VGTU talpykla > Doktorantūros skyrius / Department for Doctoral Studies > Technologijos mokslų daktaro disertacijos ir jų santraukos >

Lietuvių   English
Please use this identifier to cite or link to this item:

Title: Edge Computing Tied in Artificial Neural Network Classifiers
Other Titles: Dirbtinių neuronų tinklų klasifikatoriais susieta kraštų kompiuterija
Authors: Skirelis, Julius
Issue Date: 6-May-2021
Publisher: Vilniaus Gedimino technikos universitetas
Citation: Skirelis, J. 2021. Edge Computing Tied in Artificial Neural Network Classifiers: doctoral dissertation. Vilnius: Technika, 152 p.
Abstract: The dissertation deals with traditional and artificial neural networks based intelligent classifiers, simulations and natural experiments are performed taking into account their specifics and possible specific applications in: cell colony image parametrization and image stitching. Research is conducted to evaluate the poten tial of developed algorithms and methods application to address edge computing challenges and are therefore examined in different network topologies: centralized, decentralized and distributed. Edge computing and its benefits are compared to the currently relevant but no longer meeting the needs technology – the cloud computing. The object of the dissertation is the intelligent classifier for edge computing tasks. The aim of the work is to propose and investigate original, artificial neural network classifier based Internet of Things systems for the edge computing tasks. The dissertation is designated to contribute to the development of artificial neural network based edge computing solutions. Analytical review of artificial neural networks for edge computing is performed that underlies the work the relevance of the raise problem and explains the importance of research of the classifier application in edge computing. The dissertation consists of an introduction, four chapters and general conclusions. In the first chapter essential knowledge and progress on artificial neural networks based classifiers application in edge computing technology is presented, the relationship between the edge computing and Internet of Things is defined, traditional and intelligent classification methods are discussed, self-organizing adaptive resonance theory based classifiers are analyzed and dissertation tasks are formulated. In the second chapter three simulation experiments are described with further analysis of their results revealing quantitative and qualitative Internet of Things characteristics, lastly, Internet of Things, edge computing and sensor network specifics are revealed. The third chapter describes the proposed cell colony images parameterization method, the experiments using constructed stand are performed comparing heuristic, support vectors, and adaptive resonance theory versions 1 and 2 classifiers by evaluation of classification quality based on statistical methods. In the fourth chapter a proposed fixed-structure artificial neural network classifier based adaptive resolution selection stitching system is described, further the arranged constructed stand and obtained experiment results are discussed. The main results of the thesis were published in 9 scientific publications: three of them were printed in peer-reviewed scientific journals, two of which are listed in Clarivate Analytics Web of Science and one of them is with impact factor, six articles – in conference proceedings. The research results were presented in 11 scientific conferences.
Description: Doctoral dissertation
Appears in Collections:Technologijos mokslų daktaro disertacijos ir jų santraukos

Files in This Item:

File Description SizeFormat
Skirelio disertacija WEB.pdf1.85 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.


Valid XHTML 1.0! DSpace Software Copyright © 2002-2010  Duraspace - Feedback