精準測量並統計年輕恆星物體在不同階段的恆星形成速率(SFR)與壽命是規範恆星形成理論的必要條件之一。然而,使用能量光譜分佈(SED)來分辨年輕恆星物體(YSO)與星系是困難的工作,目前還沒有可靠的理論可以用來分辨他們因為他們都包含來自恆星與周圍灰塵的熱輻射。我們基於卷積神經網路建造了天文物體光譜判斷器(SCAO),利用SED來判斷恆星、星系與YSO。SCAO僅利用八個波段的SED就可以給出高精準(>95%)與高回復率(>98%)的結果。我們將SCAO應用在史匹哲強化影像產品(SEIP)目錄上,包含了最完整的史匹哲紅外線觀測資料,並且找到了136680個候選YSO。SCAO的網站公開在以下網址:http://scao.astr.nthu.edu.tw。我們使用SCAO所辨認的YSO,探討銀河系中恆星形成區域的SFR與氣體面密度之間的關係(SFR-氣體關係)。我們使用普朗克全天灰塵面密度圖選出六十二個區域並計算他們的SFR、氣體面密度與恆星形成效率(SFE)。由於YSO樣本的不完整度,我們發現SFR與氣體面密度不是線性關係。因為SFE在500秒差距內大約是常數(1.59%±3.45%)但會隨著距離增加而快速減少,所以我們猜測YSO的不完整度在500秒差距內是個常數。另外,我們探討SFR-氣體關係隨著消光程度的變化,而且只計入第一類與平坦類YSO因為他們尚未離開他們的出生區域。SFR-氣體關係可以用冪律擬合,從全部的樣本得出其指數N =1.36±0.15、縮小卡方18.05,從500秒差距內的樣本中得出其冪律指數N =1.73±0.19,縮小卡方9.96。先前的研究顯示用雙斷冪律可以更好地用不同的斜率表示SFR-氣體關係在重力約束及非約束的區域。單一冪律與雙斷冪律兩者都可以很好的擬合我們的SFR-氣體關係資料,因為兩者的縮小卡方是接近的。

Accurate measurements of statistical properties, such as the star formation rate (SFR) and the lifetime of young stellar objects (YSOs) in different stages, is essential for constraining star formation theories. However, it is a difficult task to separate galaxies and YSOs based on spectral energy distributions (SEDs) alone, because they contain both thermal emission from stars and dust around them and no reliable theories can be applied to distinguish them. Here we construct a machine learning model based on Convolutional Neural Network, named Spectrum Classifier of Astronomical Objects (SCAO), to classify regular stars, galaxies, and YSOs, solely based on their SEDs. It provides excellent results with high precision (>95%) and recall (>98%) for YSOs when data from only eight bands are included. We apply SCAO to Spitzer Enhanced Imaging Products (SEIP), the most complete catalog of Spitzer observations, and found 136689 YSO candidates. The website from SCAO is available at http://scao.astr.nthu.edu.tw. Using the YSOs identified with SCAO, we investigate the relation between SFR and gas surface density (SFR-gas relation) in star forming regions of the Milky way. We select 62 regions and calculate their SFR, gas surface density, and star formation efficiency (SFE) with the Planck all-sky dust surface density map. We found SFR is not linearly correlated to gas surface density possibly due to the incompleteness of YSO samples. Because SFE is roughly constant for clouds at less than 500pc (1.59±3.45%) but decreases rapidly with increasing distance, we speculate that the completeness of YSOs is constant within 500pc. In addition, we further probe the variation of SFR-gas relation as a function of extinction, and count only Class I and Flat YSOs, which did not move away from their birth place yet. The SFR-gas relation as a function of extinction can be fitted by a power law with an index N =1.36±0.15 with reduced chi square χ 2 r = 18.05 for entire samples and N =1.73±0.19 with χ 2 r =9.96 for samples within 500pc. Our result is consistent with Kennicutt-Schmidt law (N =1.2–1.6). Previous studies have shown that the SFR-gas relation is better presented with a broken power law with different slope for gravitationally bound and unbound regions. We found both the single power law and the broken power law are good models for the SFR-gas relation with comparable reduced chi square.

1 Introduction 1
1.1 Machine Learning 1
1.2 SFR-gas Relation 2
2 Searching for Young Stellar Objects through SEDs by Convolutional Neural Networks 4
2.1 Introduction 5
2.2 Data 7
2.2.1 Sources of Data 8
2.2.2 Actual Labels for our Model 9
2.2.3 Data Preparation 10
2.2.4 Typical SED 11
2.2.5 Flux-Error Correlations 12
2.3 Method 13
2.4 Results 15
2.4.1 Model I: Classification using SED 17
2.4.2 Model II: Classification using SED with Error 19
2.4.3 Model III: Classification using Normalized SED 20
2.4.4 Model IV: Classification using Partial SED 20
2.5 Discussion 22
2.5.1 Discrepancies of Predicted Results by SCAO and by c2d Method 22
2.5.2 Extinction Effects 23
2.5.3 Validation of SCAO 25
2.5.4 Comparison of the Classification Results with Marton et al.(2019) 26
2.5.5 Comparison of the Performance with Other Machine Learning Algorithms 28
2.5.6 Predicted Distribution Regime in the Flux Space 29
2.6 Conclusion 29
2.7 Appendix: SCAO YSO Candidate List from Spitzer Enhanced Imaging Products 30
2.8 Appendix: Analyzing Discrepancies of SCAO Predictions and Previous Theories 32
2.8.1 Characteristics 32
2.8.2 Identifiers 33
2.8.3 Human Judgements 33
2.8.4 Confidence Levels 34
3 Star Formation Rate and Efficiency in the Milky Way 41
3.1 Introduction 41
3.2 Data and the Definitions of Physical Quantities 44
3.3 Result 46
3.3.1 Systematic Bias 47
3.3.2 SFR-gas Relation 49
3.3.3 SFR-gas Relation as a Function of Extinction 49
3.3.4 Star Formation Efficiency 52
3.4 Discussion 52
3.5 Conclusions 54
3.6 Appendix: Samples of Star Forming Regions 55
4 Future 69

See for example, Artificial Intelligence: A Modern Approach, by Stuart Russell and Peter Norvig (Third Edition, Pearson Education 2014).
LeCun, Y., Bengio, Y., and Hinton, G. 2015, Nature 521, 436–444
de Villiers, H. 2013, Protostars and Planets VI Posters
Marton, G., Tóth, L. V., Paladini, R., et al. 2016, MNRAS, 458, 3479
Marton, G., Ábrahám, P., Szegedi-Elek, E., et al. 2019, MNRAS, 487, 2522
Miettinen, O. 2018, Ap&SS, 363, 197
Akras, S., Leal-Ferreira, M. L., Guzman-Ramirez, L., et al. 2019, MNRAS, 483, 5077
Hedges, C., Hodgkin, S., & Kennedy, G. 2018, MNRAS, 476, 2968
Gutermuth, R. A., Megeath, S. T., Pipher, J. L., et al. 2005, ApJ, 632, 397
Rebull, L. M., Padgett, D. L., McCabe, C.-E., et al. 2010, ApJS, 186, 259
Evans, N. J., II, Allen, L. E., Blake, G. A., et al. 2003, PASP, 115, 965
Evans, N. J., II, Harvey, P. M., Dunham, M. M., et al. 2007, Final Delivery of Data from the c2d Legacy Project: IRAC and MIPS (Pasadena, CA: SSC)
Evans, N. J., II, Dunham, M. M., Jørgensen, J. K., et al. 2009, ApJS, 181, 321
Harvey, P., Merı́n, B., Huard, T. L., et al. 2007, ApJ, 663, 1149
Heiderman, A., Evans, N. J., Allen, L. E., et al. 2010, ApJ, 723, 1019
Lonsdale, C. J., Smith, H. E., Rowan-Robinson, M., et al. 2003, PASP, 115, 897
Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163
Lawrence, A., Warren, S. J., Almaini, O., et al. 2007, MNRAS, 379, 1599
Hewett, P. C., Warren, S. J., Leggett, S. K., & Hodgkin, S. T. 2006, MNRAS, 367, 454
Casali, M., Adamson, A., Alves de Oliveira, C., et al. 2007, A&A, 467, 777
Hodgkin, S. T., Irwin, M. J., Hewett, P. C., & Warren, S. J. 2009, MNRAS, 394, 675
Hambly, N. C., Collins, R. S., Cross, N. J. G., et al. 2008, MNRAS, 384, 637.
Werner, M. W., Roellig, T. L., Low, F. J., et al. 2004, ApJS, 154, 1
Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, A&A, 616, A1
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2016, A&A, 586, A132
Planck Collaboration, Abergel, A., Ade, P. A. R., et al. 2014, A&A, 571, A11
Zucker, C., Speagle, J. S., Schlafly, E. F., et al. 2020, A&A, 633, A51
Koenig, X. P., & Leisawitz, D. T. 2014, ApJ, 791, 131
Draine, B. T., Dale, D. A., Bendo, G., et al. 2007, ApJ, 663, 866
Liszt, H. 2014, ApJ, 783, 17
Liszt, H. 2014, ApJ, 780, 10
Lombardi, M., & Alves, J. 2001, A&A, 377, 1023.
Meingast, S., Lombardi, M., & Alves, J. 2017, A&A, 601, A137.
Schmidt, M. 1959, ApJ, 129, 243
Kennicutt, R. C. 1989, ApJ, 344, 685
Kennicutt, R. C. 1998, ApJ, 498, 541
Martin, C. L., & Kennicutt, R. C. 2001, ApJ, 555, 301
Gao, Y., & Solomon, P. M. 2004, ApJ, 606, 271
Wu, J., Evans, N. J., Gao, Y., et al. 2005, ApJ, 635, L173
Bigiel, F., Leroy, A., Walter, F., et al. 2008, AJ, 136, 2846
Bigiel, F., Leroy, A., Walter, F., et al. 2010, AJ, 140, 1194
Bigiel, F., Leroy, A. K., Walter, F., et al. 2011, ApJ, 730, L13
Leroy, A. K., Walter, F., Brinks, E., et al. 2008, AJ, 136, 2782
Schruba, A., Leroy, A. K., Walter, F., et al. 2011, AJ, 142, 37
Sánchez, S. F. 2020, ARA&A, 58, annurev
Krumholz, M. R., & Thompson, T. A. 2007, ApJ, 669, 289
Urban, A., Martel, H., & Evans, N. J. 2010, ApJ, 710, 1343
Gutermuth, R. A., Pipher, J. L., Megeath, S. T., et al. 2011, ApJ, 739, 84
Lada, C. J., Lombardi, M., & Alves, J. F. 2010, ApJ, 724, 687
Wu, J., Evans, N. J., Shirley, Y. L., et al. 2010, ApJS, 188, 313
Murray, N. 2011, ApJ, 729, 133
Wenger, M., Ochsenbein, F., Egret, D., et al. 2000, Astronomy and Astrophysics Supplement Series, 143, 9.
Greene, T. P., Wilking, B. A., Andre, P., et al. 1994, ApJ, 434, 614
Draine, B. T. 2003, ARA&A, 41, 241
McKee, C. F., & Ostriker, E. C. 2007, ARA&A, 45, 565.
Bergin, E. A., & Tafalla, M. 2007, ARA&A, 45, 339.
Hsieh, T.-H., & Lai, S.-P. 2013, ApJS, 205, 5.
Mouschovias, T. C., & Spitzer, L. 1976, ApJ, 210, 326
McKee, C. F. 1989, ApJ, 345, 782
Kennicutt, R. C., & Evans, N. J. 2012, ARA&A, 50, 531
An, D., Ramı́rez, S. V., Sellgren, K., et al. 2011, ApJ, 736, 133
Furlan, E., Fischer, W. J., Ali, B., et al. 2016, ApJS, 224, 5
Reipurth, B. 2008, Handbook of Star Forming Regions, Volume I
Reipurth, B. 2008, Handbook of Star Forming Regions, Volume II
Chabrier, G. 2003, PASP, 115, 763
Kroupa, P. 2002, Science, 295, 82
Ninkovic, S., & Trajkovska, V. 2006, Serbian Astronomical Journal, 172, 17
Kennicutt, R. C., Armus, L., Bendo, G., et al. 2003, PASP, 115, 928
Walter, F., Brinks, E., de Blok, W. J. G., et al. 2008, AJ, 136, 2563
Kennicutt, R. C., Calzetti, D., Aniano, G., et al. 2011, PASP, 123, 1347
Galbany, L., Anderson, J. P., Rosales-Ortega, F. F., et al. 2016, MNRAS, 455, 4087
Sánchez, S. F., Kennicutt, R. C., Gil de Paz, A., et al. 2012, A&A, 538, A8
Irwin et al., 2008, in preparation
Chiu, Y.-L., Wang, D.-W., Ho, C.-T., et al., 2020, submitted
Chiu, Y.-L., & Lai, S.-P., 2020, in preparation