Georgia Tech, DARPA Work to Stop Crime Before It Happens

The U.S. government is one step closer to being able to track your every move online, thanks to the Big Data and high performance computing abilities of ADAMS (Anomaly Detection At Multiple Scales). It's designed to keep an eye on government workers with security clearances to detect future covert action on American soil, but one might legitimately ask, who's next?

It sounds like the plot from a Hollywood movie about the future, perhaps "Minority Report." But this is very real. ADAMS is a project sponsored by the Defense Advanced Research Projects Agency (DARPA) and created by Georgia Tech School of Computational Science and Engineering and the Georgia Tech Research Institute. It's basically a set of algorithms created “to detect anomalous activity before the damage is done,” according to the Defense Systems website.

Detecting activity before it happens does sound like, "Minority Report," the movie starring Tom Cruise in which in the future criminals are arrested before they commit the crime.

As we all know predictive analytics -- maybe this should be called preventive analytics -- is all about finding patterns in events. Usually the events that predictive analytics looks for are around numbers and structured data. This time it is about finding those patterns in unstructured data, such as online work activity, “everything from e-mail messages to instant messages, file access and Web traffic," according to Defense Systems. And when it finds something anomalous it sends an alert. Perhaps Men in Black will be waiting at the offender's door.

One of ADAMS's creators, David Bader, co-principal investigator of the project and a professor at Georgia Tech, says the program does not use the traditional form of pattern recognition, which gives off too many false positives. Rather, Bader, says, it uses a new kind of graphing analysis coupled with machine learning and newly designed algorithms to find those anomalous events.

Putting it in simple terms, here’s how I think of it: If a golfer gets hit by lightning on the golf course, that’s not an anomalous event worth noting. But if this old duffer shoots 10 under par, call the FBI.

Bader says the system is at present focused on insiders who are already working with a security clearance for the government or a government contractor.

“Our system tries to find those individuals who have gone down that slippery slope, but before they’ve done any crime or anything illegal. We are finding people who are on the road to going bad,” he says.

The project does what an analyst cannot do, analyze massive data sets using high performance computing systems. It monitors user activity, compiling as much as several terabytes per day by tracking everything a user might do, from accessing files to plugging in a USB key to sending e-mails. It looks at such events and either raises or lowers the threat level depending on the results of the analysis. Bader even claims the system could find individuals who might not even know they were going to turn.

“We look at someone shifting their work patterns by five minutes a day until they end up coming in at night. We are trying to understand changes about what they eat at the cafeteria or why they unexpectedly come in at 2 a.m.”

How would anyone know what you had for lunch? Simple, require a card swipe to pay for it.

So far ADAMS, also known as Prodigal, is in use only on those working in the defense industry who have traded in their right to privacy in exchange for a security clearance. The question is, Will ADAMS someday go further?

Bader says ADAMS is only used for national defense. However, he also said it could be used in the commercial world, citing as one example using ADAMS to uncover insider trading in the financial sector.

And of course, the United States, like most countries with advanced technical capabilities, sells its technology somewhat indiscriminately to democracies and not-so-democratic countries alike.

Will ADAMS, Prodigal or something like it be also be sold to governments worldwide without asking the intentions of the buyers?

There’s a good chance we may never know.