Abstract: Techniques are disclosed relating to a graphical user interface (GUI) for editing classification rules. The graphical user interface may include a display of a modified threshold value for a classification rule, where values on one side (e.g. above) the threshold value indicate a first type of classification, and values on the other side (e.g. below) indicate a second type of classification. Machine learning techniques may be used to suggest a modified threshold value to the user, who may accept the modified value, or may provide, via the GUI, their own modified value, which can be different from a suggested value. Graphical indications of accuracy for the classification rule may be displayed.

Abstract: Systems, methods, and computer program products for identifying a fraudulent device. A device analytics engine receives device data from a computing device, the device data including parameters associated with the computing device. The device analytics engine selects a set of rules in a plurality of rules that indicate at least one parameter in the plurality of parameters in the device data for determining a device identifier. The set of rules are evaluated in order until the device identifier is determined from the at least one parameter indicated in the set of rules, the device data, and previously stored data from multiple computing devices. A score is generated for the computing device using one or more of the device identifier, device data, a set of rules, and previously receive device data that corresponds to the device identifier. A computing device is identified as a fraudulent computing device based on the score.

Abstract: Systems, methods, and computer program products for identifying a fraudulent device. A device analytics engine receives device data from a computing device, the device data including parameters associated with the computing device. The device analytics engine selects a set of rules in a plurality of rules that indicate at least one parameter in the plurality of parameters in the device data for determining a device identifier. The set of rules are evaluated in an order until the device identifier is determined from the at least one parameter indicated in the set of rules, the device data, and previously stored data from multiple computing devices. A score is generated for the computing device using one or more of the device identifier, device data, a set of rules, and previously receive device data that corresponds to the device identifier. A computing device is identified as a fraudulent computing device based on the score.

Abstract: There are provided systems and methods for a shared prediction engine for machine learning model deployment. A service provider may provide AI hosting platforms that allow for clients, customers, and other end users to upload AI models for execution, such as machine learning models. A user may utilize one or more user interfaces provided to a client device by the service provider to select machine learning models to perform predictive services based on input features provided in an input string. Thereafter, a machine learning engine may host and execute the models during an instance of the engine provided to the client device. The engine may then process the input features in a processing thread remotely from the client device during the instance so that machine learning predictions may be determined. Thereafter, an output string for the predictions and model explanations may be provided to the client device.

Abstract: There are provided systems and methods for a shared prediction engine for machine learning model deployment. A service provider may provide AI hosting platforms that allow for clients, customers, and other end users to upload AI models for execution, such as machine learning models. A user may utilize one or more user interfaces provided to a client device by the service provider to select machine learning models to perform predictive services based on input features provided in an input string. Thereafter, a machine learning engine may host and execute the models during an instance of the engine provided to the client device. The engine may then process the input features in a processing thread remotely from the client device during the instance so that machine learning predictions may be determined. Thereafter, an output string for the predictions and model explanations may be provided to the client device.

Abstract: Techniques are disclosed relating to a graphical user interface (GUI) for editing classification rules. The graphical user interface may include a display of a modified threshold value for a classification rule, where values on one side (e.g. above) the threshold value indicate a first type of classification, and values on the other side (e.g. below) indicate a second type of classification. Machine learning techniques may be used to suggest a modified threshold value to the user, who may accept the modified value, or may provide, via the GUI, their own modified value, which can be different from a suggested value. Graphical indications of accuracy for the classification rule may be displayed.

Abstract: Systems, methods, and computer program products for identifying a fraudulent device. A device analytics engine receives device data from a computing device, the device data including parameters associated with the computing device. The device analytics engine selects a set of rules in a plurality of rules that indicate at least one parameter in the plurality of parameters in the device data for determining a device identifier. The set of rules are evaluated in an order until the device identifier is determined from the at least one parameter indicated in the set of rules, the device data, and previously stored data from multiple computing devices. A score is generated for the computing device using one or more of the device identifier, device data, a set of rules, and previously receive device data that corresponds to the device identifier. A computing device is identified as a fraudulent computing device based on the score.

Abstract: There are provided systems and methods for a shared prediction engine for machine learning model deployment. A service provider may provide AI hosting platforms that allow for clients, customers, and other end users to upload AI models for execution, such as machine learning models. A user may utilize one or more user interfaces provided to a client device by the service provider to select machine learning models to perform predictive services based on input features provided in an input string. Thereafter, a machine learning engine may host and execute the models during an instance of the engine provided to the client device. The engine may then process the input features in a processing thread remotely from the client device during the instance so that machine learning predictions may be determined. Thereafter, an output string for the predictions and model explanations may be provided to the client device.

Abstract: There are provided systems and methods for a shared prediction engine for machine learning model deployment. A service provider may provide AI hosting platforms that allow for clients, customers, and other end users to upload AI models for execution, such as machine learning models. A user may utilize one or more user interfaces provided to a client device by the service provider to select machine learning models to perform predictive services based on input features provided in an input string. Thereafter, a machine learning engine may host and execute the models during an instance of the engine provided to the client device. The engine may then process the input features in a processing thread remotely from the client device during the instance so that machine learning predictions may be determined. Thereafter, an output string for the predictions and model explanations may be provided to the client device.

Abstract: There are provided systems and methods for predictive data aggregations for real-time detection of anomalous data. A service provider, such as an electronic transaction processor for digital transactions, may access feature data for accounts prior to the feature data being used in a live risk analysis system, for example, at a designated time and/or for a designated time period. The service provider may predetermine data values from the feature data, such as aggregates of the feature data that are for certain time periods and utilized by the live risk analysis system. This processing may be done in a batch processing job in order to determine data values for multiple accounts. These data values are prestored in an available database for a distributed computing system of the service provider. Thereafter, when the live risk analysis system requires the data values, the data values may be immediately retrieved.

Abstract: Pure machine learning classification approaches can result in a “black box” solution where it is impossible to understand why a classifier reached a decision. This disclosure describes generating new classification rules leveraging machine learning techniques. New rules may have to meet evaluation criteria. Legibility of those rules can be improved for understanding. A machine learning classifier can be created that is used to identify possible candidate classification rules (e.g. from a group of decision trees such as a random forest classifier). Classification rules generated with the assistance of machine learning may allow for identification of transaction fraud or other classifications that a human analyst would be unable to identify. A selection process can identify which possible candidate rules are effective. The legibility of those rules can then be improved so that they can be more easily understood by humans.

Abstract: Techniques are disclosed relating to tuning fraud-detection rules using machine learning. In some embodiments, a server system may maintain rule information indicative of a plurality of fraud-detection rules for a transaction system. For example, in some embodiments, the server system may implement a fraud-detection service for the transaction system. In some embodiments, the server system may select a first rule to update, where the first rule includes one or more evaluation criteria and one or more corresponding user-defined threshold values. The server system may apply a machine learning algorithm to training data associated with the transaction system to generate an updated version of the first rule. The server system may then compare a performance of the first rule and the updated version of the first rule and, based on that performance, determine whether to suggest the updated version of the first rule to the transaction system.