ExoMiner++

• 02 Feb 2026

Source: TH

• Recently, NASA announced the open-source release of ExoMiner++ advancing the search for planets beyond our solar system. 
• About: ExoMiner++ is a deep-learning AI model developed by NASA to detect exoplanets from space-telescope data, serving as the advanced successor to the earlier ExoMiner system. 
• Working Mechanism: It analyses stellar brightness over time, identifying characteristic dips caused when a planet transits in front of its host star. 
• Key Feature: A key challenge it addresses is distinguishing real planetary signals from false positives, such as binary stars or background objects that can mimic planetary transits. 
• Unlike black-box AI models, ExoMiner++ is explainable, providing astronomers with a confidence score and clear reasoning behind each classification. 
• Data Sources Used: ExoMiner++ is trained on both Kepler Space Telescope and  the Transiting Exoplanet Survey Satellite (TESS) datasets, allowing it to analyse many more stars simultaneously and significantly scale up detection capability. 
• Significance: So far, it has identified around 7,000 potential exoplanet candidates in TESS data and has been released as open-source software, enabling global researchers to replicate results and refine the model for future missions such as the Nancy Grace Roman Space Telescope. 

Read more: Exoplanet