Using machine learning to classify the diffuse interstellar bands

Dalya Baron, Dovi Poznanski, Darach Watson, Yushu Yao, Nick L.J. Cox, J. Xavier Prochaska

Research output: Contribution to journalArticlepeer-review


Using over a million and a half extragalactic spectra from the Sloan Digital Sky Survey we study the correlations of the diffuse interstellar bands (DIBs) in the Milky Way. We measure the correlation between DIB strength and dust extinction for 142 DIBs using 24 stacked spectra in the reddening range E(B - V) < 0.2, many more lines than ever studied before. Most of the DIBs do not correlate with dust extinction. However, we find 10 weak and barely studied DIBs with correlations that are higher than 0.7 with dust extinction and confirm the high correlation of additional five strong DIBs. Furthermore, we find a pair of DIBs, 5925.9 and 5927.5 Å, which exhibits significant negative correlation with dust extinction, indicating that their carrier may be depleted on dust. We use Machine Learning algorithms to divide the DIBs to spectroscopic families based on 250 stacked spectra. By removing the dust dependence, we study how DIBs follow their local environment. We thus obtain six groups of weak DIBs, four of which are tightly associated with C2 or CN absorption lines.

Original languageEnglish
Pages (from-to)332-352
Number of pages21
JournalMonthly Notices of the Royal Astronomical Society
Issue number1
StatePublished - 1 May 2015


  • Dust, extinction
  • ISM: general
  • ISM: lines and bands
  • ISM: molecules
  • Surveys
  • Techniques: spectroscopic


Dive into the research topics of 'Using machine learning to classify the diffuse interstellar bands'. Together they form a unique fingerprint.

Cite this