prikaz prve stranice dokumenta Optimizacija modela temeljenih na Gradient Boosting metodi
  Download
PDF 5.08 MB
master's thesis
Optimizacija modela temeljenih na Gradient Boosting metodi
Pula: University of Pula, 2022. urn:nbn:hr:137:676578
Markov, Petar
University of Pula
Faculty of Informatics in Pula

Institutional repository: Digital Repository Juraj Dobrila University of Pula

Cite this document

Markov, P. (2022). Optimizacija modela temeljenih na Gradient Boosting metodi (Master's thesis). Pula: University of Pula. Retrieved from https://urn.nsk.hr/urn:nbn:hr:137:676578

Markov, Petar. "Optimizacija modela temeljenih na Gradient Boosting metodi." Master's thesis, University of Pula, 2022. https://urn.nsk.hr/urn:nbn:hr:137:676578

Markov, Petar. "Optimizacija modela temeljenih na Gradient Boosting metodi." Master's thesis, University of Pula, 2022. https://urn.nsk.hr/urn:nbn:hr:137:676578

Markov, P. (2022). 'Optimizacija modela temeljenih na Gradient Boosting metodi', Master's thesis, University of Pula, accessed 29 April 2024, https://urn.nsk.hr/urn:nbn:hr:137:676578

Markov P. Optimizacija modela temeljenih na Gradient Boosting metodi [Master's thesis]. Pula: University of Pula; 2022 [cited 2024 April 29] Available at: https://urn.nsk.hr/urn:nbn:hr:137:676578

P. Markov, "Optimizacija modela temeljenih na Gradient Boosting metodi", Master's thesis, University of Pula, Pula, 2022. Available at: https://urn.nsk.hr/urn:nbn:hr:137:676578

Metadata
TitleOptimizacija modela temeljenih na Gradient Boosting metodi
Title (english)Optimization of Models Based on Gradient Boosting Method
AuthorPetar Markov
MentorNikola Tanković (mentor)
Committee memberGoran Oreški (predsjednik povjerenstva)
Committee memberNikola Tanković (član povjerenstva)
Committee memberSiniša Sovilj (član povjerenstva)
GranterUniversity of Pula
(Faculty of Informatics in Pula)
Pula
Defense date and country2022-06-28, Croatia
Scientific / art field,
discipline and subdiscipline		SOCIAL SCIENCES
Information and Communication Sciences
Information Systems and Information Science
Abstract
Na internetu imamo sve više podataka koji svakim danom eksponencijalno rastu te se povećava potreba za obradom tih podataka. U tome nam uvelike pomaže strojno učenje i modeli koji se koriste za obradu podataka. Svaki od modela dolazi sa unaprijed postavljenim parametrima koji poboljšavaju modele. Ti parametri se nazivaju hiperparemetri modela, a njihova optimizacija je jedan od najvažnijih koraka u izgradnji modela. Optimizacija hiperparametara se može odraditi ručno, ali to je iscrpan posao koji oduzima puno vremena. Zato su napravljeni algoritmi i tehnike automatske optimizacije hiperparametara od kojih su u ovome diplomskom radu spomenute sljedeće: Pretraživanje po rešetci (eng. Grid Serach), Nausumično pretraživanje (eng. Random Serach), Bayesova optimziacija (eng. Bayesian Optimization), Treestructured Parzen Estimator i Simmulated Anneal.
Abstract (english)
Every day across the Internet the number of data is growing exponentially, and therefore the need for processing that data is growing also. Machine learning and models used in it are helping us in processing that data. Every model comes with preset parameters which improve these models. These parameters are called hyperparameters of model and their optimization is one of the most important steps in building the models. Optimization of hyperparameters can be done manually, but that is exhausting, and most of the time, time-consuming job. Therefore, algorithms and techniques for automatic optimization of hyperparameters have been developed, of which the following are mentioned in this thesis: Grid Search, Random Search, Bayesian Optimization, Treestructured Parzen Estimator and Simmulated Anneal.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:137:676578
Study programmeTitle: Master in Informatics; specializations in: Master in Informatics, Master in Informatics, specialisation: teaching, Master in Informatics, Master in Informatics, specialisation: teaching
Course: Master in Informatics
Study programme type: university
Study level: graduate
Academic / professional title: MAGISTAR/MAGISTRA INFORMATIKE - MAG.INF. (MAGISTAR/MAGISTRA INFORMATIKE - MAG.INF.)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2022-07-22 08:52:03