Open clusters (OCs) serve as invaluable tracers for investigating the properties and evolution of stars and galaxies. Despite recent advancements in machine learning clustering algorithms, accurately discerning such clusters remains challenging.
We re-visited the 3,013 samples generated with a hybrid clustering algorithm of FoF and pyUPMASK. A multi-view clustering (MvC) ensemble method was applied, which analyzes each member star of the OC from three perspectives—proper motion, spatial position, and composite views—before integrating the clustering outcomes to deduce more reliable cluster memberships.
Based on the MvC results, we further excluded cluster candidates with fewer than ten member stars and obtained 1,256 OC candidates. After isochrone fitting and visual inspection, we identified 506 candidate open clusters in the Milky Way. In addition to the 493 previously reported candidates, we finally discovered 13 high-confidence new candidate clusters.