A new automated spectral feature extraction method and its application in spectral classification and defective spectra recovery. Issue 4 (26th December 2016)

• Record Type: Journal Article
• Authors: Wang, Ke; Guo, Ping; Luo, A-Li
• Is Part Of: Monthly notices of the Royal Astronomical Society. Volume 465:Issue 4(2017)
• Page Start: 4311
• Page End: 4324
• Publication Date: 2016-12-26
• Subjects: methods: data analysis -- methods: numerical -- methods: statistical -- techniques: spectroscopic; Astronomy -- Periodicals; Periodicals; 520.5
• DOI: 10.1093/mnras/stw2894
• Languages: English
• ISSNs: 0035-8711

Abstract

Abstract: Spectral feature extraction is a crucial procedure in automated spectral analysis. This procedure starts from the spectral data and produces informative and non-redundant features, facilitating the subsequent automated processing and analysis with machine-learning and data-mining techniques. In this paper, we present a new automated feature extraction method for astronomical spectra, with application in spectral classification and defective spectra recovery. The basic idea of our approach is to train a deep neural network to extract features of spectra with different levels of abstraction in different layers. The deep neural network is trained with a fast layer-wise learning algorithm in an analytical way without any iterative optimization procedure. We evaluate the performance of the proposed scheme on real-world spectral data. The results demonstrate that our method is superior regarding its comprehensive performance, and the computational cost is significantly lower than that for other methods. The proposed method can be regarded as a new valid alternative general-purpose feature extraction method for various tasks in spectral data analysis.