"Their work, described in a new study published in the journal Astronomy & Astrophysics, is the first systematic search for astrophysical anomalies across the entire archive. To make the discoveries, the researchers used their AI tool, which they're calling AnomalyMatch, on nearly 100 million snippets of Hubble images that were only a few pixels on each side. In less than three days, the neural network identified over 1,300 anomalous objects, more than 800 of which had never been documented in scientific literature."
"The term "AI" has become a catch-all these days for all kinds of dubious tech of varying degrees of automation and reliability, but certain types have found a very practical and welcome use among astronomers. Using a custom-built AI tool, for instance, a team of scientists at the European Space Agency have identified over a thousand "anomalies" in an archive of Hubble space telescope images that have gone unnoticed for decades, according to a NASA release."
The universe contains trillions of galaxies and quadrillions of stars, producing massive astronomical datasets. Certain AI techniques can accelerate pattern recognition within these datasets. Archival Hubble observations span 35 years, providing a rich source for anomaly searches. A custom neural network called AnomalyMatch processed nearly 100 million tiny image snippets and, in less than three days, identified over 1,300 anomalous objects, with more than 800 lacking prior documentation. Many anomalies corresponded to galactic mergers. The tool also flagged jellyfish galaxies characterized by streams of star-forming gas extending from one side of a galaxy’s disk.
Read at Futurism
Unable to calculate read time
Collection
[
|
...
]