Prediction of the Atmospheric Fundamental Parameters from Stellar Spectra Using Artificial Neural Network

TL;DR

This paper develops an Artificial Neural Network algorithm to automatically classify stellar spectra and predict fundamental atmospheric parameters, demonstrating good accuracy on SDSS and observatory data.

Contribution

The paper introduces a novel ANN-based method for automated stellar spectral classification and parameter estimation, applicable to large spectral databases.

Findings

ANN predictions agree with traditional methods for about 50% of samples

The algorithm effectively classifies hot white dwarf and B-type spectra

Discrepancies are analyzed and discussed

Abstract

Innovation in the ground and space-based instruments has taken us into a new age of spectroscopy, in which a large amount of stellar content is becoming available. So, automatic classification of stellar spectra became subjective in recent years due to the availability of large observed spectral database as well as the theoretical spectra. In the present paper, we develop an Artificial Neural Network (ANN) algorithm for automated classification of stellar spectra. The algorithm has been applied to extract the fundamental parameters of some hot helium rich white dwarf stars observed by the Sloan Digital Sky Survey (SDSS) and B-type spectra observed at Onderjove observatory. We compared the present fundamental parameters and those from a minimum distance method to clarify the accuracy of the present algorithm where we found that, the predicted atmospheric parameters for the two samples are in good agreement for about 50% of the samples. A possible explanation for the discrepancies found for the rest of the samples is discussed.