[Searching for dwarf nova candidates with automatic methods in massive spectra].

Area of Science:

• Astronomy and Astrophysics
• Machine Learning in Science

Context:

• Analyzing vast astronomical datasets, such as spectra from the Sloan Digital Sky Survey (SDSS-DR9), presents significant challenges for identifying rare celestial objects.
• Dwarf novae are crucial targets for understanding stellar evolution and accretion processes.

Purpose:

• To develop and validate an automated, efficient method for discovering dwarf nova candidates within large spectral datasets.
• To leverage machine learning techniques, specifically Principal Component Analysis (PCA) and Support Vector Machine (SVM), for spectral classification.

Summary:

• An automated method combining PCA for dimensionality reduction and SVM for classification was applied to SDSS-DR9 spectra to search for dwarf nova candidates.
• The optimal feature space dimensions were determined using SVM identification accuracy on training samples.
• The method successfully identified 276 dwarf nova candidates, including 6 newly discovered ones, demonstrating its feasibility for finding special celestial bodies in massive spectral data.

Impact:

• The 6 new dwarf nova discoveries contribute to an improved dwarf nova template library, aiding in the construction of more accurate feature spaces.
• The proposed method is adaptable for searching for other special celestial objects in data from telescopes like LAMOST (Large Sky Area Multi-Object Fiber Spectroscopic Telescope).