Machine Learning Is Transforming How We See the Universe

September 11, 2018 - 5 minutes read

machine learning app developer

For centuries, astronomers have utilized math to make sense of the unfathomable amount of data set before us every night in the sky. Today, our instruments capture more data than we can humanly process and work with.

Researchers are now turning to machine learning development to help them better organize the immense amount of data and understand the universe.

Big Data = a Big Problem

Usually, it’s safe to assume that you’d find the biggest tech or scientific developments in San Francisco or another well-known innovation hub. But for the latest groundbreaking work in astronomy, you’ll have to turn your gaze to Australia, New Zealand, and South Africa.

Set to span over 2 million low-frequency antennas and 2,000 radio dishes, the Square Kilometer Array (SKA) is a work-in-progress that will be able to map gargantuan chunks of the night sky — and push data processing to the absolute limit. It’s expected to produce over an exabyte of data every day. This is more than the data required for the world’s daily internet usage.

Automating Astronomy

To handle this oncoming onslaught of data, researchers are already working on ways to automate and streamline its processing. One of the most exciting examples of this is the monitoring of explosions in space. Dr. Gemma Anderson, a research associate of the International Centre for Radio Astronomy Research, programs radio telescopes to find gamma-ray bursts.

“We have a telescope in space designed to look for explosions. The space telescope sends information back to Earth, and I have two of the big radio telescopes in Australia set up to receive that signal,” she elaborates. “When they get the signal, these telescopes stop what they’re doing and try to observe the explosion as quickly as possible.”

In the past, it was possible for astronomers to handle and analyze the data from these telescopes by themselves. But today, automation is the name of the game. “For those people interested in being an astronomer, it’s very important to get experience in computer programming. We need to become more proficient in processing and analyzing large amounts of data,” says Dr. Anderson.

For the type of data that Dr. Anderson deals with, supercomputers are needed. They allow for data analysis to take a fraction of the time it would usually require on standard computers. But it won’t be long before data processing protocols are overdue for another boost in efficiency.

Enter machine learning.

Training the Next Generation of Astronomers

Dr. Rebecca Lange works at the Curtin Institute for Computation and Astronomy Data and Computing Services. She’s been focusing on helping astronomers leverage machine learning to turn raw data into useful information. Mimicking the human brain’s process for recognizing patterns, this niche of AI helps astronomers automatically detect and classify “sources” in large images.

Sources are objects in space that emit radio signals. In this case, machine learning analyzes bright spots in an image and labels them accordingly. For example, it can tell the difference between a spiral-shaped and elliptical galaxy. “Machine learning is getting picked up because we now have the amount of data needed, ” Dr. Lange says. “When you’re doing supervised learning, you need a lot of data to train on. If you look at galaxy classification, we have done so many already that we have a great training sample for machine learning.”

Dr. Lange hopes that machine learning will give researchers a break from tedious data analysis so they can focus on more important matters. But, as she explains, it’s still a new tool: “Astronomers are still getting used to machine learning. They’re still experimenting: what algorithm works best, what kind of machine learning techniques are most useful to apply.”

It will be exciting to watch how machine learning helps to further our understanding of space. It’s already allowing us to see things from a different perspective never possible before. What other scientific fields of study do you think could benefit from a dose of AI? Let us know in the comments!

Tags: , , , , , , , , , , , ,