1. Stellar Spectrum Analysis for Automated Estimation of Atmospheric Parameter 李乡儒 2015. 11.28 Collaborators ： Ali Luo, Yongheng Zhao, Georges Comte, Fang Zuo, Q.M. Jonathan Wu, Tan Yang, Yongjun Wang, Yu Lu

2. Contents Xiangru Li, Q.M. Jonathan Wu, Ali Luo, Yongheng Zhao, Yu Lu, Fang Zuo, Tan Yang, Yongjun Wang, 2014, ApJ, 790, 105 X. Li, Yu Lu, G. Comte, Ali Luo, Yongheng Zhao, Yongjun Wang, 2015, ApJS, 218,3 Yu Lu, X. Li, 2015, MNRAS, 452(2): 1394 Tan Yang, X. Li, 2015, MNRAS, 2015, 452, 158 Problem, Available Schemes and Objective Sparse Feature Extraction Linearly Supporting Features Extraction Adaptive Basic Structure Elements and Spectral Feature Extraction

3. Problem

4. Available Schemes and Objective Template Matching Method Statistical Index Scheme Line Index Method Physical Interpretability Robustness local, sparse

5. Problem and Objective Detection Description Estimation

6. SDSS Data 50000, [4088, 9740]K for Teff, [1.015, 4.998] dex for log g, [-3.497 0.268]dex for [Fe/H]

7. LAMOST Data 33963 [3853.2, 9927] K for Teff, [0.8920, 4.9959] dex for log g, [-2.3280 0.9360] dex for [Fe/H]

8. Synthetic Data Kurucz’s NEWODF models, SPECTRUM package 18969 [4000, 9750] K for Teff, 45 values, step sizes of 100 K between 4000 and 7500, 205 K between 7750 and 9750 K [1, 5] dex for log g, 17 values, step size of 0.25 dex [-3.6 0.3]dex for [Fe/H], 27 values, step size of 0.2 dex between -3.6 and -1 dex, and 0.1 dex between -1 and 0.3 dex

9. Sparse Feature Extraction Xiangru Li, Q.M. Jonathan Wu, Ali Luo, Yongheng Zhao, Yu Lu, Fang Zuo, Tan Yang, Yongjun Wang, 2014, ApJ, 790, 105

10. Detection LASSO (least absolute shrinkage and selection operator)

11. Detection

13. Re Fiorentin, P., et al. 2007, A&A, 467, 1373 99.74%

14. Detection

20. Description and Estimation Point Description (PD) Local Integration (LI)

21. Experimental Results On Real Spectra Re Fiorentin, P., et al. 2007, A&A, 467, 1373

22. Experimental Results On Synthetic Spectra

23. Compactness 99.74% On Real Spectra Re Fiorentin, P., et al. 2007, A&A, 467, 1373

24. Linearity v.s. nonlinearity

25. Other typical non-linear estimators Feedforward neural network Generalized Additive Models Multivariate Adaptive Regression Splines Random Forest

26. Linearly Supporting Features Extraction X. Li, Yu Lu, G. Comte, Ali Luo, Yongheng Zhao, Yongjun Wang, 2015, ApJS, 218,3 Yu Lu, X. Li, 2015, MNRAS, 452(2): 1394

27. Linearly Supporting Feature Extraction

28. Dissolution of nonlinearity Dependeny of effectiveness on wavelength and frequency

36. For log g, [Fe/H] …

42. Adaptive Basic Structure Elements and Spectral Feature Extraction Tan Yang, X. Li, 2015, MNRAS, 2015, 452, 158

48. 谢谢大家！