AI finds 44 star systems that could hide Earth-like planets
AI identifies 44 star systems that may contain undiscovered Earth-like planets based on known planetary arrangements and simulated training data.Summary
A machine-learning model developed by researchers at the University of Bern has identified 44 known planetary systems that may contain undiscovered Earth-like planets. The algorithm, created by Jeanne Davoult with Romain Eltschinger and Yann Alibert, analyzes the architecture of known planets and the properties of the innermost detectable planet to predict the presence of small, temperate worlds. It was trained on tens of thousands of synthetic planetary systems generated by the Bern Model of Planet Formation and Evolution, achieving precision scores up to 99% on simulated data. When applied to 1,567 observed systems around Sun-like and smaller stars, 44 passed a 90% voting threshold and a preliminary stability assessment. However, these predicted planets remain unconfirmed, and follow-up telescope observations are needed to verify them. The target list could help prioritize searches for missions like PLATO and the proposed LIFE concept.
(Source：The Brighter Side of News)