MeerCRAB: MeerLICHT classification of real and bogus transients using deep learning

Autor: Kerry Paterson, Zafiirah Hosenie, B. W. Stappers, Paul Vreeswijk, Rudolf S. Le Poole, Robert Lyon, Simon De Wet, Elmar Körding, Bart Scheers, Paul J. Groot, Steven Bloemen, Marc Klein Wolt, Vanessa McBride, Fiorenzo Stoppa, Patrick A. Woudt, D. L. A. Pieterse
Jazyk: angličtina
Rok vydání: 2021
Předmět:
FOS: Computer and information sciences
Astrophysics - instrumentation and methods for astrophysics
Computer science
Computer Vision and Pattern Recognition (cs.CV)
Astronomy
Astrophysics - astrophysics of galaxies
Computer science - artificial intelligence
FOS: Physical sciences
Techniques: image processing
Surveys
Bogus
01 natural sciences
Convolutional neural network
Machine Learning (cs.LG)
Methods: data analysis
0103 physical sciences
Real [Transients]
Instrumentation and Methods for Astrophysics (astro-ph.IM)
010303 astronomy & astrophysics
Ground truth
Network architecture
010308 nuclear & particles physics
business.industry
Deep learning
Astronomy and Astrophysics
Pattern recognition
Deep learning [Methods]
Methods: deep learning
Matthews correlation coefficient
Pipeline (software)
Thresholding
Computer science - machine learning
Data analysis [Methods]
General [Stars]
Computer science - computer vision and pattern recognition
Artificial Intelligence (cs.AI)
Space and Planetary Science
Filter (video)
Image processing [Techniques]
Stars: general
Astrophysics of Galaxies (astro-ph.GA)
Transients: real
Artificial intelligence
business
Zdroj: Experimental Astronomy, 51, 319-344
Hosenie, Z, Lyon, R, Stappers, B, Bloemen, S & Groot, P 2021, ' MeerCRAB: MeerLICHT Classification of Real and Bogus Transients using Deep Learning ', Experimental Astronomy, vol. 51, no. 2, pp. 319–344 . https://doi.org/10.1007/s10686-021-09757-1
Experimental Astronomy, 51, pp. 319-344
Experimental Astronomy, 51, 319–344
ISSN: 0922-6435
DOI: 10.1007/s10686-021-09757-1
Popis: Astronomers require efficient automated detection and classification pipelines when conducting large-scale surveys of the (optical) sky for variable and transient sources. Such pipelines are fundamentally important, as they permit rapid follow-up and analysis of those detections most likely to be of scientific value. We therefore present a deep learning pipeline based on the convolutional neural network architecture called $\texttt{MeerCRAB}$. It is designed to filter out the so called 'bogus' detections from true astrophysical sources in the transient detection pipeline of the MeerLICHT telescope. Optical candidates are described using a variety of 2D images and numerical features extracted from those images. The relationship between the input images and the target classes is unclear, since the ground truth is poorly defined and often the subject of debate. This makes it difficult to determine which source of information should be used to train a classification algorithm. We therefore used two methods for labelling our data (i) thresholding and (ii) latent class model approaches. We deployed variants of $\texttt{MeerCRAB}$ that employed different network architectures trained using different combinations of input images and training set choices, based on classification labels provided by volunteers. The deepest network worked best with an accuracy of 99.5$\%$ and Matthews correlation coefficient (MCC) value of 0.989. The best model was integrated to the MeerLICHT transient vetting pipeline, enabling the accurate and efficient classification of detected transients that allows researchers to select the most promising candidates for their research goals.
15 pages, 13 figures, Accepted for publication in Experimental Astronomy and appeared in the 3rd Workshop on Machine Learning and the Physical Sciences, NeurIPS 2020
Databáze: OpenAIRE
Externí odkaz: