Predikcija vremenskih slijedova primjenom dubokih neuronskih mreža

master's thesis

• Title (english): Time series prediction with deep neural networks
• Author: Danijel Živčec
• Mentor: Mato Baotić (mentor)
• Committee member: Mato Baotić (član povjerenstva)
• Committee member: Nedjeljko Perić (član povjerenstva)
• Committee member: Mario Vašak (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

U ovom su radu prikazane različite strukture i načini rada dubokih neuronskih mreža s naglaskom na autoenkodere i ograničene Boltzmannove strojeve. Te strukture su istražene i ispitana je njihova primjena u predikciji vremenskih slijedova. Različiti testovi su provedeni na klasičnim primjerima vremenskih slijedova: predikcija dioničkog indeksa, potrošnja električne energije i temperaturna mjerenja. Na kraju je prikazana evaluacija različitih dubokih arhitektura.

Abstract (english)

In this thesis, deep neural network architectures and algorithms are analyzed with an emphasis on stacked auto-encoders and restricted Boltzmann machines. These structures were used to explore and test their applicability in time-series prediction. Several tests were designed to predict values of some classical time-series problems: financial time-series, consumption of electrical energy in household and temperature measurements in independent solar house. Following that, the performance of... More different network structures were evaluated and reported. Less

• Keywords: duboke neuronske mreže; ograničeni Boltzmannovi stroj; autoenkoder; MLP mreža; vremenski slijed; predikcija
• Keywords (english): deep neural networks; restricted Boltzmann machine; stacked auto-encoder; MLP Network; time series; prediction
• Language: croatian
• URN:NBN: urn:nbn:hr:168:444619
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 14:37:04