Fraud Detection and Prevention: A Synopsis of Artificial Intelligence Intervention in Financial Services Smart Card Systems

Abstract

Due to the rapid advancement in electronic commerce technology, credit and debit cards theft have increased. As credit/debit card becomes the most popular mode of payment for both online and regular purchase in Nigeria, cases of fraud keeps increasing. Financial fraud has increased significantly with the development of internet technology and the global superhighways of communication, resulting in the loss of billions of dollars worldwide each year. These fraudulent transactions are scattered with genuine transactions and simple pattern matching techniques are not often sufficient to detect these frauds accurately.

Implementation of efficient fraud detection systems has thus become imperative for all credit and debit card issuing banks in order to minimize their losses. Many modern techniques have been suggested such as Artificial Intelligence, Data mining, Neural Network, Bayesian Network, Fuzzy logic, Artificial Immune System, K- Nearest Neighbour algorithm, Support Vector Machine, Decision Tree, Fuzzy Logic Based System, Machine learning, Sequence Alignment, Genetic Programming etc., have evolved in detecting various debit card fraudulent transactions.

By developing a system that utilizes machine learning to determine the similarity in trends of financial institution customers through three stages; preprocessing, principal component analysis, and recognition, banks can be better fortified for securing their customers details and preventing fraudulent transactions on their portals. Through machine learning preprocessing, character traits and attributes are stored in a database. The principal component analysis is applied to find the accurate access characteristics of an account user which are important for identification. The naïve Bayesian Approach is then used to calculate the likelihood of an account owner accessing his account without making mistakes. Based on these trends, a data mining pattern is drawn in verifying any account holder; this ensures that infiltrators can be easily identified by the artificial intelligence system used to monitor such accounts. A neural network is used to create the log-in database, and recognize and authenticate the correct account users by using these weights. In this work, a separate Bayesian network is developed for each account holder. The input log-in details are mapped to a particular computer system’s MAC address with provisions for modifications after due verification that the account is owned by the correct account accessor. Each logged-in detail is used as input to each account owner’s network while cataloguing the accesses. The one with maximum output is selected and reported as the host if it passes a predefined recognition threshold. The algorithms that have been developed are tested on the developed Artificial Intelligence engine.

Testing the designed AI engine shows that fraudsters will find it harder breaking into the account of bank customers while the engine is active hence an effective anti-fraud and intervention tool.

Keywords: machine Learning, fraud detection, fraud prevention, artificial inelligence

JEL Classification: C11, M15