Exploring symbolic regression and genetic algorithms for astronomical object classification
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917954494398464 |
|---|---|
| author | Llorella, Fabio R. Cebrián, José A. |
| author_facet | Llorella, Fabio R. Cebrián, José A. |
| contents | This study explores the use of symbolic regression (SR) combined with genetic algorithms (GA) to classify astronomical objects. Using the SDSS17 dataset from Kaggle, which includes 100,000 observations of stars, galaxies, and quasars, we applied SR to 10\% of the data to derive a mathematical expression capable of distinguishing these classes. A genetic algorithm was then employed to optimize the hyperparameters of the expression, refining the model's performance. The final model achieved a Cohen's kappa value of 0.81, indicating a strong agreement with true classifications. Our results demonstrate that the SR+GA approach can produce interpretable and accurate models for the classification of astronomical objects, offering a promising alternative to traditional black-box machine learning methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_09220 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Exploring symbolic regression and genetic algorithms for astronomical object classification Llorella, Fabio R. Cebrián, José A. Astrophysics of Galaxies Instrumentation and Methods for Astrophysics This study explores the use of symbolic regression (SR) combined with genetic algorithms (GA) to classify astronomical objects. Using the SDSS17 dataset from Kaggle, which includes 100,000 observations of stars, galaxies, and quasars, we applied SR to 10\% of the data to derive a mathematical expression capable of distinguishing these classes. A genetic algorithm was then employed to optimize the hyperparameters of the expression, refining the model's performance. The final model achieved a Cohen's kappa value of 0.81, indicating a strong agreement with true classifications. Our results demonstrate that the SR+GA approach can produce interpretable and accurate models for the classification of astronomical objects, offering a promising alternative to traditional black-box machine learning methods. |
| title | Exploring symbolic regression and genetic algorithms for astronomical object classification |
| topic | Astrophysics of Galaxies Instrumentation and Methods for Astrophysics |
| url | https://arxiv.org/abs/2503.09220 |