Active Learning and Autoencoders in Banking Fraud Detection

• Research unit: INNOSUISSE
• Project number: 34137.1 IP-ICT

Short description (German)

Active Learning and Autoencoders in Banking Fraud Detection

Short description (English)

Active Learning and Autoencoders in Banking Fraud Detection

Abstract (German)

This project aims to enhance the methods used to detect and understand fraud in the banking sector by using two innovative scientific methods. These are active learning to identify cases that are worth investigating and autoencoders, a novel machine-learning method for outlier detection.