Artificial Intelligence Uncovers Dozens of Hidden Early Galaxies in Webb Data
A new AI system called ASTERIS successfully strips noise from James Webb Space Telescope images to reveal extremely faint galaxies from the early universe. The technique effectively doubles the detection rate of distant cosmic structures without requiring additional expensive observation time.
A new artificial intelligence network named ASTERIS gives astronomers a powerful tool to see deeper into the cosmos without needing extra observation time. By effectively stripping away background noise from space images, this self-supervised transformer-based system reveals celestial features that remain completely hidden in standard observations.
The technology tackles a fundamental challenge in astronomy where faint objects vastly outnumber bright ones but stay obscured by visual noise. While scientists traditionally stack multiple long exposures to beat down this interference, ASTERIS achieves similar or better results by processing existing data, making it a highly cost-effective solution for expensive observatories like the James Webb Space Telescope.
During a proof-of-concept test using Webb's Advanced Deep Extragalactic Survey, ASTERIS more than doubles the number of detected distant galaxies. The AI successfully uncovers dozens of early galaxies that formed during the universe's first 500 million years, pushing the boundaries of what scientists can extract from current deep-space imagery.