Istraživanje asteroida primjenom algoritama grupiranja u strojnom učenju

master's thesis

• Title (english): The application of machine learning clustering algorithms to the study of asteroids
• Author: Vedrana Đunđek
• Mentor: Goranka Bilalbegović (mentor)
• Committee member: Goranka Bilalbegović (predsjednik povjerenstva)
• Committee member: Davor Horvatić (član povjerenstva)
• Committee member: Mihael Makek (član povjerenstva)
• Committee member: Krešimir Kumerički (član povjerenstva)
• Committee member: Željko Skoko (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Physics); Zagreb
• Defense date and country: 2017-08-30, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Physics

Abstract

S napretkom teleskopa i razvojem algoritama strojnog učenja povećava se mogućnost uočavanja novih asteroida i određivanja njihovih osobina. U ovom diplomskom radu analizirani su asteroidi u blizini Zemlje pomoću nenadziranog strojnog učenja. Koristili smo dva algoritama grupiranja u programskom jeziku Python: KMEANS i DBSCAN. Pokušali smo grupirati asteroide u blizini Zemlje koristeći podatke američke državne uprave za zrakoplovna i svemirska istraživanja (NASA). Usporedili smo rezultate s poznatim potencijalno opasnim asteroidima. Trenutne verzije korištenih programa dostupne su na adresi https://github.com/lib686/avenge-the-dinosaurs. Opisan je projekt znanosti za građanstvo Asteroids@home kao metodički dio diplomskog rada. Diskutiran je značaj ovog projekta za popularizaciju astronomije, informatike te znanosti općenito.

Abstract (english)

With the advancement of telescopes and the development of machine learning algorithms we increase the ability to detect new asteroids and determine their characteristics. In this master’s degree thesis, Near-Earth Asteroids (NEAs) were analyzed using unsupervised machine learning. We were using two clustering algorithms in programming language Python: KMEANS and DBSCAN. We attempted to cluster NEAs using data from National Aeronautics and Space Administration (NASA). We compared the results with known Potentially Hazardous Asteroids (PHA). Current versions of the programs used can be found at https://github.com/lib686/ avenge-the-dinosaurs. We described the citizen science project Asteroids@home as a teaching method part of the thesis. The importance of this project for the popularization of astronomy, computer science, and science in general is discussed.

• Keywords: asteroidi; potencijalno opasni asteroidi; strojno učenje; algoritmi grupiranja; algoritam k-srednjih vrijednosti; DBSCAN algoritam
• Keywords (english): asteroids; potentially hazardous asteroids; machine learning; clustering algorithms; k-means algorithm; DBSCAN algorithm
• Language: croatian
• URN:NBN: urn:nbn:hr:217:058781
• Study programme: Title: Physics and Computer Science; specializations in: Education; Course: Education; Study programme type: university; Study level: integrated undergraduate and graduate; Academic / professional title: magistar/magistra fizike i informatike (magistar/magistra fizike i informatike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access; Embargo expiration date: 2018-08-30
• Created on: 2017-09-13 11:19:25