California-based renowned chip maker, Intel has been selected by the US military research wing, DARPA in order to lead a machine learning security technology led by DARPA itself. The research wing aims to enhance cyber-defenses against deception attacks on machine learning models.
As it is known that machine learning is a kind of AI that allows systems to enhance over time with new experiences along with the new data. Today, the most common use case of this technology is object recognition. Machine learning can also be helpful with other computers including autonomous vehicles, to identify the objects and signs on the road.
But, according to DARPA, the deception attacks can meddle with machine learning algorithms and subtle changes to real-world objects can have disastrous consequences. As an example, McAfee researchers had tricked a Tesla by adding a two-inch piece of tape on a speed limit sign and the vehicle got accelerated by 50 miles per hour above its intended speed. The research was just to experience how one can manipulate a device’s machine learning algorithms.
And that’s the thing where DARPA wants to eradicate these types of problems. The research arm was previously working on a program named the Guaranteeing AI Robustness against Deception (GARD). According to the research team, the existing mitigations against machine learning attacks are typically rule-based and pre-defined, against which, DARPA is focused to develop GARD into a system that will have broader defenses to address a number of different kinds of attacks.
In this selection, it has been decided that Intel will serve as the prime contractor for the four-year program with Georgia Tech.
Jason Martin, the principal Engineer at Intel Labs who leads Intel’s GARD team states, “Intel and Georgia Tech will work together to enhance object detection and to improve the ability for AI and machine learning to respond to adversarial attacks.”