Researchers from University of California – Santa Cruz have developed a type of ‘deep learning’ algorithm that can be trained on images from cosmological simulations. Tests using the platform have proved successful at classifying real galaxies via Hubble telescope images.
Deep learning
Deep learning is an established method for face recognition, together with speech-recognition applications. While this has tended to be confined to assess people (see, for example the Digital Journal article “Face recognition technology works at night“). The development of algorithms that can scan images of galaxies represents an advance in the field. Such information can help astronomers and astrophysicists understand how galaxies form and evolve.
This adds to other applications of artificial intelligence with astronomy. For example, astronomers have used machine learning to combat contamination in radio data, according to Wired. Here the scientists trained a neural network to recognize and then mask the human-made radio interference that comes from satellites, airports, WiFi routers, and so on, in order to achieve better image clarity.
In another instance, Digital Journal has reported on astronomers assessing the potential of using convolutional neural networks to determine the positions and sizes of craters on the Moon. The information has helped to improve the international Lunar digital elevation map.
New algorithm for astronomy
With the new research, scientists have devised a computer simulations of galaxy formation in order to train an especially developed deep learning algorithm. To develop the algorithm, scientists took output from the simulations to produce mock images of galaxies, of the form they would look in observations by the Hubble Space Telescope. The generated images were then used to train the deep learning system to recognize important phases of galaxy evolution.
Next, the researchers then gave the system a large set of actual Hubble images to classify. The results were deemed successful, showing a high level of consistency. According to lead researcher Professor Joel Primack, this came as somewhat of a surprise: “We were not expecting it to be all that successful. I’m amazed at how powerful this is…We know the simulations have limitations, so we don’t want to make too strong a claim. But we don’t think this is just a lucky fluke.”
Of particular interest to the researchers was the evolution of gas-rich galaxies. These occur when when large flows of gas move into the center of a galaxy and fuel formation of a small, dense, star-forming region called a “blue nugget.”
The research has been published in the Astrophysical Journal. The research paper is titled “Deep Learning Identifies High-z Galaxies in a Central Blue Nugget Phase in a Characteristic Mass Range.”
This is NGC 1300. A spiral galaxy located about 69 million light years away in the direction of the Constellation Eridanus. Is one of the galaxies having a large central bar. This bar contains a spiral of 3300 light years in diameter (Photographed by Hubble telescope).
A colossal cluster of galaxies
In related astronomy news, astronomers have recently witnessed the birth of a colossal cluster of galaxies. The new observations reveal some 14 galaxies packed into an area just four times the diameter of the Milky Way’s galactic disk. Furthermore, computer simulations of these galaxies predict that the cluster will, at some point in time, assemble into one of the most massive structures in the modern universe. This observation has been reported to the journal Nature (“A massive core for a cluster of galaxies at a redshift of 4.3”).
Essential Science
This article is part of Digital Journal’s regular Essential Science columns. Each week Tim Sandle explores a topical and important scientific issue. Last week’s focus was about the U.S. Great Lakes and human drug waste. Research has uncovered levels of antidepressants not only in the water but also in the brains of fish. This signals the polluting effect of wastewater and sewage treatment facilities.
The week before we looked into a new medical discovery. Here coffee, which is well known as a psychostimulant, was revealed to effect the body’s metabolism in a similar way to cannabis.
