Keynote Address:Analytic Techniques for Combating Financial Fraud

No ratings

Presented at Financial Cryptography and Data Security 2013 by

Scott Zoldi

Effective detection and prevention of financial transaction fraud requires analytic techniques capable of characterizing the transaction stream and determining the likelihood of fraud or misuse in real time. The nature of fraud makes the development of transaction monitoring models challenging, and inspires new Analytic innovations to respond to the ever changing tactics taken on by fraudsters. In this talk, we will review traditional payment card fraud detection analytic techniques such as transaction profiling and neural network models, and then expand to new analytic techniques. These new techniques include Adaptive Analytics which allows a model in production to adjust its scores dynamically based on analyst fraud case dispositions; Self-Calibrating Analytics which determines outlier fraud features in streaming transactions through real-time estimations of variable distributions; and Global Intelligent Profiles which enable profiling of various entities in the fraud decision such as ATMs, countries, and compromise clusters.