Against “Fraudulent Transaction” with Deep Neural Network
As an E-Commerce we have some issues about fraudulent transaction. We see that there are two kind cases fraudulent transaction in online transaction. First, someone who do fraud by stealing other people's accounts so they can doing transaction using people’s money and it against the law. Second, someone who do abusing the online system, in order for the system or its business rivals can not perform transactions such as lock the item stock or exploit the vulnerabilities such as abusing the discounts or vouchers. The solution before is to analyzing the fraudulant pattern then use the expert system to create the fraud detection system (FDS). Today machine intelligence make the FDS can improving contiunously and still update with the possible pattern that can come in the future. Deep learning is a very promising technology that can be implemented to be solution for fraudulent transaction problem.
Senior Software Engineer at blibli.com (ecommerce in Indonesia), working in Research and Development division for Intelligence System. Develop fraud detection system for B2C ecommerce using machine learning especially deep learning. He received his B.S from Indonesian Computer University (UNIKOM), and his M.Sc from Bandung Instute of Technology (ITB), Indonesia. His research for Master’s theses is performance comparisson between standard backpropagation and deep neural network for repriceing gap prediction of conventional bank in Indonesia. Currently research in machine learning for several problems such as fraud and abusing detection system, and to understand the profile and the behaviour of the users.