Abstract

Most existing star-galaxy classifiers use the reduced summary information from catalogues, requiring careful feature extraction and selection. The latest advances in machine learning that use deep convolutional neural networks (ConvNets) allow a machine to automatically learn the features directly from the data, minimizing the need for input from human experts. We present a star-galaxy classification framework that uses deep ConvNets directly on the reduced, calibrated pixel values. Using data from the Sloan Digital Sky Survey and the Canada-France-Hawaii Telescope Lensing Survey, we demonstrate that ConvNets are able to produce accurate and well-calibrated probabilistic classifications that are competitive with conventional machine learning techniques. Future advances in deep learning may bring more success with current and forthcoming photometric surveys, such as the Dark Energy Survey and the Large Synoptic Survey Telescope, because deep neural networks require very little, manual feature engineering.

Original languageEnglish (US)
Pages (from-to)4463-4475
Number of pages13
JournalMonthly Notices of the Royal Astronomical Society
Volume464
Issue number4
DOIs
StatePublished - Jan 1 2017

Fingerprint

galaxies
stars
machine learning
telescopes
classifiers
dark energy
France
Canada
pattern recognition
learning
catalogs
pixels
engineering
pixel
energy

Keywords

  • Galaxies: statistics
  • Methods: data analysis
  • Methods: statistical
  • Stars: statistics
  • Surveys
  • Techniques: image processing

ASJC Scopus subject areas

  • Astronomy and Astrophysics
  • Space and Planetary Science

Cite this

Star-galaxy classification using deep convolutional neural networks. / Kim, Edward J.; Brunner, Robert J.

In: Monthly Notices of the Royal Astronomical Society, Vol. 464, No. 4, 01.01.2017, p. 4463-4475.

Research output: Contribution to journalArticle

@article{41288c7399e34999bf739857d6cb75e1,
title = "Star-galaxy classification using deep convolutional neural networks",
abstract = "Most existing star-galaxy classifiers use the reduced summary information from catalogues, requiring careful feature extraction and selection. The latest advances in machine learning that use deep convolutional neural networks (ConvNets) allow a machine to automatically learn the features directly from the data, minimizing the need for input from human experts. We present a star-galaxy classification framework that uses deep ConvNets directly on the reduced, calibrated pixel values. Using data from the Sloan Digital Sky Survey and the Canada-France-Hawaii Telescope Lensing Survey, we demonstrate that ConvNets are able to produce accurate and well-calibrated probabilistic classifications that are competitive with conventional machine learning techniques. Future advances in deep learning may bring more success with current and forthcoming photometric surveys, such as the Dark Energy Survey and the Large Synoptic Survey Telescope, because deep neural networks require very little, manual feature engineering.",
keywords = "Galaxies: statistics, Methods: data analysis, Methods: statistical, Stars: statistics, Surveys, Techniques: image processing",
author = "Kim, {Edward J.} and Brunner, {Robert J}",
year = "2017",
month = "1",
day = "1",
doi = "10.1093/mnras/stw2672",
language = "English (US)",
volume = "464",
pages = "4463--4475",
journal = "Monthly Notices of the Royal Astronomical Society",
issn = "0035-8711",
publisher = "Oxford University Press",
number = "4",

}

TY - JOUR

T1 - Star-galaxy classification using deep convolutional neural networks

AU - Kim, Edward J.

AU - Brunner, Robert J

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Most existing star-galaxy classifiers use the reduced summary information from catalogues, requiring careful feature extraction and selection. The latest advances in machine learning that use deep convolutional neural networks (ConvNets) allow a machine to automatically learn the features directly from the data, minimizing the need for input from human experts. We present a star-galaxy classification framework that uses deep ConvNets directly on the reduced, calibrated pixel values. Using data from the Sloan Digital Sky Survey and the Canada-France-Hawaii Telescope Lensing Survey, we demonstrate that ConvNets are able to produce accurate and well-calibrated probabilistic classifications that are competitive with conventional machine learning techniques. Future advances in deep learning may bring more success with current and forthcoming photometric surveys, such as the Dark Energy Survey and the Large Synoptic Survey Telescope, because deep neural networks require very little, manual feature engineering.

AB - Most existing star-galaxy classifiers use the reduced summary information from catalogues, requiring careful feature extraction and selection. The latest advances in machine learning that use deep convolutional neural networks (ConvNets) allow a machine to automatically learn the features directly from the data, minimizing the need for input from human experts. We present a star-galaxy classification framework that uses deep ConvNets directly on the reduced, calibrated pixel values. Using data from the Sloan Digital Sky Survey and the Canada-France-Hawaii Telescope Lensing Survey, we demonstrate that ConvNets are able to produce accurate and well-calibrated probabilistic classifications that are competitive with conventional machine learning techniques. Future advances in deep learning may bring more success with current and forthcoming photometric surveys, such as the Dark Energy Survey and the Large Synoptic Survey Telescope, because deep neural networks require very little, manual feature engineering.

KW - Galaxies: statistics

KW - Methods: data analysis

KW - Methods: statistical

KW - Stars: statistics

KW - Surveys

KW - Techniques: image processing

UR - http://www.scopus.com/inward/record.url?scp=85014847810&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85014847810&partnerID=8YFLogxK

U2 - 10.1093/mnras/stw2672

DO - 10.1093/mnras/stw2672

M3 - Article

AN - SCOPUS:85014847810

VL - 464

SP - 4463

EP - 4475

JO - Monthly Notices of the Royal Astronomical Society

JF - Monthly Notices of the Royal Astronomical Society

SN - 0035-8711

IS - 4

ER -