prikaz prve stranice dokumenta Automated Classification of LSST Images Using Convolutional Neural Networks
Access restricted to students and staff of home institution
master's thesis
Automated Classification of LSST Images Using Convolutional Neural Networks
Rijeka: University of Rijeka, 2021. urn:nbn:hr:194:924805
Mrakovčić, Karlo
University of Rijeka
Faculty of Physics

Institutional repository: Repository of the University of Rijeka, Faculty of Physics

Cite this document

Mrakovčić, K. (2021). Automated Classification of LSST Images Using Convolutional Neural Networks (Master's thesis). Rijeka: University of Rijeka. Retrieved from https://urn.nsk.hr/urn:nbn:hr:194:924805

Mrakovčić, Karlo. "Automated Classification of LSST Images Using Convolutional Neural Networks." Master's thesis, University of Rijeka, 2021. https://urn.nsk.hr/urn:nbn:hr:194:924805

Mrakovčić, Karlo. "Automated Classification of LSST Images Using Convolutional Neural Networks." Master's thesis, University of Rijeka, 2021. https://urn.nsk.hr/urn:nbn:hr:194:924805

Mrakovčić, K. (2021). 'Automated Classification of LSST Images Using Convolutional Neural Networks', Master's thesis, University of Rijeka, accessed 29 November 2024, https://urn.nsk.hr/urn:nbn:hr:194:924805

Mrakovčić K. Automated Classification of LSST Images Using Convolutional Neural Networks [Master's thesis]. Rijeka: University of Rijeka; 2021 [cited 2024 November 29] Available at: https://urn.nsk.hr/urn:nbn:hr:194:924805

K. Mrakovčić, "Automated Classification of LSST Images Using Convolutional Neural Networks", Master's thesis, University of Rijeka, Rijeka, 2021. Available at: https://urn.nsk.hr/urn:nbn:hr:194:924805

Metadata
TitleAutomated Classification of LSST Images Using Convolutional Neural Networks
Title (croatian)Automatizirana klasifikacija LSST slika korištenjem konvolucijskih neuronskih mreža
AuthorKarlo Mrakovčić
MentorDijana Dominis Prester (mentor)
MentorŽeljko Ivezić (komentor)
Committee memberRajka Jurdana-Šepić (predsjednik povjerenstva)
Committee memberDijana Dominis Prester (član povjerenstva)
Committee memberŽeljko Ivezić (član povjerenstva)
GranterUniversity of Rijeka
(Faculty of Physics)
Rijeka
Defense date and country2021-09-13, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Physics
Abstract
With its 3.2 gigapixel camera, the Vera Rubin observatory can produce images of up to 10 million changing celestial sources per night. Its first principal project, the Legacy Survey of Space and Time (LSST), will be a multi-band large-area time-domain astronomical survey. Because the data is impossible to classify and analyze in the conventional way, the telescope will need to take a different approach in order to generate quality scientific products. Because of the volume of data collected (20 TB each night), accurate data categorization and false positive detection automation must be created. Images of three types of sources were generated to approximate anticipated LSST images: ”stars,” ”trails,” and ”dipoles.” Unsupervised learning was utilized to automate the categorization of generated LSST images using a combination of convolutional autoencoder and Kmeans. Hyperparameter search was done on HPC Bura to explore hyperparameter space, and cMetric, a measure for evaluating different models, was devised. The accuracy of 91.2 percent was reached, presenting us with a promising tool that can be integrated one day into LSST pipeline.
Abstract (croatian)
Sa svojom kamerom od 3,2 gigapiksela, opservatorij Vera Rubin može proizvesti do 10 milijuna slika u jednoj noći. Njegov prvi glavni projekt, Legacy Survey of Space and Time (LSST), bit ´ce višepojasni astronomski pregled neba na velikom području vremenske domene. Budući da je podatke nemoguće klasificirati i analizirati na konvencionalan način, obrada podataka će morati biti automatizirana, kako bi se generirali kvalitetni znanstveni rezultati. Zbog opsega prikupljenih podataka (20 TB svake noći), potrebno je stvoriti točnu kategorizaciju podataka i automatizaciju otkrivanja lažno pozitivnih podataka. Generirane su tri vrste slika, kako bi se to simuliralo: ”zvijezde”, ”staze” i ”dipoli”. Nenadzirano učenje korišteno je za automatiziranje kategorizacije generiranih LSST slika, pomoću kombinacije konvolucijskog autoenkodera i algoritma Kmeans. Na HPC Buri izvršeno je pretraživanje hiperparametara, u svrhu istraživanja hiperparametarskog prostora, a osmišljen je i cMetric, mjera za ocjenu različitih modela. Dostignuta je točnost od 91,2 posto koja predstavlja obećavajući alat kojeg će biti moguće integrirati u LSST sustav.
Keywords
Languagecroatian
URN:NBNurn:nbn:hr:194:924805
Study programmeTitle: physics
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra fizike (magistar/magistra fizike)
Type of resourceText
File originBorn digital
Access conditionsAccess restricted to students and staff of home institution
Terms of useLicence In copyright icon
Created on2021-10-21 08:59:52