Machine learning-based system for dynamic variable determination and labeling

A method, comprising:

• receiving, by a computing platform, the computing platform having at least one processor and memory, and from a plurality of data sources, historical user data;
• training, by the at least one processor and using the historical user data, a machine learning model to generate a plurality of dynamic variable profiles applicable to a plurality of users, each profile of the dynamic variable profiles being associated with a different user variable, and evaluate data for potential unauthorized activity based on the plurality of dynamic variable profiles;
• receiving, by the at least one processor, user specific data for a user, wherein the user specific data includes data associated with a requested user event;
• associating, by the at least one processor, one or more dynamic variable profiles of the plurality of dynamic variable profiles with the user, wherein associating the one or more dynamic variable profiles with the user includes identifying one or more first categories associated with the user based on at least a job role of the user, identifying one or more user variables associated with each first category of the one or more first categories, and identifying the one or more dynamic variable profiles associated with the identified one or more user variables associated with each first category of the one or more first categories associated with the user;
• executing, by the at least one processor and using the user specific data and the associated one or more dynamic variable profiles as inputs, the machine learning model to detect an anomaly in at least one dynamic variable profile;
• responsive to not detecting an anomaly, processing, by the at least one processor, the requested user event;
• and
• responsive to detecting an anomaly, flagging, by the at least one processor, the user and requested user event as potentially unauthorized, wherein flagging includes:
• identifying, by the at least one processor a mitigating action;
• and
• transmitting, by the at least one processor, the mitigating action to one or more computing devices, wherein transmitting the mitigating action to one or more computing devices causes the one or more computing devices to execute the mitigating action;
• and
• updating, by the at least one processor, the machine learning model based on the detected anomaly and the identified mitigating action.

One or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, memory, and a communication interface, cause the computing platform to:

• receive historical user data from a plurality of data sources;
• train, using the historical user data, a machine learning model to generate a plurality of dynamic variable profiles applicable to a plurality of users, each profile of the dynamic variable profiles being associated with a different user variable, and evaluate data for potential unauthorized activity based on the plurality of dynamic variable profiles;
• receive user specific data for a user, wherein the user specific data includes data associated with a requested user event;
• associate one or more dynamic variable profiles of the plurality of dynamic variable profiles with the user, wherein associating the one or more dynamic variable profiles with the user includes identifying one or more first categories associated with the user based on at least a job role of the user, identifying one or more user variables associated with each first category of the one or more first categories, and identifying the one or more dynamic variable profiles associated with the identified one or more user variables associated with each first category of the one or more first categories associated with the user;
• execute, using the user specific data and the associated one or more dynamic variable profiles as inputs, the machine learning model to detect an anomaly in at least one dynamic variable profile;
• responsive to not detecting an anomaly, process the requested user event;
• and
• responsive to detecting an anomaly, flag the user and requested user event as potentially unauthorized, wherein flagging includes:
• identifying a mitigating action;
• and
• transmitting the mitigating action to one or more computing devices, wherein transmitting the mitigating action to one or more computing devices causes the one or more computing devices to execute the mitigating action;
• and
• update the machine learning model based on the detected anomaly and the identified mitigating action.