News Release

Team of ORNL 'agents' working to keep people safe

Keep people safe

Peer-Reviewed Publication

DOE/Oak Ridge National Laboratory

OAK RIDGE, Tenn., June 16, 2004 -- Thousands of special agents created at the Department of Energy's Oak Ridge National Laboratory are on missions 24 hours a day as they work to uncover threats to national security.

These agents, which are actually intelligent software programs, scan the Internet, satellite images, hundreds of newspapers and electronic databases worldwide as they search for anything that even hints at a plot. In addition, the agents reproduce and spin off special-purpose agents that assist in the massive effort to scan more data than is humanly possible to analyze.

"The challenge is to take an incredible amount of information and very quickly determine what represents a true threat to our safety," said Thomas Potok, who leads a team of researchers in the lab's Computational Sciences & Engineering Division. "It's like having a stack of 100,000 pages and having to find the 20 pages that contain information critical to national security."

By using computers to gather data and reduce the information to what is relevant, the intelligence community can concentrate on analyzing just the meaningful information. Thus, people can make quick and accurate decisions based on hard data instead of relying on instincts or gut reactions.

"We're trying to marry what people do best with what computers do best so we can reduce the amount of information that we have to deal with," Potok said. ORNL's intelligent software agent research actually began in the late 1980s and has involved a number of clients, including the military, the intelligence community, Battelle and DOE.

For the military, Potok describes a future battlefield in which intelligent software agents gather information from multiple sources, analyze it instantly and send information to a commander or command center. As a result, instead of a commander being bombarded with information and having to mentally determine priorities, the officer is fed information in order of its importance.

Turning to the homeland, cameras at airports, seaports, sporting events and other major public gathering places can be linked to agents that quickly scan the scene for threats. Researchers create agents that do this, for example, by programming them to look for objects that are out of place or anything that has changed since the last scan.

"With satellite imaging and the many other tools we have at our disposal, it gets rather sophisticated," said Potok, adding that his colleagues are a creative lot who write programs that result in highly effective agents.

"In creating intelligent agents, we think beyond the normal bounds of software engineering," Potok said. "We look at biological and natural models such as a school of piranha or a flock of birds as patterns for our agents. We also look at the breeding and natural selection as a blueprint for finding solutions to complex problems."

The challenges are immense, but so are the rewards.

"Ultimately, our goal is to be able to detect an imminent threat that no one has been able to see with conventional methods," Potok said. "So designing systems that safeguard people and help the military be more effective makes this job rewarding, and it makes coming to work every day exciting."

Remaining challenges include scalability, which involves figuring out how thousands -- or even millions -- of sensors and agents can communicate with each other and people. Another challenge involves developing agents that more closely mimic brain functions.

While ORNL has made great progress in the area of intelligent software agents, Potok expects greater progress as ORNL's computing power continues to increase. In May, DOE selected ORNL as the site where it will build the world's largest supercomputer.

###

ORNL is managed by UT-Battelle for the Department of Energy. Funding for the project is provided by the departments of Defense and Homeland Security.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.