magistrsko delo
Lucija Brezočnik (Author), Vili Podgorelec (Mentor)

Abstract

V magistrskem delu smo razvili metodo FS-BPSO, ki združuje postopek izbire atributov z algoritmom optimizacije z rojem delcev. Glavni namen te metode je njena uporabnost pri reševanju naslednjega dobro znanega problema. Ko so v podatkovni množici primerki z ogromnim številom atributov, je med njimi težko najti tiste, ki so najbolj informativni oziroma reprezentativni. Tega problema smo se lotili s predlaganim hibridnim algoritmom binarne optimizacije z rojem delcev v kombinaciji s klasifikacijskimi metodami C4.5, Naive Bayes in SVM v cenitveni funkciji za izbiro informativnih atributov. Dobljeni rezultati so statistično analizirani in razkrivajo, da predlagani hibridni algoritem prekaša znane klasifikacijske metode C4.5, Naive Bayes in SVM.

Keywords

računalniška inteligenca;umetna inteligenca;optimizacija z rojem delcev;klasifikacija;magistrske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: L. Brezočnik
UDC: 004.89(043.2)
COBISS: 20148502 Link will open in a new window
Views: 1150
Downloads: 207
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: PARTICLE SWARM OPTIMIZATION IN FEATURE SELECTION FOR CLASSIFICATION
Secondary abstract: In this master's thesis, we have developed an FS-BPSO method that joins a feature selection procedure with a particle swarm optimization algorithm. The main purpose of this method is its usability in addressing the following well-known problem: When there are instances with a huge number of attributes in a data set, it is hard to select the most representative ones among them. In order to cope with this problem, we propose the use of a hybrid binary particle swarm optimization algorithm combined with C4.5, Naive Bayes, and SVM as the classifiers in the fitness function for the selection of informative attributes. The results obtained were statistically analysed and revealed that the proposed algorithm outperformed known classifiers, e.g. C4.5, Naive Bayes, and SVM.
Secondary keywords: computational intelligence;particle swarm optimization;feature selection;classification;
URN: URN:SI:UM:
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Informatika in tehnologije komuniciranja
Pages: X, 95 f.
ID: 9160761