Automated selection of CVs from massive SDSS photometric data

We apply Support Vector Machine (SVM) to construct the photometric models for cataclysmic variable stars (CVs) and identify CVs in massive photometric data of Sloan Digital Sky Survey (SDSS). Based on the study of spectral characteristics and the color information, orthogonal criterion combined with SVM is used for modeling the photometric data. We present 10 groups of representative photometric models constrained by SVM, which can be used as features for classification and also for photometric criterion. In the experiment, we develop and demonstrate the automatic selection method for CVs. We have also made hybrid verification through logistic regression and artificial neural network to verify the experiment results. The photometric model improve the identification accuracy of CVs in massive SDSS spectra data. The result shows that the photometric model can improve the classify accuracy on the research of five color feathers in CVs classification tasks.