[Automated classification of celestial spectra based on support vector machines].

Area of Science:

• Astronomy
• Astrophysics
• Machine Learning

Background:

• Automatic classification of celestial spectra is crucial for astronomical surveys.
• Distinguishing active from non-active celestial objects is challenging due to low signal-to-noise ratio (SNR) and unknown red-shift values.
• Traditional methods struggle with accurate spectral line extraction in such conditions.

Purpose of the Study:

• To develop an automated method for classifying celestial spectra.
• To accurately differentiate between active and non-active celestial objects using spectral data.
• To address the challenges posed by low SNR and unknown red-shift values in spectral analysis.

Main Methods:

• A novel automatic classification method combining Support Vector Machines (SVM) and Principal Component Analysis (PCA).
• Utilizing SVM for spectral classification and PCA for dimensionality reduction and feature extraction.
• The method is designed to handle spectra with unknown red-shift values.

Main Results:

• The proposed method successfully separates active from non-active celestial objects based on their spectra.
• It demonstrates robustness in classifying spectra with low SNR and unknown red-shift values.
• Effective automatic recognition of active celestial object spectra is achieved.

Conclusions:

• The combined PCA-SVM approach offers an effective solution for automatic celestial spectral classification.
• This method is applicable to the automated processing of large astronomical datasets from sky surveys.
• It overcomes limitations of traditional methods in identifying active celestial objects with unknown red-shift.