Abstract: A system for classifying tasks based on workflow patterns detected on workflows through a real time video feed that shows steps being performed to accomplish a plurality of tasks. Each task is associated with a different set of steps. The system accesses a first set of steps known to be performed to accomplish a first task on the webpages. The first set of steps is represented by a first set of metadata. The system extracts a second set of metadata from the video feed. The second set of metadata represents a second set of steps to perform a second task. The system determines whether the second set of metadata corresponds to the first set of metadata. If it is determined that the second set of metadata corresponds to the first set of metadata, the system classifies the second task in a class to which the first task belongs.

Abstract: A system for detecting errors in task workflows from a real time video feed records. The video feed that shows a plurality of steps being performed to accomplish a plurality of tasks through an automation process system. The system splits the video feed into a plurality of video recordings which are valid breakpoints determined through cognitive Machine Learning Engine, where each video recording shows a single task. For each task from among the plurality of tasks, the system determines whether the task fails and the exact point of failure for that task. If the system determines that the task fails, the system determines a particular step where the task fails. The system flags the particular step as a failed step. The system reports the flagged step for troubleshooting.

Abstract: Systems, methods, and apparatus are provided for a dynamic contract payment term (“payterm”) generator. A machine learning algorithm may generate a replacement payment term for a contract based on market-based parameters and blockchain metadata for the contract. The blockchain metadata may encode hierarchical interdependencies between contracts using blockchain encryption. The blockchain metadata may be applied to auto-generate machine learning inputs for related contracts having interdependent payment terms. The machine learning inputs may include contract parameters that have been extracted and encrypted as blockchain metadata, as well as market-based parameters extracted from enterprise sources.

Abstract: Configuration parameter values associated with executing an application on a computing system may be determined by computational optimization based on configuration parameter values and/or monitored performance metrics associated with executing the application on the computing system. Configuration parameter values associated with executing the application on the computing system may be updated based on monitored performance metrics associated with executing the application.