Machine learning approach for stellar light curve analysis

Machine learning has emerged as a powerful tool in astrophysics, transforming the way we analyze and interpret complex datasets. In this talk, I will present a machine learning approach for analyzing stellar light curves, with a focus on predicting stellar rotation periods. The proposed method utilizes a novel deep learning architecture that combines self-supervised and simulation-based training, achieving superior results compared to traditional methods. Leveraging this model, we have created the largest stellar rotation period catalog for main-sequence stars. I will also discuss how a detailed error analysis revealed surprising insights learned by the model, highlighting the fundamental differences between data-driven and heuristic approaches.