Astronomers use AI to discover hundreds of weird galaxies in Hubble archive
In what is a perfect example of the proper use of artificial computer intelligence (AI), astronomers have now used this programming to analyze almost 100 million images taken by the Hubble Space Telescope over the decades to find any galaxies hidden there that have “anomalies” or unusual shapes.
The team analyzed nearly 100 million image cutouts from the Hubble Legacy Archive, each measuring just a few dozen pixels (7 to 8 arcseconds) on a side. They identified more than 1,300 objects with an odd appearance in just two and a half days — more than 800 of which had never been documented in scientific literature.
The six galaxies to the right are just a small sample. All six were previously unidentified, and include “three lenses with arcs distorted by gravity, one galactic merger, one ring galaxy, and one galaxy that defied classification” (the galaxy at the top left). From the European Space Agency’s (ESA) press release:
The strange, bi-polar galaxy seen here is certainly anomalous, with its compact, swirling core and two open lobes at the sides. Exactly what kind of galaxy it is is unclear, and it was not previously known to astronomers.
As noted in the first link, the volume of data that astronomers are now collecting from ground-based and orbiting telescopes — many of which are survey telescopes that photograph the entire sky repeatedly — has actually become a problem. They have great data, but don’t have the time or human resources to study it sufficiently. Even employing large numbers of ordinary citizens, working at home with their own computers, can’t get the job done.
This is the kind of grunt work that AI is ideally made for. It can quickly review the data and identify objects that don’t fit normal expectations. Humans then can do the real work, finding the most interesting of these strange objects, such as the top left galaxy, and devote human creativity to studying it.
In what is a perfect example of the proper use of artificial computer intelligence (AI), astronomers have now used this programming to analyze almost 100 million images taken by the Hubble Space Telescope over the decades to find any galaxies hidden there that have “anomalies” or unusual shapes.
The team analyzed nearly 100 million image cutouts from the Hubble Legacy Archive, each measuring just a few dozen pixels (7 to 8 arcseconds) on a side. They identified more than 1,300 objects with an odd appearance in just two and a half days — more than 800 of which had never been documented in scientific literature.
The six galaxies to the right are just a small sample. All six were previously unidentified, and include “three lenses with arcs distorted by gravity, one galactic merger, one ring galaxy, and one galaxy that defied classification” (the galaxy at the top left). From the European Space Agency’s (ESA) press release:
The strange, bi-polar galaxy seen here is certainly anomalous, with its compact, swirling core and two open lobes at the sides. Exactly what kind of galaxy it is is unclear, and it was not previously known to astronomers.
As noted in the first link, the volume of data that astronomers are now collecting from ground-based and orbiting telescopes — many of which are survey telescopes that photograph the entire sky repeatedly — has actually become a problem. They have great data, but don’t have the time or human resources to study it sufficiently. Even employing large numbers of ordinary citizens, working at home with their own computers, can’t get the job done.
This is the kind of grunt work that AI is ideally made for. It can quickly review the data and identify objects that don’t fit normal expectations. Humans then can do the real work, finding the most interesting of these strange objects, such as the top left galaxy, and devote human creativity to studying it.





























