This thesis analyzes the concept of Anomaly Detection, understood as the identification of anomalies, through the use of machine learning techniques to detect events that deviate from expected and predefined behaviors. The document begins with an extensive introduction to the topic, presenting the various approaches and algorithms employed in the field, and highlighting their differences, advantages, and limitations from both theoretical and computational perspectives. Particular attention is devoted to the DBSCAN method, an algorithm used to identify suspicious behaviors based on data density within a multidimensional space. Its underlying principles are explained and its performance is analyzed in comparison with other anomaly detection methods
Questo elaborato di tesi analizza il concetto di Anomaly Detection, inteso come rilevamento di anomalie, attraverso l’utilizzo di tecniche di machine learning per identificare eventi atipici rispetto a comportamenti attesi e prestabiliti. Il documento si apre con un’ampia introduzione sul tema, presentando i diversi approcci e gli algoritmi impiegati nel settore, evidenziandone differenze, vantaggi e limiti sia teorici sia computazionali. Viene inoltre approfondito il metodo DBSCAN, algoritmo utilizzato per rilevare comportamenti sospetti basandosi sulla densità dei dati in uno spazio multidimensionale. Ne vengono illustrati i principi di funzionamento e analizzate le prestazioni, mettendole a confronto con altri metodi di anomaly detection.
Anomaly Detection: analisi teorica e ricerca sul metodo DBSCAN
SANTORO, GIUSEPPE
2024/2025
Abstract
This thesis analyzes the concept of Anomaly Detection, understood as the identification of anomalies, through the use of machine learning techniques to detect events that deviate from expected and predefined behaviors. The document begins with an extensive introduction to the topic, presenting the various approaches and algorithms employed in the field, and highlighting their differences, advantages, and limitations from both theoretical and computational perspectives. Particular attention is devoted to the DBSCAN method, an algorithm used to identify suspicious behaviors based on data density within a multidimensional space. Its underlying principles are explained and its performance is analyzed in comparison with other anomaly detection methods| File | Dimensione | Formato | |
|---|---|---|---|
|
Santoro_Giuseppe.pdf
accesso aperto
Dimensione
2.69 MB
Formato
Adobe PDF
|
2.69 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/97853