|
|
Paper: |
Effect of the Signal to Noise Ratio on the Accuracy of the Automatic Spectral Classification of Stellar Spectra |
Volume: |
521, Astronomical Data Analysis Software and Systems XXVI |
Page: |
351 |
Authors: |
Corral, L.; Jiménez, S. G. N.; Villavicencio, E. |
Abstract: |
The signal to noise ratio (S/N) of spectra obtained in survey missions are not always ideal for spectral classification;
the actual surveys are acquiring thousands or even millions of spectra, like the GAIA, LAMOST and other surveys.
This fact motivated the development of methods to automatize the spectral classification. Some AI technics, like artificial neural networks are highly robust against noise and lost data.
The signal to noise ratio is an important parameter that greatly affects the accuracy of the automatic spectral classification. We present here the analysis made over the automatic spectral classification of stellar spectra with different levels of signal to noise ratio (S/N). We trained specialized neural networks with spectra at different S/N levels in order to minimize such effect and present here the quantitative analysis of the accuracy in the spectral classification. |
|
|
|
|