Improved fraud detection in e-commerce transactions

Abstract

Online transactions have gained popularity in the recent years with an impact of increasing fraud cases associated with it. Fraud increases as new technologies and weaknesses are found, resulting in tremendous losses each year. Since the transactions associated with e-commerce are large in number, the dataset associated with them is also large; therefore, it requires fast and efficient algorithms to identify fraudulent transactions. Most of the methods used for fraud detection are rule-based or are systems that require re-training when newer patterns of fraud occurs. Detecting fraud as it is happening or within a short time span is not easy and requires advanced techniques. As the demand has arisen for self-learning predictive systems, the main objective is to detect the fraudulent transactions by using Adaptive Neuro-Fuzzy Inference System, which is a hybrid of neural networks along with fuzzy inference, wherein the system can adapt to newer instances of fraud.