Aristotle University Of Thessaloniki (AUTH)
Μεθοδολογίες ανάπτυξης ασαφών συστημάτων ταξινόμησης με χρήση εξελικτικών αλγορίθμων: εφαρμογή σε προβλήματα υψηλής διαστατικότητας
Abstract
dc:descriptionThis doctoral dissertation deals with the general problem of classification using fuzzy logic techniquesand evolutionary algorithms. The main objective is to develop methodologies that efficientlyconstruct interpretable classification models for solving problems characterized by highdimensionality of the feature space, while minimizing the structural complexity of the derivedclassifiers. To this end, five new methodologies for developing genetic fuzzy rule-based classifiersare proposed, which address effectively a wide range of real-world classification problems. Thesemethodologies are complementary. Each displays comparative advantages in certain classes ofproblems, depending on the input space’s dimensionality and the desired balance between classificationaccuracy and structural complexity. A particular aribute of the proposed systems is thecreation of simple and understandable classification models, which are made up from rules thatresemble the simple deductive reasoning of humans. The minimization of the classifiers’ structuralcomplexity is accomplished through the progressive decomposition of complex optimizationtasks into separate subproblems, the sequential solution of which is ultimately easier than solvingthe original problems. Furthermore, the training algorithm evaluates deterministic information regardingeach feature’s suitability throughout the derivation of the final classification model, thusenabling the effective treatment of highly dimensional classification problems. The proposed classifiersare primarily applied to land cover classification tasks, using both multispectral and hyperspectralsatellite imagery. Within this framework, higher-order features are extracted throughthe wavelet transform, which increase the class separability in remote sensing classification tasksusing hyperspectral data. Finally, the proposed systems are also applied to modern biomedicalclassification problems, which comprise up to tens of thousands of available features.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2013
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Stavrakoudis, Dimitris
- Σταυρακούδης, Δημήτριος
Subjects
dc:subject × 22- Ασαφείς ταξινομητές
- Εξελικτικοί αλγόριθμοι
- Γενετικά ασαφή συστήματα ταξινόμησης βασισμένα σε κανόνες
- Τηλεπισκόπηση
- Ταξινόμηση κάλυψης γης
- Υπερφασματικές δορυφορικές εικόνες
- Προβλήματα ταξινόμησης πολύ υψηλής διαστατικότητας
- Fuzzy classifiers
- Evolutionary algorithms
- Genetic fuzzy rule-based classification systems
- Remote sensing
- Land cover classification
- Hyperspectral satellite imagery
- Highly-dimensional classification tasks
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Natural Sciences
- Computer and Information Sciences
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
dc:rights- Statement dc:rights
-
- BY
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/39930