标题:Automatic classification of massive LAMOST spectra with RBF Neural Network
作者:Jiang Bin; Wang Wenyu; Qu Meixia; Wang Wei; Gao Jun
通讯作者:Gao, J
作者机构:[Jiang Bin; Wang Wenyu; Qu Meixia; Wang Wei; Gao Jun] Shandong Univ Weihai, Sch Mech Elect & Informat Engn, Weihai, Peoples R China.
会议名称:3rd International Conference on Information Science and Control Engineering (ICISCE)
会议日期:JUL 08-10, 2016
来源:2016 3RD INTERNATIONAL CONFERENCE ON INFORMATION SCIENCE AND CONTROL ENGINEERING (ICISCE)
出版年:2016
页码:244-247
DOI:10.1109/ICISCE.2016.61
关键词:Data mining; Spectra; RBF Neural Network
摘要:Cataclysmic variable stars (CVs) are used to study evolutionary theories especially models for close binary stars. The further research of CVs is constrained by the number of samples especially spectroscopic samples since CVs is a kind of rare and special celestial objects. LAMOST-DR2 is the second data release of Guoshoujing telescope, which provides an unprecedented opportunity to search for CVs. In this paper, we proposed a RBF neural network approach to search for CVs in massive LAMOST data based on spectral features. The hyper-parameters of the network are determined in a practical and theoretical way to improve the accuracy of the model. Totally 436 CVs candidates are selected by applying the classification method to about four million LAMOST spectra. Seven CVs spectra are verified as new discoveries including four dwarf nova and three nova like variables. The newly found CVs with a catalog of spectra, positions and colors is presented which enriches the current spectra library. Experiment results show that RBF neural network method is well suited to deal with massive spectra from large survey telescope data.
收录类别:CPCI-S
WOS核心被引频次:1
资源类型:会议论文
TOP