Artificial intelligence (AI) has confirmed the existence of 50 new planets by analysing old data from NASA telescopes.
As part of their mission to find new planets, scientists use telescopes such as NASA’s Transiting Exoplanet Survey Satellite (TESS) to look for signs that a planet is passing between the telescope and a star.
This movement, known as transiting, results in a telltale dip in light from the star. However, the dip isn’t always caused by a planet, and could instead be the result of a binary star system, interference from an object in the background or even slight errors in the camera.
If the cause of the dip isn’t a planet it leads to a false positive in the telescope data, meaning scientists have to sift through the data to figure out which findings are actually planets, and which are false positives.
Understandably, this is a big job, but now scientists have found a way to have AI do much of the heavy lifting.
Researchers from the University of Warwick’s Departments of Physics and Computer Science, as well as The Alan Turing Institute, developed a new machine learning algorithm that is trained to analyse a sample of potential planets and determine which are real and which are ‘fakes’.
The AI was trained to recognise real planets using two samples of confirmed planets and false positives from the now-retired Kepler mission. Researchers then used the algorithm on a dataset of planetary candidates from Kepler that were yet to be confirmed.
Using its knowledge of real planets and false positives, the AI was able to confirm the existence of 50 new planets, the first to be validated by machine learning.
The newly confirmed planets range from being as large as Neptune to smaller than Earth, with orbits lasting as long as 200 days to as little as a single day.
Dr David Armstrong, from the University of Warwick Department of Physics, said researchers hope to apply the AI algorithm to large samples of candidates from current and future missions like TESS and PLATO.
In terms of planet validation, no-one has used a machine learning technique before. Machine learning has been used for ranking planetary candidates but never in a probabilistic framework, which is what you need to truly validate a planet.
Rather than saying which candidates are more likely to be planets, we can now say what the precise statistical likelihood is. Where there is less than a 1% chance of a candidate being a false positive, it is considered a validated planet.
We still have to spend time training the algorithm, but once that is done it becomes much easier to apply it to future candidates. You can also incorporate new discoveries to progressively improve it.
Fast, automated systems like this that can take us all the way to validated planets in fewer steps let us do that efficiently.
Once it is built and trained, the algorithm works faster than existing techniques and can be completely automated, meaning it could analyse the potentially thousands of planetary candidates observed in current surveys like TESS, and allow astronomers to prioritise the confirmed planets for further observations.
The researchers argue that the algorithm should be one of the tools to be collectively used to validate planets in future.
If you have a story you want to tell, send it to UNILAD via [email protected]
University of Warwick