Strizhak A. Development of mathematical and software tools to minimize risks in payment systems

The dissertation is devoted to developing applied mathematics and specialized software that includes new mathematical models and algorithms for solving complex problems of security protection and minimization of risks in payment system; implementation of this models, algorithms and appropriate application software to improve efficiency of payment systems and reduce the losses of its members. The payment systems are analyzed as complex information, analytical geographically distributed systems; explored their architecture and different types of relationships between the constituent parts. Several new heterogeneous algorithms are proposed to identify suspicious transactions: the classification algorithm of fraud detection based on Kohonen maps, and algorithm of transaction's analysis based on expert rules specified in terms of fuzzy sets. This approach helps improve the overall accuracy assessment of transactions, reduce the number of false classifications. The architecture of specialized application software for monitoring transactions and fraud detection in payment systems is developed, which can be integrated with existing systems and software of the processing center. A software program is implemented and tested.