Machine learning technique reveals hidden exoplanet in James Webb data

Astronomers have developed a new technique using machine learning to remove glare from James Webb Space Telescope data, revealing a previously hidden giant exoplanet. The findings were published Nov. 4 in the journal Nature. The method, called FLIP, creates a "digital twin" of JWST to correct for telescope movement and light distortions, allowing planets to emerge from stellar brightness. It successfully detected 51 Eridani b, a planet around a star 97 light-years away that was obscured in earlier observations. Researchers hope FLIP will help identify more exoplanets similar to those in our solar system, which are typically harder to spot than larger, more reflective worlds. The team emphasizes the approach is physics-based, not AI-driven.