Προσομοίωση πολυδιάστατων δεδομένων κίνησης με τεχνικές μηχανικής μάθησης
Simulation of multiple-aspect trajectories with machine learning methods
View/
Keywords
Machine learning ; Synthetic Trajectory Generation ; Generative Adversarial Networks (GANs)Abstract
The rapid growth in the collection and utilization of trajectory data, such as data derived from GPS, sensors, and mobile applications, has created the need for the generation of synthetic yet realistic data that represent human or object mobility across space and time.
This diploma thesis focuses on the study and experimental evaluation of machine learning techniques for the simulation of multidimensional, semantically enriched movement data. More specifically, it examines methods that aim to generate synthetic trajectory data with realistic characteristics, while preserving both spatial and temporal coherence as well as the semantic properties that often accompany such data, such as entity type, activities, or points of interest.
The thesis emphasizes the efficiency, accuracy, and generalizability of simulation models, with the objective of supporting research scenarios, testing procedures, and mobility analysis without relying on real, sensitive, or hard-to-obtain trajectory data.
Postgraduate Studies Programme
Εφαρμοσμένη ΣτατιστικήDepartment
Σχολή Χρηματοοικονομικής και Στατιστικής. Τμήμα Στατιστικής και Ασφαλιστικής ΕπιστήμηςNumber of pages
123Language
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