Planet-hunting telescopes have excelled at the task of finding new potential candidates, as thousands have been observed beyond the borders of our solar system. However, researchers have to analyze a massive amount of data to determine which are real planes.
A team of British researchers has harnessed the power of artificial intelligence to make the task of tracking down the real planet a lot easier, developing a new tool that will allow astronomers to process a higher amount of data at a faster pace.
Real planets and pretenders
The Transiting Exoplanet Survey Satellite and other telescopes find planets by tracking the way in which the brightness of a star dips when an object passes in front of it. However, the dip in brightness can also be caused by a variety of other factors, including a glitch, asteroids, or dust.
A machine learning algorithm was created and then trained with the help of data related to confirmed planets and false positives, which were collected during the Kepler mission. Once the AI was ready, the researchers fed it a batch of unconfirmed planet candidates, and 50 planets were found.
The validation of the planet is an essential step as it allows researchers to focus on areas without wasting time. An object which features a less than 1% chance of being a false positive is deemed to be a valid planet, according to the researchers.
TESS has already spotted more than 2,100 candidates, and the PLATO Mission will contribute to the number. According to scientists who contributed to the project, the algorithm requires further training at this point, but in the future, it will be able to be even more efficient as it continues to process and learn new information.
More information can be found in a paper that was published in a journal.