be categorized into a respective type of a plurality of types. A data quality score for each respective data field of the plurality of data fields may be determined based on the respective type of the respective data field. A system and computer program product are also provided.

Abstract: Provided is a computer-implemented method for monitoring and improving data quality of transaction data that may include conducting data pre-processing on transaction data associated with a plurality of payment transactions; determining feature values associated with a textual data field in each transaction record of a plurality of transaction records included in the transaction data associated with the plurality of payment transactions, wherein the feature values are used in a parsing layer of a natural language processing (NLP) model after conducting data pre-processing on the transaction data associated with the plurality of payment transactions; and determining whether the feature values associated with the textual data field satisfy one or more rules associated with the parsing layer of the NLP model.