Stellar spectral subclasses classification based on Isomap and SVM

Bu, Yude; Chen, F.; Pan, Jiabao

doi:10.1016/j.newast.2013.09.007

Cited by 40 publications

(17 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It has been extensively used in data science and made a break into astronomy as a classification algorithm (Matijevič et al 2017;Lochner et al 2016;Valentini et al 2017;Traven et al 2017), along with other manifold learning algorithms (e.g. Vanderplas & Connolly 2009;Daniel et al 2011;Bu et al 2014). We extend its use as a pure manifold learning algorithm to find structure in a 13-dimensional Cspace.…”

Section: T-distributed Stochastic Neighbour Embedding (T-sne)mentioning

confidence: 99%

The GALAH survey: chemical tagging of star clusters and new members in the Pleiades

Kos

Bland-Hawthorn

Freeman

et al. 2017

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

The technique of chemical tagging uses the elemental abundances of stellar atmospheres to 'reconstruct' chemically homogeneous star clusters that have long since dispersed. The GALAH spectroscopic survey -which aims to observe one million stars using the Anglo-Australian Telescope -allows us to measure up to 30 elements or dimensions in the stellar chemical abundance space, many of which are not independent. How to find clustering reliably in a noisy high-dimensional space is a difficult problem that remains largely unsolved. Here we explore t-distributed stochastic neighbour embedding (t-SNE) -which identifies an optimal mapping of a high-dimensional space into fewer dimensions -whilst conserving the original clustering information. Typically, the projection is made to a 2D space to aid recognition of clusters by eye. We show that this method is a reliable tool for chemical tagging because it can: (i) resolve clustering in chemical space alone, (ii) recover known open and globular clusters with high efficiency and low contamination, and (iii) relate field stars to known clusters. t-SNE also provides a useful visualization of a high-dimensional space. We demonstrate the method on a dataset of 13 abundances measured in the spectra of 187,000 stars by the GALAH survey. We recover 7 of the 9 observed clusters (6 globular and 3 open clusters) in chemical space with minimal contamination from field stars and low numbers of outliers. With chemical tagging, we also identify two Pleiades supercluster members (which we confirm kinematically), one as far as 6 • -one tidal radius away from the cluster centre.

show abstract

Section: T-distributed Stochastic Neighbour Embedding (T-sne)mentioning

confidence: 99%

The GALAH survey: chemical tagging of star clusters and new members in the Pleiades

Kos

Bland-Hawthorn

Freeman

et al. 2017

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

show abstract

“…Thus, in contrast to MDS, Isomap can also capture non-linear manifold structures. In astronomy, it was recently applied to spectroscopic classification by Bu et al (2014).…”

Section: Isomapmentioning

confidence: 99%

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

Sasdelli

Ishida

Vilalta

et al. 2016

Mon. Not. R. Astron. Soc.

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. One major challenge inevitably met when trying to infer the existence of one or more subclasses is the time consuming, and subjective, process of subclass definition. In this work, we show how machine learning tools facilitate identification of subtypes of SNe Ia through the establishment of a hierarchical group structure in the continuous space of spectral diversity formed by these objects. Using Deep Learning, we were capable of performing such identification in a 4 dimensional feature space (+1 for time evolution), while the standard Principal Component Analysis barely achieves similar results using 15 principal components. This is evidence that the progenitor system and the explosion mechanism can be described by a small number of initial physical parameters. As a proof of concept, we show that our results are in close agreement with a previously suggested classification scheme and that our proposed method can grasp the main spectral features behind the definition of such subtypes. This allows the confirmation of the velocity of lines as a first order effect in the determination of SN Ia subtypes, followed by 91bg-like events. Given the expected data deluge in the forthcoming years, our proposed approach is essential to allow a quick and statistically coherent identification of SNeIa subtypes (and outliers). All tools used in this work were made 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

“…The resulting pairwise geodesic distance matrix is then transformed using classical multidimensional scaling [48]. Isomap was used for instance in classification of stellar spectral subclasses in SDSS data [5] and for discovering white dwarf + main sequence for the same survey [53]. In both cases as the classification engine Support Vector Machine method was employed, with the superiority of this solution over the one using PCA being demonstrated once more.…”

Section: Techniques Of Feature Extractionmentioning

confidence: 99%

Survey of Object-Based Data Reduction Techniques in Observational Astronomy

et al. 2016

View full text Add to dashboard Cite

Abstract:Dealing with astronomical observations represents one of the most challenging areas of big data analytics. Besides huge variety of data types, dynamics related to continuous data flow from multiple sources, handling enormous volumes of data is essential. This paper provides an overview of methods aimed at reducing both the number of features/attributes as well as data instances. It concentrates on data mining approaches not related to instruments and observation tools instead working on processed object-based data. The main goal of this article is to describe existing datasets on which algorithms are frequently tested, to characterize and classify available data reduction algorithms and identify promising solutions capable of addressing present and future challenges in astronomy.

show abstract

Stellar spectral subclasses classification based on Isomap and SVM

Cited by 40 publications

References 9 publications

The GALAH survey: chemical tagging of star clusters and new members in the Pleiades

The GALAH survey: chemical tagging of star clusters and new members in the Pleiades

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

Survey of Object-Based Data Reduction Techniques in Observational Astronomy

Contact Info

Product

Resources

About