Přístupnostní navigace
E-application
Search Search Close
Master's Thesis
Author of thesis: Ing. Tomáš Kuběna
Acad. year: 2025/2026
Supervisor: Ing. Tomáš Marada, Ph.D.
Reviewer: Ing. Daniel Zuth, Ph.D.
This thesis deals with the implementation and comparison of artificial intelligence methods for optimizing the parameters of a fuzzy classifier. In testing preprocessing methods, normalization achieved the highest accuracy of classification, but the use of principal component analysis reduced computation time by 90% with a loss of accuracy by few percent. Cluster analysis was used for the initial parameter settings, where the Fuzzy C-means method proved to be more successful and was subsequently used for optimization. The main part of the work is a comparison of the genetic algorithm (GA) and particle swarm optimization (PSO). Comparison results on the Iris, Wine, Dry Bean and Vibro datasets showed that GA is more demanding in terms of proper configuration but achieves higher accuracy and optimization stability. PSO lags behind in speed and is more likely to get stuck in local minimum. When applied in technical practice, it is advisable to use principal component analysis to reduce computation time, the FCM method for initial parameter tuning, and then the genetic algorithm, which is more robust and has good repeatability. Beyond the scope of this work, an application was developed for configuring algorithms, running optimization, and visualizing results.
Fuzzy classifier, cluster analysis, principal component analysis, membership function, fitness function, genetic algorithm, particle swarm optimization.
Date of defence
16.06.2026
Result of the defence
Defended (thesis was successfully defended)
Grading
C
Process of defence
Prezentace: - Fuzzy systém - Vybrané datasety pro testování - Struktura algoritmu - Předzpracování dat - Shluková analýza - Metody optimalizace parametrů - Vývoj Fitness funkce - Metody optimalizace parametrů - Standalone aplikace - Zhodnocení a závěr Student seznámil členy komise s výsledky své práce a odpověděl na otázky oponenta. Otázky členů komise: - Proč byly zvoleny konkrétně tyto datasety? Zodpovězeno. - Jak lze využít výsledky práce? Zodpovězeno částečně. Student odpověděl na otázky členů komise s drobnými nedostatky.
Language of thesis
Czech
Faculty
Fakulta strojního inženýrství
Department
Institute of Production Machines, Systems and Robotics
Study programme
Production Machines, Systems and Robots (N-VSR-P)
Composition of Committee
Ing. Tomáš Marek, Ph.D. (člen) Ing. Jan Vetiška, Ph.D. (člen) Ing. Jan Vlček (člen) doc. Ing., Dipl.-Ing Michal Holub, Ph.D., FEng. (předseda) Ing. Rostislav Huzlík, Ph.D. (člen) doc. Ing. Petr Kolář, Ph.D. (místopředseda)
Supervisor’s reportIng. Tomáš Marada, Ph.D.
Grade proposed by supervisor: B
Reviewer’s reportIng. Daniel Zuth, Ph.D.
Grade proposed by reviewer: B
Responsibility: Mgr. et Mgr. Hana Odstrčilová