Skip to main navigation Skip to search Skip to main content

Interpreting Stellar Spectra with Unsupervised Domain Adaptation

  • Teaghan O’Briain
  • , Yuan-Sen Ting
  • , S Fabbro
  • , Kwang Moo Yi
  • , Kim Venn
  • , Spencer Bialek

    Research output: Chapter in Book/Report/Conference proceedingConference Paper

    Abstract

    We discuss how to achieve mapping from large sets of imperfect simulations and observational data with unsupervised domain adaptation. Under the hypothesis that simulated and observed data distributions share a common underlying representation, we show how it is possible to transfer between simulated and observed domains. Driven by an application to interpret stellar spectroscopic sky surveys, we construct the domain transfer pipeline from two adversarial autoencoders on each domains with a disentangling latent space, and a cycle-consistency constraint. We then construct a differentiable pipeline from physical stellar parameters to realistic observed spectra, aided by a supplementary generative surrogate physics emulator network. We further exemplify the potential of the method on the reconstructed spectra quality and to discover new spectral features associated to elemental abundances.
    Original languageEnglish
    Title of host publicationInternational Conference for Machine Learning ICML 2020
    EditorsSubhashini Venugopalan, Michael Brenner, Scott Linderman, Been Kim
    Place of PublicationVirtual
    PublisherSelf-published
    PagesSession 2, P#24
    DOIs
    Publication statusPublished - 2020
    EventML Interpretability for Scientific Discovery Workshop: International Conference for Machine Learning ICML 2020 - Online
    Duration: 1 Jan 2020 → …
    https://icml.cc/virtual/2020/workshop/5740

    Conference

    ConferenceML Interpretability for Scientific Discovery Workshop: International Conference for Machine Learning ICML 2020
    Period1/01/20 → …
    OtherJuly 17, 2020
    Internet address

    Fingerprint

    Dive into the research topics of 'Interpreting Stellar Spectra with Unsupervised Domain Adaptation'. Together they form a unique fingerprint.

    Cite this