A Self-Organizing Map Model for Network Fraud Pattern Classification
Ikechukwu F. C. Onah, Inyiama H. Chibueze · SSRN Electronic Journal · 2012
As financial transactions continue to be processed electronically, advanced fraudbusting software solutions are needed to protect e-businesses from the constant problems of network-related frauds. Many popular fraud detection algorithms were used to adaptively profile legitimate customer behavior in a transaction database. This paper presents a Self-Organizing Map Neural Network (SOMNN) software agent based approach to detect such frauds and provides a means to gather relevant information about the nature of fraud that can be used for forensic analysis. The technology dynamically updates and educates itself each time it handles a new transaction. This enables operators to respond to fraud by detection, service denial and prosecutions against fraudulent users. The applicability of the approach has been demonstrated for a network fraud scenario caused by the Routing Information Protocol (RIP) attack.