Applying Machine Learning to the Universe's Mysteries


Researchers at the U.S. Department of Energy's Lawrence Berkeley National Laboratory and international collaborators have demonstrated computers' readiness to solve the universe's mysteries. The team used thousands of images from simulated high-energy particle collisions to train neural networks to identify important features. They found the networks were up to 95-percent successful in recognizing important features in a sampling of about 18,000 images. The researchers say machine-learning algorithms enable the networks to improve their analysis as they process more images, with the underlying technology employed in facial recognition and other types of image-based object recognition applications. "With this type of machine learning, we are trying to identify a certain pattern or correlation of patterns that is a unique signature of the equation of state," says Long-Gang Pan of the University of California, Berkeley.