Searching for hot subdwarf stars from the LAMOST spectra. III. classifying the hot subdwarf stars from LAMOST DR4 using deep learning method

TL;DR

This paper introduces a deep learning approach combining CNN and SVM to accurately classify hot subdwarf stars from LAMOST spectra, significantly aiding large-scale stellar surveys.

Contribution

The study develops and demonstrates a novel CNN+SVM method for hot subdwarf star classification, achieving high accuracy and recall on LAMOST DR4 data.

Findings

Classification accuracy of 88.98%

Recall rate of 94.38%

Provides a new machine learning tool for large spectroscopic surveys

Abstract

Hot subdwarf stars are core He burning stars located at the blue end of the horizontal branch, also known as the extreme horizontal branch. The properties of hot subdwarf stars are important for our understanding of the stellar astrophysics, globular clusters and galaxies. The spectra of hot subdwarf stars can provide us with the detailed information of the stellar atmospheric parameters (such as effective temperature, gravity, and helium abundances), which is important to clarify the astrophysical and statistical properties of hot subdwarf stars. These properties can provide important constraint on the theoretical models of hot subdwarf stars. Searching for hot subdwarf stars from the spectra data obtained by the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) can significantly enlarge the sample size of hot subdwarf stars, and help us better study the nature of hot subdwarf stars. In this paper we study a new method of searching for hot subdwarf stars from LAMOST spectra using convolutional neural networks and support vector machine (CNN+SVM). The experiment on the spectra from LAMOST DR4 shows that CNN+SVM can classify the hot subdwarf stars accurately: the accuracy is 88.98$\%$ and the recall is 94.38 $\%$. Our research provides a new machine learning tool for searching for hot subdwarf stars in large spectroscopic surveys.