magistrsko delo
Lucijano Berus (Author), Simon Klančnik (Mentor)

Abstract

Magistrsko delo obravnava področje umetne inteligence, strojnega učenja, razvrščanja kompleksnih vzorcev in metode določitve značilk. Predstavljeno je delovanje nekaterih najpogosteje uporabljenih razvrščevalnih algoritmov. Izdelan je bil algoritem za zaznavo Parkinsonove bolezni na podlagi zajetega zvočnega signala. Meritve zvoka so bile narejene na štiridesetih posameznikih. Od tega je bila polovica zdravih in polovica z Parkinsonovo boleznijo. Namen naloge je razviti robusten sistem za zaznavo prisotnosti Parkinsonove bolezni. Za izboljšanje natančnosti razvrščanja, so bile uporabljene različne tehnike določitve značilk (Pearsonov korelacijski koeficient, Khendallov korelacijski koeficient in Samoorganizacijske gruče) in topologije nevronskih mrež. S pomočjo usmerjene nevronske mreže, je bila dosežena 86,47 % natančnost razvrščanja. Omenjena natančnost je bila dosežena z uporabo redukcije značilk na podlagi Pearsonovega korelacijskega koeficienta.

Keywords

umetna inteligenca;klasifikacija;strojno učenje;Parkinsonova bolezen;umetna nevronska mreža;magistrska dela;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FS - Faculty of Mechanical Engineering
Publisher: [L. Berus]
UDC: 004.923.021(043.2)
COBISS: 21149974 Link will open in a new window
Views: 1406
Downloads: 229
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Classification of patterns with use of intelligent methods
Secondary abstract: This Master’s thesis discusses artificial intelligence, machine learning, classification of complex patterns and feature selection procedure. Some of the most used classification algorithms are introduced. Algorithm for the detection of Parkinson’s disease based on sound measures has been made. Sound measurements of forty individuals were used as a dataset. Half of the individuals are healthy and half have the Parkinson’s disease. Purpose of this thesis is to present robust system for Parkinson’s disease detection. Few different feature selection techniques (Pearson’s correlation coefficient, Khendall’s correlation coefficient and Self-organizing maps) and neural network topologies have been used for improving classification accuracy. With the use of feed-forward neural network 86,47 % accuracy was achieved based on Pearson’s correlation coefficient.
Secondary keywords: artificial intelligence;classification;machine learning;Parkinson's disease;artificial neural network;
URN: URN:SI:UM:
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za strojništvo, Računalniško inženirsko modeliranje
Pages: X, 49 f.
ID: 10859777
Recommended works:
, delo diplomskega seminarja
, s Pythonom do prvega klasifikatorja