Download
PDF 887.17 KB
master's thesis
42   11
Mathematical foundations of neural networks
Matea Kalinić (2017)
University of Split
University of Split, Faculty of science
Department of Mathematics
Cite this document...

Kalinić, M. (2017). Matematičke osnove neuronskih mreža (Master's thesis). Retrieved from https://urn.nsk.hr/urn:nbn:hr:166:112258

Kalinić, Matea. "Matematičke osnove neuronskih mreža." Master's thesis, University of Split, University of Split, Faculty of science, 2017. https://urn.nsk.hr/urn:nbn:hr:166:112258

Kalinić, Matea. "Matematičke osnove neuronskih mreža." Master's thesis, University of Split, University of Split, Faculty of science, 2017. https://urn.nsk.hr/urn:nbn:hr:166:112258

Kalinić, M. (2017). 'Matematičke osnove neuronskih mreža', Master's thesis, University of Split, University of Split, Faculty of science, accessed 22 March 2019, https://urn.nsk.hr/urn:nbn:hr:166:112258

Kalinić M. Matematičke osnove neuronskih mreža [Master's thesis]. Split: University of Split, University of Split, Faculty of science; 2017 [cited 2019 March 22] Available at: https://urn.nsk.hr/urn:nbn:hr:166:112258

M. Kalinić, "Matematičke osnove neuronskih mreža", Master's thesis, University of Split, University of Split, Faculty of science, Split, 2017. Available at: https://urn.nsk.hr/urn:nbn:hr:166:112258

Metadata
TitleMatematičke osnove neuronskih mreža
Title (english)Mathematical foundations of neural networks
AuthorMatea Kalinić
MentorIvo Ugrina (mentor)
Committee memberIvo Ugrina (predsjednik povjerenstva)
Committee memberSnježana Braić (član povjerenstva)
Committee memberGoran Erceg (član povjerenstva)
GranterUniversity of Split
University of Split, Faculty of science
(Department of Mathematics)
Split
Defense date and country2017-06-13, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Mathematics
Abstract
Glavni cilj ovog rada objasniti je matematički model po kojem računalo oponaša proces čovjekovog učenja. U tu svrhu definirane su jednostavne i višeslojne neuronske mreže kao objekt podložan učenju te je oprimjerena njihova upotreba na logičkim funkcijama. Također, objašnjen je algoritam s povratnim proslijeđivanjem pogreške, čija je glavna ideja utemeljena na metodi gradijentnog spusta, koji mijenjanjem parametara neuronskih mreža omogućava učenje istih. Algoritam je proveden na jednoslojnoj neuronskoj mreži. Naposljetku, uvedena je Elmanova mreža kao najjednostavniji primjer povratnih neuronskih mreža i primjerom pokazano učenje te mreže.
Abstract (english)
The main objective of this master’s thesis is to introduce a mathematical model which enables computers to imitate human’s learning process. For this purpose, simple and multi-layer neural networks are defined as objects that can be trained and a few examples for their usage are provided. Also, Backpropagation algoritam, whose main idea is based on gradient descent method and which enables neural network to learn by changing their parametar, is explained. The algorithm was implemented on a single-layer neural network. Lastly, Elman’s network was introduced as the simplest example of reccurent neural networks and its learning is demonstrated.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:166:112258
Study programmeTitle: Matematika; smjerovi: teorijski, računarski, nastavnički
Study programme type: sveučilišni
Study level: diplomski
Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
Type of resourcetext
File originBorn digital
Access conditionsOpen access
Terms of usehttp://creativecommons.org/licenses/by-nc-nd/4.0/
Created on2018-05-18 15:43:46