‘Facial recognition’ for galaxies helps reveal secrets at heart of young galaxies

Image credit: Dreamstime

Researchers at the University of California (UC) Santa Cruz have developed a ‘facial recognition’ system for categorising galaxies.

Galaxies evolve over billions of years; astronomers cannot observe the evolution of a single galaxy and must instead look back through time at the most distant galaxies to observe them as they appeared during their early years. Due to these limitations, computer simulations of galaxies are valuable for the study of galaxies and their evolution.

Researchers at UC Santa Cruz have been using observations and simulations to study a phenomenon known as ‘blue nuggets’ in early galaxies, whereby gas flows into the centre of a galaxy form dense, star-forming regions (the nuggets).

In order to assist them with their work, the researchers used deep learning in order to teach an algorithm to categorise galaxies at different phases in their lives.

Deep learning - an approach to machine learning - uses algorithms approximately reflecting the structure of the biological brain (an artificial neural network). During deep learning, data is passed through layers of a neural network, each layer drawing features from the data. This approach is commonly used in image recognition and voice-to-text programs.

In this case, the researchers trained a deep learning algorithm using images from cosmological simulations, which generated images of galaxies as they would look when photographed from the Hubble Space Telescope. These mock-ups were used to train the algorithm to recognise three phases in the evolution of a galaxy: pre-blue nugget, blue nugget and post-blue nugget.

Slightly to their surprise, the algorithm proved very good at analysing real photographs of galaxies taken by the Hubble Space Telescope.

“We were not expecting it to be all that successful. I’m amazed at how powerful that is,” said Joel Primack, Emiritus Professor of Physics at the UC Santa Cruz’ Institute for Particle Physics. “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.”

The algorithm - working through mock-ups and photographs - found that the ‘blue nuggets’ only occur in galaxies with certain masses. This phenomenon is followed by quenching of star formation, leading to a ‘red nugget’ phase.

According to David Koo, Emeritus Professor of Astronomy and Astrophysics at UC Santa Cruz, deep learning could play a role in identifying patterns in observational data that humans cannot identify.

“Deep learning looks for patterns and the machine can see patterns that are so complex that we humans don’t see them,” said Koo. “We want to do a lot more testing of this approach, but in this proof-of-concept study, the machine seemed to successfully find in the data the different stages of galaxy evolution identified in the simulations.”

According to Koo, astronomers and computers of the future will be equipped with far more observational data due to major new survey projects such as the James Webb Space Telescope and the Wide-Field Infrared Survey Telescope.

“This is the beginning of a very exciting time for using advanced artificial intelligence in astronomy,” said Koo.

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them