International Business Machines Corporation

Kirk Woods

Manus Kevin McHugh

Barry Michael Graham

Schuyler Bruce Matthews

Lee Thomas

In an approach to predicting software security exploits by monitoring software events, sets of design events are received from a compliance monitor, where the sets of design events are captured by the compliance monitor during execution of a software application. A pattern of the sets of design events performed by the software application is detected, where the pattern is a specific sequence of the sets of design events. The pattern of the sets of design events performed by the software application is compared to a database of learned patterns using a deep learning model. A security exploit is predicted based on the comparison of the pattern of the sets of design events performed by the software application to the learned patterns.

Predicting software security exploits by monitoring software events

In capturing software element state changes during software application runtime, a computing device executes a software application that a set of software sensors integrated within executable code of the software application. The software sensors correspond to software elements in the software application. During runtime, a given software sensor corresponding to a given software element continuously determines whether the given software element changes state. In response to a change in state, the given software sensor captures and stores the current state of the given software element. An application control module associated with the software application obtains state data stored by the software sensors. The application control module determines commands for modifying a behavior of the software application based on an analysis of the state data and executes the commands.

Growing community of inventors

Lathrop, MO, United States of America

Kirk Woods