Unsupervised learning models in credit card fraud detection

A joint seminar of two structural divisions was held at the Institute of Information Technology under the Ministry of Science and Education. At the seminar, Eltun Ahmadov, an employee of the institute, gave a presentation titled "Unsupervised learning models in anomaly-based detection of credit card fraud."

E. Ahmadov provided a comprehensive analysis of the rapid development of e-commerce and its impact on financial security, including the rise in credit card fraud and the cybersecurity measures being implemented in this sector. The speaker noted that credit card fraud is causing significant economic losses both globally and in Azerbaijan. He also touched upon issues related to the legal regulation of cybersecurity in our country.

Highlighting the main challenges of credit card fraud, the speaker informed the audience about key shortcomings such as imbalanced data, real-time detection, similarity between normal and fraudulent behavior, data accessibility, privacy and ethical concerns, resource limitations, computational complexity, limited evaluation metrics, and others.

The researcher drew attention to experiments conducted in the detection of fraudulent cases using unsupervised learning models, including the Isolation Forest, Local Outlier Factor, Autoencoder, and Robust Covariance methods.

He stated that as a result of the proposed one-class classification approach, the Autoencoder model stood out with the highest performance, adding, "This result serves to ensure effectiveness and reliability in fraud detection systems."

Discussions were held relating to the presentation, and the questions of the seminar participants were answered.

Ramiz Aliguliyev, Head of Department of the Institute and corresponding member of ANAS, commented on the report. The scientist emphasized that the work carried out in this direction contributes to strengthening cybersecurity and increasing the defense potential of the national digital infrastructure.

© All rights reserved. Citing www.iсt.az is necessary when using news.