Exploring the spectroscopic diversity of Type Ia supernovae with dracula: a machine learning approach

Sasdelli, Michele; Ishida, Emille E. O.; Vilalta, Ricardo; Aguena, M.; Busti, V. C.; Chavez, Hugo Orlando Camacho; Trindade, A. M. M.; Gieseke, Fabian; Souza, Rafael S. de; Fantaye, Y.; Mazzali, P. A.

doi:10.1093/mnras/stw1228

Cited by 35 publications

(39 citation statements)

References 97 publications

(115 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A recommended approach is to use kernel principal component analysis (Kernel PCA) [14]; the technique is a generalization of principal component analysis (PCA) by using non-linear components. Recently, new approaches have emerged that compute robust non-linear combinations of features, such as those found in deep-learning architectures [5,13].…”

Section: An Application Of Transfer Learning In Astronomymentioning

confidence: 99%

Transfer Learning in Astronomy: A New Machine-Learning Paradigm

Vilalta

2018

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

The widespread dissemination of machine learning tools in science, particularly in astronomy, has revealed the limitation of working with simple single-task scenarios in which any task in need of a predictive model is looked in isolation, and ignores the existence of other similar tasks. In contrast, a new generation of techniques is emerging where predictive models can take advantage of previous experience to leverage information from similar tasks. The new emerging area is referred to as transfer learning. In this paper, I briefly describe the motivation behind the use of transfer learning techniques, and explain how such techniques can be used to solve popular problems in astronomy. As an example, a prevalent problem in astronomy is to estimate the class of an object (e.g., Supernova Ia) using a generation of photometric lightcurve datasets where data abounds, but class labels are scarce; such analysis can benefit from spectroscopic data where class labels are known with high confidence, but the data sample is small. Transfer learning provides a robust and practical solution to leverage information from one domain to improve the accuracy of a model built on a different domain. In the example above, transfer learning would look to overcome the difficulty in the compatibility of models between spectroscopic data and photometric data, since data properties such as size, class priors, and underlying distributions, are all expected to be significantly different.

show abstract

Section: An Application Of Transfer Learning In Astronomymentioning

confidence: 99%

Transfer Learning in Astronomy: A New Machine-Learning Paradigm

Vilalta

2018

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

show abstract

“…We compiled a set of publicly available SN Ia spectra from a variety of sources (see Sasdelli et al 2016 for a complete reference list). These were smoothed with a Savitzky-Golay filter and derived as described in Sasdelli et al 2015.…”

Section: Datamentioning

confidence: 99%

“…Our goal is to provide a proof of concept, showing that the algorithm is able to leverage the same set of spectral features one would choose through visual recognition. This work was originally reported in Sasdelli et al 2016 and the tools used here are implemented in the DRACULA Python package (Dimensionality Reduction And Clustering for Unsupervised Learning in Astronomy) and are publicly available within the COINtoolbox †.…”

Section: Introductionmentioning

confidence: 99%

Exploring the spectroscopic diversity of type Ia supernovae with Deep Learning and Unsupervised Clustering

et al. 2016

View full text Add to dashboard Cite

The existence of multiple subclasses of type Ia supernovae (SNeIa) has been the subject of great debate in the last decade. In this work, we show how machine learning tools facilitate identification of subtypes of SNe Ia. Using Deep Learning for dimensionality reduction, we were capable of performing such identification in a parameter space of significantly lower dimension than its principal component analysis counterpart. This is evidence that the progenitor system and the explosion mechanism can be described with a small number of initial physical parameters. All tools used here are publicly available in the Python package DRACULA (Dimensionality Reduction And Clustering for Unsupervised Learning in Astronomy) and can be found within COINtoolbox (https://github.com/COINtoolbox/DRACULA).

show abstract

“…for example of type Ia supernovae (Sasdelli et al 2016). These approaches provide information about the specific measured quantities but do not provide a whole picture of the transient.…”

Section: Introductionmentioning

confidence: 99%

Spectral modeling of type II supernovae

Vogl

Kerzendorf

Noebauer

et al. 2020

A&A

View full text Add to dashboard Cite

There are now hundreds of publicly available supernova spectral time series. Radiative transfer modeling of this data provides insight into the physical properties of these explosions, such as the composition, the density structure, and the intrinsic luminosity, which is invaluable for understanding the supernova progenitors, the explosion mechanism, and for constraining the supernova distance. However, a detailed parameter study of the available data has been out of reach due to the high dimensionality of the problem coupled with the still significant computational expense. We tackle this issue through the use of machine-learning emulators, which are algorithms for high-dimensional interpolation. These use a pre-calculated training dataset to mimic the output of a complex code but with run times that are orders of magnitude shorter. We present the application of such an emulator to synthetic type II supernova spectra generated with the TARDIS radiative transfer code. The results show that with a relatively small training set of 780 spectra we can generate emulated spectra with interpolation uncertainties of less than one percent. We demonstrate the utility of this method by automatic spectral fitting of two well-known type IIP supernovae; as an exemplary application, we determine the supernova distances from the spectral fits using the tailored-expanding-photosphere method. We compare our results to previous studies and find good agreement. This suggests that emulation of TARDIS spectra can likely be used to perform automatic and detailed analysis of many transient classes putting the analysis of large data repositories within reach.

show abstract

Exploring the spectroscopic diversity of Type Ia supernovae with dracula: a machine learning approach

Cited by 35 publications

References 97 publications

Transfer Learning in Astronomy: A New Machine-Learning Paradigm

Transfer Learning in Astronomy: A New Machine-Learning Paradigm

Exploring the spectroscopic diversity of type Ia supernovae with Deep Learning and Unsupervised Clustering

Spectral modeling of type II supernovae

Contact Info

Product

Resources

About