Exploration and proposal of time dependent features for use in supervised machine learning and star classification in the GAIA south ecliptic pole field
Supervised learning methods bear tools such that jobs like classification and analysis become much more efficient. In general, these technological tools have a large pool of possible uses among various fields. In particular, in the field of astronomy these tools have clear uses on star classificatio...
- Autores:
-
León Figueroa, Benjamín
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2020
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/51268
- Acceso en línea:
- http://hdl.handle.net/1992/51268
- Palabra clave:
- Estrellas
Astronomía
Aprendizaje automático (Inteligencia artificial)
Física
- Rights
- openAccess
- License
- http://creativecommons.org/licenses/by-nc-sa/4.0/
| Summary: | Supervised learning methods bear tools such that jobs like classification and analysis become much more efficient. In general, these technological tools have a large pool of possible uses among various fields. In particular, in the field of astronomy these tools have clear uses on star classification. In relation to this, 15 new statistical features were proposed in an attempt to obtain information from light curves in the Large Magellanic Cloud. Moreover, random forest, k-neighbours, decision trees and support vector machines classifiers were trained with a series of widely used features and proposed features in order to classify the set of stars described by previous works. After selecting and optimizing the appropriate model, we obtain an F1 score of 0.8130 en relation to the results in the literature using a random forest classifier. About this, the classifier was trained with a default number of 1000 light curves per star class with... |
|---|