Mind Meld for Sat Sort
Sorting through satellite imagery is tough. There are tons and tons of material, only a fraction of which can be reviewed in anything resembling a timely fashion. And very little of that is of any military use at all. Software systems can help, a bit. But, according to the mad scientists at Darpa, "the human visual system is still the best target detection apparatus" there is.
The agency would like to harness that system better. Not just the conscious mind. But the automatic and instant firing of neurons that goes on every time we take a look at something.
"Preliminary research shows that an analyst’s brain registers the discovery long before the [imagery] analyst becomes cognitively aware of it. Thus, the brain can signal the discovery three times faster than the analyst can respond," agency program manager Amy Kruse told the DarpaTech conference last August.
As part of her "Neurotechnology for Intelligence Analysts" (NIA) effort, Kruse wants researchers to "discover and characterize the neural signatures for target detection events in the human brain." The goal of the year-long study is to demonstrate "an image 'triage' system in which subjects are rapidly shown static imagery. Signals are classified in real time and the corresponding imagery shown is then sorted based on the classification of the neural signatures into sets of images that contain targets/regions of interest versus those that contain none."
Lotsa luck.
My guess is that DARPA is really looking for methodologies for cortical identification that could be tied to "lie detection" among many other things. It is true that the p300 cortical evoked potential has some application, but utilization and interpretation of those signals is difficult and have to be tuned to the questions. My guess is that proper interface design and application of novel methods of clustering data would be far more beneficial to the elucidation of remote sensing data. It's too bad because I have a project to analyze retinal and neural anatomy and physiology using technologies derived from remote sensing applications that would be right up their alley if they chose to fund "more practical" projects.
Posted by: BWJones at October 31, 2005 1:53 AM