Photometric redshift estimation of galaxies with Convolutional Neural Network

作     者：Yong-Huan Mu Bo Qiu Jian-Nan Zhang Jun-Cheng Ma Xiao-Dong Fan Yong-Huan Mu;Bo Qiu;Jian-Nan Zhang;Jun-Cheng Ma;Xiao-Dong Fan

作者机构：School of Electronics and Information EngineeringHebei University of TechnologyTianjin 300401China CAS Key Laboratory of Optical AstronomyNational Astronomical ObservatoriesChinese Academy of SciencesBeijing 100101China 

出 版 物：《Research in Astronomy and Astrophysics》 (天文和天体物理学研究（英文版）)

年 卷 期：2020年第20卷第6期

页      面：193-202页

核心收录：

学科分类：0709[理学-地质学] 07[理学] 0708[理学-地球物理学] 070401[理学-天体物理] 0704[理学-天文学] 0825[工学-航空宇航科学与技术] 

基　　金：supported by the Joint Research Fund in Astronomy(U1531242)under cooperative agreement between the National Natural Science Foundation of China(NSFC)and Chinese Academy of Sciences(CAS) 

主　　题：galaxies distances and redshifts-techniques photometric-method data analysis 

摘      要：The abundant photometric data collected from multiple large-scale sky surveys give important opportunities for photometric redshift ***,low accuracy is still a serious issue in the current photometric redshift estimation *** this paper,we propose a novel two-stage approach by integration of Self Organizing Map(SOM)and Convolutional Neural Network(CNN)methods *** SOM-CNN method is tested on the dataset of 150000 galaxies from Sloan Digital Sky Survey Data Release 13(SDSS-DR13).Inthe first stage,we apply the SOM algorithm to photometric data clustering and divide the samples into early-type and *** the second stage,the SOM-CNN model is established to estimate the photometric redshifts of ***,the precision rate and recall rate curves(PRC)are given to evaluate the models of SOM-CNN and Back Propagation(BP).It can been seen from the PRC that the SOM-CNN model is better than BP,and the area of SOM-CNN is 0.94,while the BP is ***,we provide two key error indicators:mean square error(MSE)and *** results show that the MSE of early-type is 0.0014 while late-type is 0.0019,which are better than the BP algorithm 22.2%and 26%,*** compared with Outliers,our result is optimally 1.32%,while the K-nearest neighbor(KNN)algorithm has 3.93%.In addition,we also provide the error visualization figures aboutΔZ andδ.According to the statistical calculations,the early-type with an error of less than 0.1 accounts for 98.86%,while the late-type is 99.03%.This result is better than those reported in the literature.