크기변환_김대원

ECE Colloquium: Dae-Won Kim(ETRI) “Machine Learning Application in Astronomy”

March 22, 2017 / 16:00 ~ 17:15

Speaker : Dae-Won Kim

 

Abstract

After the identification of the first variable star in 1638 by Johannes Holwarda, the conventional belief of never-ending immutability of a starry sky turned out to be wrong: “The Heavens are Changing”. Since that time, monitoring brightness of various sources in the sky has provided crucial informations for astronomers for centuries. For instance, Lemaitre (1927) and Hubble (1929) found the first evidence of the expanding universe using a relation between distance and radial velocity of galaxies, where the distance was derived using a Period-Luminocity relation (P-L relation) of Cepheid variables. An another astonishing discovery in 20th century was a detection of an observational evidence for accelerated expansion of the universe using lightcurves of Type Ia supernova explosions (Riess et al., 1998). Not surprisingly, this groundbreaking discovery awarded him and his colleagues the Nobel Prize.

In this talk, I will introduce several astronomical survey projects and explain how they use machine learning to classify stellar objects in the sky and to estimate physical parameters of them. Given that ongoing wide-field surveys such as Gaia, LSST, SkyMapper, Pan-STARRS, etc. produce a vast volume of data, automated detection methods are crucial because manual insepection is nearly impossible.

 

Bio

Kim received his PhD in 2012 at the Department of Astronomy, Yonsei University, South Korea. During his 5 years PhD period, he had been affiliated as a predoc at the Harvard-Smithsonian Center for Astrophysics for analyzing astronomical time series and QSO classification using machine learning. He then joined the Max-Planck Institute for Astronomy for the ESA satellite project, Gaia, as a postdoc and worked for the classification of astronomical objects using machine learning and statistical data analysis. He recently joined ETRI for researching general application of machine learning.

 

Venue

104-E207