Ηλεκτρονική απάτη στον τραπεζικό τομέα
Fraud detection in banking
View/
Keywords
Ηλεκτρονικές απάτες ; Σύστημα ανίχνευσης απάτης ; Τεχνικές μηχανικής μάθησης ; Τεχνητή νοημοσύνηAbstract
Financial fraud in the banking sector is now a topic of major importance and worthy of study. The rapid
evolution of technology, combined with the digitalization of banking services and transactions, has had the
sharp consequence of the growth of cybercrime in cyberspace, thus affecting the economic activity of financial
institutions and affecting their solvency. Among the extensive volume of transactions that take place every
minute, detecting and anticipating illegal movements in a timely and swift manner is a matter of primary
importance, as it is a fact that financial fraud is carried out by constantly changing and evolving methods. This
has consequently led to the need for effective shielding of banking systems using modern machine learning
and artificial intelligence techniques, which would previously have seemed impossible. Daily efforts and tests
are being made by the banking sector, to create a security system which will be able to provide all possible
security to the bank and its clientele, and to create a leery public by information providers. In this thesis we
will analyze the existing fraud incidents that a modern financial institution is confronted with, the machine
learning models and techniques used to detect and predict them, as well as the challenges that arise for their
proper assessment and management. Then, for the programmatic Part 1 of the paper, we analyze real data
from credit card fraud cases, to which we apply machine learning algorithms which we then compared and
evaluated. For a better understanding of the topic, we have chosen for part 2, to design a fraud detection
system from scratch, inspired by real scenarios and challenges, which categorizes and detects fraudulent
movements within a financial institution.
The programming part and data processing were done using Python. The conclusions drawn from both
parts of the work showed the prediction accuracy of the algorithms, but at the same time provided us with
important information about the challenges and vulnerabilities that exist in a fraud prediction and detection
system.
Postgraduate Studies Programme
Κυβερνοασφάλεια και Επιστήμη ΔεδομένωνDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα ΠληροφορικήςNumber of pages
65Language
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