Advanced Search+
Xiaomeng LI (李晓萌), Huili LU (陆慧丽), Jianhong YANG (阳建宏), Fu CHANG (常福). Semi-supervised LIBS quantitative analysis method based on co-training regression model with selection of effective unlabeled samples[J]. Plasma Science and Technology, 2019, 21(3): 34015-034015. DOI: 10.1088/2058-6272/aaee14
Citation: Xiaomeng LI (李晓萌), Huili LU (陆慧丽), Jianhong YANG (阳建宏), Fu CHANG (常福). Semi-supervised LIBS quantitative analysis method based on co-training regression model with selection of effective unlabeled samples[J]. Plasma Science and Technology, 2019, 21(3): 34015-034015. DOI: 10.1088/2058-6272/aaee14

Semi-supervised LIBS quantitative analysis method based on co-training regression model with selection of effective unlabeled samples

Funds: This work was supported by National Natural Science Foundation of China (No. 51674032).
More Information
  • Received Date: July 27, 2018
  • The accuracy of laser-induced breakdown spectroscopy (LIBS) quantitative method is greatly dependent on the amount of certified standard samples used for training. However, in practical applications, only limited standard samples with labeled certified concentrations are available. A novel semi-supervised LIBS quantitative analysis method is proposed, based on co-training regression model with selection of effective unlabeled samples. The main idea of the proposed method is to obtain better regression performance by adding effective unlabeled samples in semi- supervised learning. First, effective unlabeled samples are selected according to the testing samples by Euclidean metric. Two original regression models based on least squares support vector machine with different parameters are trained by the labeled samples separately, and then the effective unlabeled samples predicted by the two models are used to enlarge the training dataset based on labeling confidence estimation. The final predictions of the proposed method on the testing samples will be determined by weighted combinations of the predictions of two updated regression models. Chromium concentration analysis experiments of 23 certified standard high-alloy steel samples were carried out, in which 5 samples with labeled concentrations and 11 unlabeled samples were used to train the regression models and the remaining 7 samples were used for testing. With the numbers of effective unlabeled samples increasing, the root mean square error of the proposed method went down from 1.80% to 0.84% and the relative prediction error was reduced from 9.15% to 4.04%.
  • [1]
    Li X W et al 2015 Plasma Sci. Technol. 17 621
    [2]
    Guo Y M et al 2018 Plasma Sci. Technol. 20 065505
    [3]
    Yu J L et al 2018 Anal. Methods 10 281
    [4]
    Xie S C et al 2018 J. Anal. At. Spectrom. 33 975
    [5]
    Tognoni E et al 2002 Spectrochim. Acta B 57 1115
    [6]
    Labutin T A et al 2013 Spectrochim. Acta B 87 57
    [7]
    Li L Z et al 2011 J. Anal. At. Spectrom. 26 2274
    [8]
    Wang Z et al 2012 Spectrochim. Acta B 68 58
    [9]
    Zaytsev S M et al 2018 Spectrochim. Acta B 140 65
    [10]
    Zhang T L et al 2014 J. Anal. At. Spectrom. 29 2323
    [11]
    Dingari N C et al 2012 Anal. Chem. 84 2686
    [12]
    Wang Z et al 2011 J. Anal. At. Spectrom. 26 2175
    [13]
    Tsuda K and R?tsch G 2005 IEEE Trans. Image Process. 14 737
    [14]
    Feng W et al 2009 J. Vis. Lang. Comput. 20 188
    [15]
    Yu Z T et al 2010 Pattern Recognit. Lett. 31 1975
    [16]
    Yuan H J, Wang C R and Liu J 2011 Adv. Mater. Res. 267 1065
    [17]
    Zhou Z H and Li M 2005 Semi-supervised regression with co- training Proc. 19th Int. Joint Conf. on Artificial Intelligence (Edinburgh) (ACM) p 908
    [18]
    Shahshahani B M and Landgrebe D A 1994 IEEE Trans. Geosci. Remote Sens. 32 1087
    [19]
    Baluja S 1998 Probabilistic modeling for face orientation discrimination: learning from labeled and unlabeled data Proc. 11th Int. Conf. on Neural Information Processing Systems 11 (Denver) (ACM) p 854
    [20]
    Belkin M and Niyogi P 2004 Mach. Learn. 56 209
    [21]
    Niyogi P 2013 J. Mach. Learn. Res. 14 1229
    [22]
    Zhou Z H and Li M 2005 IEEE Trans. Knowl. Data Eng. 17 1529
    [23]
    Blum A and Mitchell T 1998 Combining labeled and unlabeled data with co-training Proc. 11th Annual Conf. on Computational Learning Theory (Madison, WI) (ACM) p 92
    [24]
    Suykens J A K and Vandewalle J 1999 Neural Process. Lett. 9 293
    [25]
    Pontes M J C et al 2005 Chemometr. Intell. Lab. Syst. 78 11
    [26]
    Poli R, Kennedy J and Blackwell T 2007 Swarm Intell. 1 33
  • Related Articles

    [1]Haochen FAN, Guoqiang LI, Jinping QIAN, Xuexi ZHANG, Xiaohe WU, Yuqi CHU, Xiang ZHU, Hui LIAN, Haiqing LIU, Bo LYU, Yifei JIN, Qing ZANG, Jia HUANG. Kinetic equilibrium reconstruction with internal safety factor profile constraints on EAST tokamak[J]. Plasma Science and Technology, 2024, 26(4): 045102. DOI: 10.1088/2058-6272/ad0d48
    [2]Hailong GAO (高海龙), Tao XU (徐涛), Zhongyong CHEN (陈忠勇), Ge ZHUANG (庄革). Plasma equilibrium calculation in J-TEXT tokamak[J]. Plasma Science and Technology, 2017, 19(11): 115101. DOI: 10.1088/2058-6272/aa7f26
    [3]Linghan WAN (万凌寒), Zhoujun YANG (杨州军), Ruobing ZHOU (周若冰), Xiaoming PAN (潘晓明), Chi ZHANG (张弛), Xianli XIE (谢先立), Bowen RUAN (阮博文). Design of Q-band FMCW reflectometry for electron density profile measurement on the Joint TEXT tokamak[J]. Plasma Science and Technology, 2017, 19(2): 25602-025602. DOI: 10.1088/2058-6272/19/2/025602
    [4]GAO Yu (高宇), WANG Yumin (王嵎民), ZHANG Tao (张涛), ZHANG Shoubiao (张寿彪), QU Hao (屈浩), HAN Xiang (韩翔), WEN Fei (文斐), KONG Defeng (孔德峰), HUANG Canbin (黄灿斌), CAI Jianqing (蔡剑青), SUN Youwen (孙有文), LIANG Yunfeng (梁云峰), GAO Xiang (高翔), EAST Team. Preliminary Study of the Magnetic Perturbation Effects on the Edge Density Profiles and Fluctuations Using Reflectometers on EAST[J]. Plasma Science and Technology, 2016, 18(9): 879-883. DOI: 10.1088/1009-0630/18/9/01
    [5]KE Xin (柯新), CHEN Zhipeng (陈志鹏), BA Weigang (巴为刚), SHU Shuangbao (舒双宝), GAO Li (高丽), ZHANG Ming (张明), ZHUANG Ge (庄革). The Construction of Plasma Density Feedback Control System on J-TEXT Tokamak[J]. Plasma Science and Technology, 2016, 18(2): 211-216. DOI: 10.1088/1009-0630/18/2/20
    [6]QU Hao (屈浩), ZHANG Tao (张涛), ZHANG Shoubiao (张寿彪), WEN Fei (文斐), WANG Yumin (王嵎民), KONG Defeng (孔德峰), HAN Xiang (韩翔), YANG Yao (杨曜), GAO Yu (高宇), HUANG Canbin (黄灿斌), CAI Jianqing (蔡剑青), GAO Xiang (高翔), the EAST team. Q-Band X-Mode Reflectometry and Density Profile Reconstruction[J]. Plasma Science and Technology, 2015, 17(12): 985-990. DOI: 10.1088/1009-0630/17/12/01
    [7]LIU Chao(刘超), LIU Yue(刘悦), MA Zhaoshuai(马照帅). Effect of Equilibrium Current Profiles on External Kink Modes in Tokamaks[J]. Plasma Science and Technology, 2014, 16(8): 726-731. DOI: 10.1088/1009-0630/16/8/02
    [8]ZHANG Chongyang (张重阳), LIU Ahdi (刘阿娣), LI Hong (李弘), LI Bin (李斌), et al.. X-Mode Frequency Modulated Density Profile Reflectometer on EAST Tokamak[J]. Plasma Science and Technology, 2013, 15(9): 857-862. DOI: 10.1088/1009-0630/15/9/04
    [9]XU Chao (许超), OU Yongsheng (欧勇盛), Eugenio SCHUSTER, and YU Xin(于欣). Computing Open-Loop Optimal Control of the q-Profile in Ramp-Up Tokamak Plasmas Using the Minimal-Surface Theory[J]. Plasma Science and Technology, 2013, 15(5): 403-410. DOI: 10.1088/1009-0630/15/5/02
    [10]LI Li(李莉), LIU Yue (刘悦), XU Xinyang(许欣洋), XIA Xinnian(夏新念). The Effect of Equilibrium Current Profiles on MHD Instabilities in Tokamaks[J]. Plasma Science and Technology, 2012, 14(1): 14-19. DOI: 10.1088/1009-0630/14/1/04

Catalog

    Article views (162) PDF downloads (260) Cited by()

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return