A “maven” is an expert—someone who knows everything there is to know about their area of expertise, be it fashion, art, food, or, in the case of DoD’s new Algorithmic Warfare Cross-Functional Team (AWCFT), machine learning algorithms.

Established in April by Deputy Defense Secretary Bob Work, the AWCFT—code-named “Project Maven”—is being led by Lt. Gen. John N.T. “Jack” Shanahan, director for defense intelligence, warfighter support with the Office of the Under Secretary of Defense for Intelligence. Shanahan described how his team is pursuing machine learning mavenry during a keynote address Wednesday at GEOINT 2017.

Adding 1,000 more intelligence analysts is neither realistic nor feasible in today’s fiscal environment. — Lt. Gen. John N.T. “Jack” Shanahan, OUSD(I)

Project Maven couldn’t come at a better or more urgent time, said Shanahan, noting the ubiquity of high-resolution commercial imagery, the pervasiveness of machine-to-machine data transmission, and the rise of activity-based intelligence as indications of a new and rapidly changing world the United States must learn to navigate faster and better than its adversaries.

Keynote: Lt. Gen. John N.T. “Jack” Shanahan, Director for Defense Intelligence, Warfighter Support, OUSD(I) from Trajectory On Location on Vimeo.

“The era of incremental change will soon fade into the rearview mirror,” Shanahan said. “We no longer live in a ‘y = mx + b’ world. The slope of the technology-changing curve is becoming so steep at each new point of the curve … that we will find it hard to keep up over the next decade. But we must do something, or risk ceding valuable ground to our potential adversaries.”

The problem is the same as the opportunity: an embarrassment of data. More analysts aren’t the answer, however. Rather, it’s smarter analysis.

“The first and perhaps most important step [to solving our data problem] is to understand that it is not possible to solve these problems with brut force alone. Adding 1,000 more intelligence analysts is neither realistic nor feasible in today’s fiscal environment,” Shanahan explained. “We must instead find creative ways to adapt to this new environment in which we are already deeply immersed. … Artificial intelligence, machine learning, and deep learning [are] the critical base ingredients in the recipe for future success.”

Enter Project Maven, whose charge is acquiring and developing machine learning algorithms that can be applied to intelligence data in a way that lets human analysts “make the best use of their time working on the hardest problems.”

Project Maven’s first objective will be augmenting or automating processing, exploitation, and dissemination (PED) of full-motion video (FMV) captured by unmanned aerial systems in support of DoD’s campaign to defeat ISIS. By the end of this calendar year, the team plans to deliver its first capabilities by: organizing a data-labeling effort; developing, acquiring, and/or modifying algorithms to accomplish object detection, classification, and alerts for FMV PED; identifying required computational resources; determining a path to fielding necessary infrastructure; and integrating algorithmic-based technology with programs of record.

In the past, for reasons entirely understandable, the approach across DoD to developing and fielding AI/machine learning capabilities was to minimize disruption. — Lt. Gen. John N.T. “Jack” Shanahan, OUSD(I)

Although it’s an ambitious agenda, the AWCFT will achieve its goals by borrowing talent, resources, and workflows—working in 90-day sprints, for example—from Silicon Valley, whose business models will help DoD achieve the “Third Offset” it so desperately needs.

“In the past, for reasons entirely understandable, the approach across DoD to developing and fielding AI/machine learning capabilities was to minimize disruption,” Shanahan concluded. “That approach is no longer [viable]. We are playing catch-up from a position of deep disadvantage, and we must move aggressively and with alacrity.”

Related


Posted by Matt Alderton