A group of scientists at the University of Georgia has developed a novel technique, facilitated by machine learning, to detect and categorize distant planets beyond Earth.
By leveraging artificial intelligence, they successfully identified an exoplanet by analyzing protoplanetary disks, the gas surrounding newly formed stars. This breakthrough signifies an initial stride towards utilizing machine learning to uncover previously overlooked exoplanets.
Jason Terry, the lead author of the study, explained that while they confirmed the planet’s existence using conventional methods, their models guided them to conduct simulations and pinpointed the precise location of the planet.
By applying these models to older observations, they identified a disk that had not previously been recognized to host a planet, despite undergoing prior analysis. Subsequent simulations confirmed that a planet could account for the observed data.
The models employed by Terry and his team detected the presence of a planet through detailed scrutiny of multiple photographs, revealing a specific feature in the disk—remarkable variations in gas velocity—indicative of a planet’s influence. Cassandra Hall, assistant professor of computational astrophysics and principal investigator of the Exoplanet and Planet Formation Research Group at UGA, expressed excitement over this proof of concept, highlighting its potential to make groundbreaking discoveries in addition to identifying known exoplanets. She emphasized how machine learning and their models, in particular, can rapidly and accurately identify crucial information that may be overlooked by human observers, potentially accelerating analysis and facilitating new theoretical insights.
Terry further noted that analyzing an entire catalogue and finding strong evidence for a new planet in a specific location only took around an hour, suggesting the significant role these techniques can play as datasets continue to expand. The team envisions machine learning’s ongoing contribution in advancing exoplanet research, enabling scientists to explore and understand the universe more efficiently.
