Machine Learning Enhances Exoplanet Detection by Reducing Starlight

• A team led by Northwestern University developed a machine learning technique to reduce starlight in telescope images.
• The method improved sensitivity to faint exoplanets by up to 15 times in tests on James Webb Space Telescope data.
• The technique uses a physics-based telescope model combined with optimization techniques from machine learning.
• This approach could make imaging exo-Earths easier for future observatories.
• The method overcomes the telescope-stability barrier that has limited previous exoplanet detection methods.