prikaz prve stranice dokumenta Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža
  Download
PDF 20.81 MB
master's thesis
Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža
2024. urn:nbn:hr:126:910636
Strapač, Luka
Josip Juraj Strossmayer University of Osijek
School of Applied Mathematics and Informatics

Cite this document

Strapač, L. (2024). Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics. Retrieved from https://urn.nsk.hr/urn:nbn:hr:126:910636

Strapač, Luka. "Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics, 2024. https://urn.nsk.hr/urn:nbn:hr:126:910636

Strapač, Luka. "Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics, 2024. https://urn.nsk.hr/urn:nbn:hr:126:910636

Strapač, L. (2024). 'Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža', Master's thesis, Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics, accessed 10 November 2024, https://urn.nsk.hr/urn:nbn:hr:126:910636

Strapač L. Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics; 2024 [cited 2024 November 10] Available at: https://urn.nsk.hr/urn:nbn:hr:126:910636

L. Strapač, "Primjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža", Master's thesis, Josip Juraj Strossmayer University of Osijek, School of Applied Mathematics and Informatics, Osijek, 2024. Available at: https://urn.nsk.hr/urn:nbn:hr:126:910636

Please login to the repository to save this object to your list.
Metadata
TitlePrimjena pristupa meta-učenja kao metoda brzog učenja neuronskih mreža
Title (english)Applying Meta-Learning Approches As Methods For Quick Learning Of Neural Networks
AuthorLuka Strapač
MentorDomagoj Ševerdija (mentor)
Committee memberDomagoj Matijević (predsjednik povjerenstva)
Committee memberDomagoj Ševerdija (član povjerenstva)
Committee memberMateja Đumić (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
School of Applied Mathematics and Informatics
Osijek
Defense date and country2024-10-30, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Mathematics
Mathematical Logic and Accounting
Abstract
U ovom radu opisujemo koncept meta-učenja i njegovu ulogu kod brzog učenja i prilagodbe neuronskih mreža na nove zadatke. U današnje vrijeme se u području umjetne inteligencije još uvijek susrećemo s problemom loše generalizacije modela na nove zadatke. Zbog toga, za svaki novi zadatak moramo konstruirati novi skup podataka što može biti vrlo skupocjeno. Meta-učenje se bavi upravo tim problemom, odnosno, prilagođavanjem modela na novi zadatak uz minimalan broj primjera. Kroz rad definiramo meta-učenje, navodimo najpoznatije pristupe te implementiramo modele meta-učenika triju pristupa koristeći learn2learn paket u programskom jeziku Python.
Abstract (english)
In this paper we describe meta-learning concept and its role in quick learning and adaptation of neural networks on new tasks. Nowadays in artifical intelligence area we still face problem of poor generalization of models on new tasks. Consequently, for every new task we need to construct new dataset, which can be very expensive. Meta-learning deals with exactly this problem, i.e., adapting model on new task with a minimal number of examples. Through paper we define metalearning, list the most popular approaches and implement meta learner models of 3 approaches using learn2learn package in programming language Python.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:126:910636
Study programmeTitle: Mathematics; specializations in: Financial and Statistical Mathematics, Mathematics and Computer Science, Industrial and Applied Mathematics
Course: Mathematics and Computer Science
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 on2024-11-04 11:11:26