A new AI system that detects flares erupting from stars could help astronomers find habitable planets, according to the tool’s inventors.

The neural network finds the light patterns of a stellar flare — that may incinerate the atmospheres of planets forming nearby. The frequency and location of these flares can hence indicate the top places to look for habitable planets.

Astronomers normally start looking for the flares through a time consuming procedure of assessing dimensions of star brightness. The AI tool can make their job quicker and more successful.
Also read: Computer Vision: Businesses Transforming from an Unrealistic Idea to Practical Reality

The researchers trained the neural network on a dataset of recognized flares and not-flares, then applied it to a dataset of over 3,200 stars. It found over 23,000 flares across tens of thousands of young celebrities.

The researchers discovered that stars like our sun merely have a few endings, which appear to diminish in quantity after about 50 million decades.

“That is excellent for fostering planetary atmospheres — more moderate stellar surroundings usually means the atmospheres have a better prospect of living,” stated Adina Feinstein, a University of Chicago graduate student and first author on the paper.

But, cooler stars known as red dwarfs tended to have more regular flares, which might make it difficult for the planets that they host to keep any air.
Also read: Use of Artificial Intelligence in 2020

The researchers plan to alter the neural network so that it can search for planets near young stars.

“This will lead to some’growth of the machines’ where we could use machine learning algorithms to discover a lot of exciting new planets with the very same procedures,” said UNSW Sydney’s Dr Ben Montet, co-author of this analysis.