New space computer program thinks like us
A computer algorithm modeled after the human brain will see billions of galaxies as we do and categorize them.
Thu, Jun 03, 2010 at 01:34 PM
Many of us are familiar with HAL, the feisty computer from the film "2001: A Space Odyssey". The HAL 9000 flew a ship carrying several astronauts into space — and “he” famously seemed to act and even feel like a human. It may be 2010, but it looks like computers that think like people are a step closer to reality. Space.com reports on a new study about a computer algorithm based on the human brain. This computer can recognize different types of galaxies just as humans do. Researchers are optimistic that such a computer can sort spiral, elliptical, barred, irregular and possibly undiscovered kinds of galaxies much more easily.
Scientists at University College London and the University of Cambridge point out that the new computer has agreed with human classifications of galaxies about 90 percent of the time. Astronomers say this device should help them keep up with the immense amount of galaxy images provided from places such as the Sloan Digital Sky Survey and the Galaxy Zoo. Experts say classifying galaxies will go a long way towards understanding their origins and evolution.
Manda Banerji is an astronomer at the University of Cambridge. As she told Space.com, “Next generation telescopes now under construction will image hundreds of millions and even billions of galaxies over the coming decade … The numbers are overwhelming and every image cannot viably be studied by the human eye." The algorithm was developed after 250,000 people helped astronomers at the Galaxy Zoo sort galaxies. These classifications were then used to train the computers to sort the stars as humans do.
Called the artificial neural network, this new algorithm looks at the shape, size and color of astrophysical objects. Using a “thought process” similar to one used in the human brain, it then picks the appropriate galaxy type. This apparently mimics the biological neural network found in people. Including the objects studied in the Galaxy Zoo project, the algorithm was trained on 75,000 sky objects from the Sloan Digital Sky Survey. It was then tested on 1 million objects in space.
Ofer Lahav is an astrophysicist at the University College London. He tells Space.com that this is a huge step towards understanding our universe. According to Lahav, "While human eyes are very efficient in recognizing patterns, clever computational techniques that can reproduce this behavior are essential as we begin to push the boundaries of our observable Universe and detect more distant galaxies."
For further reading: