Ταξινόμηση με χρήση αλγορίθμων data mining και ασαφούς λογικής σε δεδομένα σημάτων αστέρων νετρονίων
Classification using data mining algorithms and fuzzy logic on neutron star signal data
View/
Keywords
Μηχανική μάθηση ; Ασαφής λογική ; Ταξινόμηση ; Machine learning ; Fuzzy logic ; ClassificationAbstract
This study investigates the performance of both classical and fuzzy classification algorithms in predicting the authenticity of pulsar signals, using a widely accepted feature set proposed as a standard reference for future screening research. Six models were trained and evaluated, including four classical machine learning algorithms (Logistic Regression, Random Forest, SVM and XGBoost), as well as two custom fuzzy logic classifiers (Fuzzy k-NN and Fuzzy Decision Tree). The fuzzy models incorporate fuzzification mechanisms and rule extraction procedures to enhance interpretability. Results demonstrate that the classical models achieve high accuracy and robustness, with Random Forest and SVM performing particularly well. However, the fuzzy approaches offer meaningful advantages in decision transparency, making them useful in domains where interpretability is critical. The study also confirms the suitability of the selected features for reliable pulsar signal screening and suggests future research directions to further assess their adequacy. The study was implemented using the appropriate libraries of the Python programming language.
Postgraduate Studies Programme
Πληροφοριακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
55Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα