Identification of asteroid families' members

Domingos R.C.; Huaman M.; Lourenço M.V.F. · Machine Learning for Small Bodies in the Solar System (2024) · DOI: 10.1016/b978-0-44-324770-5.00007-6

Resumen

Asteroid families are groupings of objects with a common origin (parent body) generated by collision events, rotational fission of the parent body, characteristic ejection velocities, and a consequence of the dynamic region where they survive. The hierarchical clustering approach identifies these groupings in proper elements or frequency domains. However, HCM needs to improve accuracy in regions of high population density, where it is almost impossible to differentiate members between neighboring families. The gradual increase in large, reliable databases of asteroid proper elements has generated the need to use more sophisticated algorithms, such as machine learning or genetic algorithms. This chapter reviews supervised, unsupervised, and genetic algorithms that classify new asteroid family members. The best free hyperparameters (FP) were compared to determine the most effective algorithm. In comparison, genetic algorithms were observed as a more optimal tool; an efficient and faster alternative was obtained by obtaining more optimal hyperparameters.

Instituciones: Universidad Tecnológica del Perú