Supervised classification of variable stars using time series light-curve data
Loading...
Authors
Garlipp, Bernhardt Günter Conrad
Issue Date
2024-02-17
Type
Dissertation
Language
en
Keywords
Variable stars , Time series , Convolutional neural networks , Long short-term neural networks , Hybrid neural networks , Artificial intelligence , Machine learning , Space study and Square Kilometer Area , Fourth Industrial Revolution and Digitalisation
Alternative Title
Abstract
This research study investigates the current state-of-the-art variable star classification. We analyse current supervised classification methods for application to the problem of classifying variable star data obtained from a radio telescope. Since variable stars are a time series problem, we investigated various models to compare them with classification methods in the literature. We implemented these models and derived metrics that are directly comparable to the current literature results using a quantitative methodology. This research problem has not yet been investigated; consequently, we applied deep learning image classification to the problem. Thereafter, we compared published work with the performance of all the models and literature and inferred which models performed the best.