Banksealer: Automatic Banking Fraud Detection
We started this project because we wanted to analyze banking and credit-card transactions and, with as little knowledge as possible, predict whether new ones are fraudulent or not (e.g., due to a banking trojan working on the victim’s computer, made by a cyber criminal with stolen credentials).
BankSealer is based on lightweight statistical learning on feature models (e.g., amount, timestamp, recipient country, description) extracted by each transaction. BankSealer is currently deployed at one of the largest Italian banks and has been proven effective at detecting frauds, to the point that my co-authors have created a startup out of it!
Banksealer is now a fintech startup: https://banksealer.com
References
(2016).
BankSealer: An Online Banking Fraud Analysis and Decision Support System.
ICT Systems Security and Privacy Protection.
(2014).