Machine learning and advanced statistics in astronomy: two applications

Date Issued
2015
Author(s)
Abstract
In the last decade, the advances in technology have permitted the development of highly automated surveys in many elds of astronomy. One of the most ambitious is the ESA mission Gaia. Mainly devoted to astrometric measurements in the Galaxy, Gaia will provide also spectroscopic and photometric data. All this information will amount to thousands of terabytes. The same goes for surveys designed to observe transient events: the ongoing Panoramic Survey Telescope and Rapid Response System (PanSTARSS) and Dark Energy Survey (DES) (Bernstein et al. 2009), and the planned Large Synoptic Survey Telescope (LSST) (Ivezic et al. 2008), will produce a huge amount of data.1 The data production is thus quickly increasing, and is most likely to increase more with he future surveys. Astronomy is facing an era of data-ooding, where there will be much more data then we are able to analyse with classical methods. The way to deal with this ood, the way in which we can extract scienti c information in a short time scale, is using techniques developed in the eld of statistics and computer science. In this framework, in this work are presented two applications, one using spectroscopic data and the other photometric data. The rst is the use of an automatic method called MATISSE to determine atmospheric parameters from stellar spectra. The second is the development of a data driven classi er for supernovae using photometric information alone.
File(s)
Thumbnail Image
Name

depascale_phd_thesis.pdf

Description
PhD Thesis
Size

12.04 MB

Format

Adobe PDF

Checksum (MD5)

07a62ae894e305429d0f557acb79cbf6