prikaz prve stranice dokumenta Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom
  Download
PDF 18.4 MB
master's thesis
Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom
2019. urn:nbn:hr:126:647855
Jaganjac, Danijela
Josip Juraj Strossmayer University of Osijek
Department of Mathematics

Cite this document

Jaganjac, D. (2019). Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Department of Mathematics. Retrieved from https://urn.nsk.hr/urn:nbn:hr:126:647855

Jaganjac, Danijela. "Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom." Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, 2019. https://urn.nsk.hr/urn:nbn:hr:126:647855

Jaganjac, Danijela. "Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom." Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, 2019. https://urn.nsk.hr/urn:nbn:hr:126:647855

Jaganjac, D. (2019). 'Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom', Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, accessed 23 December 2024, https://urn.nsk.hr/urn:nbn:hr:126:647855

Jaganjac D. Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Department of Mathematics; 2019 [cited 2024 December 23] Available at: https://urn.nsk.hr/urn:nbn:hr:126:647855

D. Jaganjac, "Primjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom", Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, Osijek, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:126:647855

Please login to the repository to save this object to your list.
Metadata
TitlePrimjena dubokog učenja u analizi rezultata pretraga koljena magnetskom rezonancom
Title (english)Application of Deep Learning in Analysis of Knee Magnetic Resonance Imaging
AuthorDanijela Jaganjac
MentorDomagoj Matijević (mentor)
Committee memberKristian Sabo (predsjednik povjerenstva)
Committee memberDomagoj Matijević (član povjerenstva)
Committee memberDomagoj Ševerdija (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Department of Mathematics
Osijek
Defense date and country2019-09-17, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Mathematics
Mathematical Logic and Accounting
Abstract
U ovom diplomskom radu je objašnjen pojam neuronskih i konvolucijskih mreža. Njihovo djelovanje se definira matematičkim izrazima i formulama. Ulazni podaci u neuronskim mreža prolaze kroz niz slojeva pri čemu se transformiraju i kao rezultat dobijemo vrijednosti koje dalje interpretiramo sukladno početnom problemu. Svaki model neuronskih i konvolucijskih mreža ima parametre čije vrijednosti utječu na točnost predikcija. U svrhu povećavanje točnosti, parametri se podešavaju u procesu učenja. Pristup procesu učenja je, u suštini, matematička optimizacija nekih nelinearnih funkcija koje ovise o tim parametrima. Stoga se koriste neki standardni algoritmi za optimizaciju, ali isto tako postoje algoritmi razvijeni specifično za probleme dubokog učenja. Kako bi se povećala kvaliteta modela, potrebno je provesti postupke obrade podataka prije samog učenja i regularizacije. Motiv za kreiranjem kvalitetnih mreža za duboko učenje je njihova praktična upotreba u raznim znanstvenim granama i djelatnostima. Jedna od primjena je u medicini, za klasifikaciju slika s magnetske rezonance, gdje bi dobro naučeni model mogao pridonijeti efikasnijem dijagnosticiranju ozljeda i bolesti.
Abstract (english)
This master thesis describes and presents the concept of neural and conventional networks. The basic flow of actions in those networks is defined by mathematical expressions and formulas. Input data are being transformed while flowing through the network and the obtained output is then used to form an answer to the initial problem. Network parameters are crucial for the process of learning and their values directly effect precision. In order to obtain better performance parameters are being tuned in the learning process which is basically the optimization of some nonlinear functions with respect to the mentioned parameters. Optimization is using standard algorithms from numerical mathematics, but there are also special algorithms which are results of optimization in deep learning. Building a deep learning model includes preprocessing data and regularization. High-quality deep learning networks are useable in various scientific disciplines and businesses. One of the widespread application is for medical diagnosis and medical imaging classification. This kind of model can contribute to decision-making in a way that it gives automated diagnosis and localization of abnormalities.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:126:647855
Study programmeTitle: Mathematics; specializations in: Financial and Statistical Mathematics, Mathematics and Computer Science, Industrial and Applied Mathematics
Course: Financial and Statistical Mathematics
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2019-09-19 06:24:15