English

Debunking Generalization Error or: How I Learned to Stop Worrying and Love My Training Set

Instrumentation and Methods for Astrophysics 2020-12-02 v1 Astrophysics of Galaxies

Abstract

We aim to determine some physical properties of distant galaxies (for example, stellar mass, star formation history, or chemical enrichment history) from their observed spectra, using supervised machine learning methods. We know that different astrophysical processes leave their imprint in various regions of the spectra with characteristic signatures. Unfortunately, identifying a training set for this problem is very hard, because labels are not readily available - we have no way of knowing the true history of how galaxies have formed. One possible approach to this problem is to train machine learning models on state-of-the-art cosmological simulations. However, when algorithms are trained on the simulations, it is unclear how well they will perform once applied to real data. In this paper, we attempt to model the generalization error as a function of an appropriate measure of distance between the source domain and the application domain. Our goal is to obtain a reliable estimate of how a model trained on simulations might behave on data.

Keywords

Cite

@article{arxiv.2012.00066,
  title  = {Debunking Generalization Error or: How I Learned to Stop Worrying and Love My Training Set},
  author = {Viviana Acquaviva and Chistopher Lovell and Emille Ishida},
  journal= {arXiv preprint arXiv:2012.00066},
  year   = {2020}
}

Comments

Accepted for 2020 NeurIPS workshop "Machine Learning and the Physical Sciences"; comments welcome!

R2 v1 2026-06-23T20:37:05.669Z