What good is gathering scientific data if you gather so much that it's impossible to find the time to study it?
The answer is computers programmed with artificial intelligence, according to a new study from Lawrence Technological University that was chosen for a prominent position on the research website of the American Astronomical Society.
The research paper, by LTU computer science undergraduate student Evan Kuminski and Lior Shamir, associate professor of mathematics and computer science, was published online April 6 in The Astrophysical Journal.
Then, on April 27, Susanna Kohler, editor of AASNova.org, the online journal of research highlights from the AAS, wrote an article about the paper for the website.
The paper details how Kuminski and Shamir developed a machine learning algorithm to classify millions of galaxies whose images have been captured by automated telescopes -- but almost all of which have never been studied by scientists.
Shamir said the problem is simply one of numbers.
“The thing is, you have robotic telescopes that keep collecting images of celestial objects, hundreds of millions of images, soon to be billions of images, and there are no tools to analyze those images,” Shamir said. “No person can study and categorize that many images manually. That's not going to work.”
Shamir noted that Kohler pointed out in her article that the scientific community tried crowdsourcing the work: “They recruited over a million people worldwide, non-experts, to do that manually, but that produced about 55,000 objects after three years, and that’s not a lot out of a catalog of millions of objects.”
And the problem will only grow worse when the Large Synoptic Survey Telescope (LSST), now under construction in Chile, goes online in the late teens. The internationally financed telescope and its 3,200-megapixel camera will image the entire visible universe every few weeks, allowing scientists to observe faint, distant objects -- and changes in them -- like never before.
Shamir has used computer intelligence to analyze everything from modern art to Beatles music over his years at LTU, and had already started working on astronomy analysis when Kuminski arrived as a freshman. “When he came here, that's what he wanted to do,” Shamir said of Kuminski. “He had already published a paper by his sophomore year.”
Kuminski said astronomy “has been something I’ve been interested in going back as far as I can remember, but it was never something I planned to do anything with in college. But I’m really glad it has, because it’s become a way to put together two subjects (astronomy and computer science) that have always been a fascination of mine.”
Kuminski said the study means “we’ve found a method that can obtain useful information quickly from the data that these sky studies are generating so rapidly. It could be useful to people studying aspects of galaxies. It’s an easy way to get a large sample size very quickly.”
Kuminski, a junior from Strongsville, Ohio, said he is still unsure about his plans after graduation from LTU a year from now, but that he probably will attend graduate school. “I know I want to do something in the field of computer science, but I’ve always thought I’ll make that decision after I finish school,” he said.
The latest publication by AAS demonstrates the way undergraduates can participate in groundbreaking research at Lawrence Tech, university officials said.
“Advances in technology have transformed the way we do science. Generation of big data is outpacing human analysis in almost every field,” said Hsiao-Ping Moore, Dean of the College of Arts and Sciences at LTU. “The work by Kuminski and Shamir elegantly illustrates the direction science must take in the future. Involving undergraduate students in science discovery will also be the way of the future in science education.”
LTU President Virinder Moudgil also lauded “this very important and extremely impressive study,” and said of the authors, “I am proud of their work and its impact on science and our understanding the relationship between computers and the human mind.”