Προσέγγιση δεικτών βάσεων δεδομένων με μηχανική μάθηση

• Κύριος συγγραφέας: Σπυρόπουλος, Σπυρίδων
• Άλλοι συγγραφείς: Σιούτας, Σπυρίδων
• Μορφή: Thesis
• Γλώσσα: Greek
• Έκδοση: 2020
• Θέματα: Μεγάλου όγκου δεδομένα; Κατηγοριοποίηση; Νευρωνικά δίκτυα; Μηχανική μάθηση; Big data; Classification; Neural networks; Machine learning

For this thesis, neural network models have been developed to experiment with classification on numerical data. The architectures used these years give the opportunity to manage large volumes of data and export correlations. Every minute, the number of data is constantly increasing and the target is to make better use of it for various problems. Due to the large amount of data it is imperative to find data with similar characteristics. In this thesis a combinational analysis technique based on a "balls in bins" game for the purpose of classification is implemented. Classification combined with machine learning techniques enables prediction. This would not be possible without the proper computing tools and libraries that make the task of the developer much easier. The work is divided into two parts. The bibliographical section presents the theoretical documentation of machine learning techniques as well as a general overview of the main knowledge studied for the writing of this work. The experimental section attempts to implement a technique, combinational analysis for data processing, and three neural networks for predicting data categories.