How Things Work: Future Attribute Screening Technology

It is hard to argue that flying home for mid-semester break has gotten easier since Sept. 11. Security is tighter, lines are longer, and anybody with a standard tube of toothpaste could be a potential terrorist. In a few more years, however, these inconveniences may be only bad memories, thanks to a project under development at the Department of Homeland Security (DHS). This project, known as the Future Attribute Screening Technology, aims to revolutionize security screening in airports, stadiums, and other potential terrorist targets.

The Future Attribute Screening Technology (FAST), formerly known as Project Hostile Intent, aims to detect potential security threats based on behavioral factors. Instead of trying to find bomb components in a suitcase, future security screeners will be able to identify telltale signs of deception, including subtle reactions to nervousness and aggression. FAST consists of an array of sensors that work together to detect possible signs of hostile intent. Almost all of these sensors are commercially available, but the complete array and the software that makes it all work, are being developed by the Human Factors Division of the DHS’s Science and Technology Directorate. Work on the system began in 2007, and tests have been conducted throughout the past two years.

More realistic tests with a more compact array and random subjects have recently been conducted, and researcher Daniel Martin claims that results have turned out “significantly better than chance,” according to an article published Oct. 6 on The FAST system is in many ways the technological successor to the polygraph. Both are based on the premise that physiological reactions are indicative of psychological attitudes. By combining data from all the available sensors, analysts — be they human or software — should be able to identify individuals who match certain suspicious profiles. According to a May 2007 Department of Homeland Security newsletter, FAST involves two main processes: developing models that match physiological cues to threat level, and the actual creation of the hardware suite to detect these cues.
The combined system would be able to identify threats without a human screener, thereby eliminating both time and subjective bias from the screening process. One of the stated goals of the project is to perform this task “without ever impeding the flow of travelers.” While both of these tasks are on the cutting edge of science, DHS researchers have made significant progress. Model matching algorithms, particularly those dealing with human physical responses, have been at the forefront of machine learning research for several years, with many advances happening here at Carnegie Mellon. The individual components of the FAST system, while available by themselves, have not previously been used in such a complex environment, and researchers also had to combine the data from these sensors into a meaningful result.

Proponents of FAST and similar technologies believe that the future of security screening lies in being able to screen people, not their belongings. Their goal is to make screening more accurate and more rapid while reducing the controlled chaos that is an airport security line. These supporters hold that the variety of data sources will eliminate most false positives. As with any new technology, particularly new security Technologies, FAST has no shortage of critics. Any system that collects the amount of data FAST does has significant privacy implications. While members of the project maintain that no data will be saved about an individual after screening, this does not satisfy skeptics who worry that what equates to a detailed medical examination is being conducted without the necessary safeguards.

Furthermore, some experts believe that the premise of physiological detection is invalid. In an interview for the same story, Carnegie Mellon’s Stephen Fienberg, a university professor in the statistics and machine learning departments, said, “I haven’t seen any research that shows that those measures from the autonomic nervous system ... measuring blood pressure, measuring breathing, measuring heat on the face, are at all related to intent.” Researchers aim to begin field testing in 2011 and hope to start implementing the system within five years.