Computers to Sift Through Stars

Computer Scientists at Carnegie Mellon University are turning to the stars for their next project. The school picked up a $1.6 million grant to build computer algorithms that will help astrophysicists sift through hard drives full of data. CMU associate research professor Jeff Schneider says astronomers set up telescopes to continually scan the sky, taking hundreds of thousands of pictures of whatever is out there. He says there is no way researchers can look at them all. That is where the computer scientists come into the mix. The pictures will be fed into the computers and then scanned using the new algorithms. Schneider says at first the computers will learn what known objects look like and then be asked to look for things out of the ordinary. They may also be told to look for a specific color or light intensity. From there Schneider says he hopes to build algorithms that will allow the computers to start making decisions on their own. He says eventually the computers may be able to scan a set of pictures, compare the images of various galaxies and then tell astrophysicists what they may find interesting. Schneider says, “Computers have long helped scientists make discoveries by processing and analyzing observational data, but now we will need computer programs that also can make discoveries on their own.” The CMU researchers will team up with others at Johns Hopkins University and the University of Washington to do the work over the next three years.
“Most algorithms for discovering dynamic evolution assume that you have a sequence of observations to analyze,” Schneider says. “But in cosmology, you never get to see things evolve. Instead, you see a bunch of objects that are at different points on the evolutionary path. We need a way to look at those objects and use them to infer the evolutionary path and where each object might be on that path.” Schneider says some of the algorithms will run over just a few hours but as they get more complex and the data sets get larger they may run for days or even weeks. The researchers will be using data from the Sloan Digital Sky Survey, a New Mexico telescope that during its first eight years of operation amassed a dataset that includes 930,000 galaxies, 120,000 quasars and 460,000 stars.