Back to Volume
Paper: Application of Artificial Neural Networks for the Automatic Stellar Spectral Classification
Volume: 522, Astronomical Data Analysis Software and Systems XXVII
Page: 401
Authors: Villavicencio-Arcadia, E.; Navarro, S. G.; Corral, L. J.
Abstract: In this work, we present the application of Artificial Neural Networks (ANNs) for the automatic stellar spectral classification process. For the experiment were used two spectral datasets, the first one contained 4722 spectra with signal to noise ratio (S/N) as low as 15, and the second dataset from the First Data Release (DR1) of LAMOST, this second dataset consisted of 50731 spectra with S/N above 20. All spectra in the two datasets were normalized and for each spectrum were measured 36 spectral indices. The ANNs were trained and tested several times with the first dataset, and were used to provide a new classification for the LAMOST spectra. Results showed that Artificial Neural Networks are a powerful technique which could provide a reliable and accurate spectral classification over the existent and upcoming astronomical data.
Back to Volume