Despite the proliferation of AI based research lately, sometimes researchers need a human eye to make true discoveries. That collaboration was in evidence in a recent paper by Dr. Veselin Kostov, a research scientist at the SETI Institute and NASA’s Goddard Space Flight Center, who led a team of almost 1,800 to review a dataset from the Transiting Exoplanet Survey Satellite (TESS) that led to the discovery of almost 8,000 new eclipsing binary systems.
An eclipsing binary is a star system where two stars orbit each other, with one passing in front of the other from our perspective. The way to find them is similar to that of exoplanets - watch a star and look for dips in brightness. If the dip is large enough, then instead of an exoplanet, there is likely another star (albeit a faint one) orbiting that star.
Data from TESS is great for this work, as it covers around 98% of the sky watching for exactly these types of transits. However, that doesn’t mean that the researchers just handed TESS data directly to a set of volunteers. The data went through several pre-processing steps before being handed over to the public.
First, they limited the dataset to only stars with a magnitude brighter than 15. After narrowing down the sheer number of stars to look at they used a tool developed in Python called the ELEANOR pipeline to create a massive dataset of millions of light curves. Those light curves were then artificially padded to a uniform number of data points and scaled to ensure periodic changes from TESS’s observational instrument weren’t mistaken for eclipses.
After all that preprocessing was done, the researchers fed their millions of updated light curves to (you probably guessed it) an AI. This one is a relatively simple convolutional neural network that was trained to find the shape of an eclipse rather than any given periodic signal, making them more adept at finding eclipses no matter their periodicity. It was trained on some of the data with manually labeled results, and then set loose on the catalog of TESS and even Kepler data on eclipses. It successfully found around 85% of known eclipsing binaries in TESS’s data, and around 56% of them from Kepler’s data sets. Intriguingly, it also managed to find about 32% of the exoplanet candidates in TESS’s data set.
Even after all that AI processing, the dataset still wasn’t quite ready for volunteer help yet. The research team, along with a select group of trained amateurs, used a platform called Exogram to identify 10,000 initial targets, which were then released to the public on Zooniverse, a crowd-based research platform. Between September 2024 and March 2025 1,800 volunteers performed 320,000 classifications of eclipsing binary systems, while also verifying their period and assessing the quality of data used to identify them.
Fraser discusses the discoveries from TESSThe outcome of all the work resulted in the identification of 10,001 eclipsing binary systems. 7,936 of them are new to science, while the other 2,065 were previously known, but the study provided updated, more accurate, parameters for their periods, as TESS’ dataset provided better insight. There were also some particularly interesting systems that could hold new discoveries, including several that had variable eclipse timings, and plenty that might have a third star, and some that show a significant dynamic between the star being orbited and the one doing the orbiting.
All of those systems await further research, but there’s another, unspoken factor at play in this data - exoplanets. TESS was originally designed as an exoplanet hunter, and this kind of large scale AI/human collaboration of lightcurve analysis is exactly the kind of work that could potentially produce even more accurate exoplanet catalogues, as evidenced by some of the work already done in this paper. That seems to be the next step for this dataset, with Dr. Kostov telling an interviewer “I can’t wait to search them for exoplanets!” Given the data has already been collected, and the team has already been assembled, it’s very likely he’ll get his chance soon.
Learn More:
NASA - NASA Citizen Scientists Find New Eclipsing Binary Stars
UT - Astronomy Jargon 101: Eclipsing Binary
UT - Rare Eclipsing Binary Stars Provide Refined Measurements in the Universe