Stvaranje slika pomoću dubokog učenja

master's thesis

• Title (english): Generating Images Using Deep Learning
• Author: Ivan Rezić
• Mentor: Marin Šilić (mentor)
• Committee member: Marin Šilić (predsjednik povjerenstva)
• Committee member: Siniša Srbljić (član povjerenstva)
• Committee member: Goran Delač (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Zbog svog potencijala, nenadzirano učenje jako je važno područje strojnog učenja. Ono nam omogućuje iskorištavanje velike količine neoznačenih podataka. U ovom radu bavimo se modelima nenadziranog učenja. U uvodnim poglavljima opisani su sastavni dijelovi dubokih modela. Dan je pregled postojećih generativnih modela, a poseban fokus bio je na generativne suparničke modele za stvaranje slika. Predstavljena su četiri modela za stvaranje slika i jedan model za povećavanje rezolucije slika te su implementirani korištenjem biblioteke Keras. Na kraju su prikazani i opisani dobiveni rezultati nad Fashion MNIST i DIV2K podatkovnim skupovima.

Abstract (english)

Because of its potential, unsupervised learning is a very important field of machine learning. It allows us exploiting large amounts of unlabeled data. In this paper we are dealing with models of unsupervised learning. The introductory chapters describe the component parts of deep models. There was given an overview of existing generative models, with emphasis on generative adversarial models for image generation. Four image-creation models and one image-enhancing model were introduced and implemented using library Keras. Finally, in the lasts parts of paper the results over the Fashion MNIST and DIV2K data sets are shown and described.

• Keywords: duboko učenje; generativni suparnički modeli; nenadzirano učenje; stvaranje slika
• Keywords (english): deep learning; generative adversarial networks; unsupervised learning; image generation
• Language: croatian
• URN:NBN: urn:nbn:hr:168:810541
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 21:50:29