Hybridization of SOM and PSO for Detecting Fraud in Credit Card

Hybridization of SOM and PSO for Detecting Fraud in Credit Card

Suman Arora, Dharminder Kumar
DOI: 10.4018/IJISSS.2017070102
(Individual Articles)
No Current Special Offers


Fraud Detection is a detection of criminal activity that generally occurs in commercial organization. Detection of such fraud can prevent a great economic loss. Credit card fraud depends upon usage of card, its unusual transactions behavior or any unauthorized activity on a credit card. Clustering process can divide the data into subsets and it can be very helpful in credit card fraud detection where outlier may be more interesting than common cases. Self-organizing Map (SOM) is unsupervised clustering technique which is very efficient and handling large and high dimensional dataset. Particle Swarm Optimization (PSO) is another stochastic optimization technique based on intelligent of swarms. In the present study, we combine these two methods and present a new hybrid approach self-organizing Particle Swarm Optimization (SOPSO) in detection of credit card fraud. In order to apply our method, we demonstrated an example and its results are compared with previous techniques. Some challenges shown in the previous researches such as time and space complexity, false positive rate and supervised techniques. Our approach is efficient as it implements one of the optimization technique and unsupervised approach which results in less time and space complexity and false positive rate is very low. Domain independency is also achieved in our approach.
Article Preview


E-commerce has become an important sales channel for worldwide company (Ifinedo, 2015). From the last few years credit card users, credit card issuers and online merchants are rapidly growing. Uses of credit card for buying any article also increased due to rapid development of e-commerce. With the growth in credit card transaction fraudulent activity of credit card fraud has also been increased. Sophisticated techniques are used by fraudsters to perpetrate credit card fraud. The fraudulent activities worldwide present unique challenges to banks and other financial institutions who issue credit cards. Economically, financial fraud is becoming an increasingly serious problem. In BBC news report 2007, cost of 1.6 billion pounds a year claims by the fraudulent insurance in UK. Researchers (Ngai, Hu, Wong, Chen, Sun, 2011) discussed that the financial fraud detection (FFD) is fundamental for the avoidance of the often-occurring financial fraud. The overall losses caused by financial fraud are incalculable. Fraud Detection refers to detecting a pattern that does not simulate with normal behavior. The broad categories of fraud can be classified as (Chandola, Banerjee, & Kumar, 2009): Offline Fraud is a type of fraud that is performed manually like by the help of stolen credit card, Online Frauds is performed electronically via internet, web or in absence of Credit Card Holder. Influence of Credit card fraud is least on card owners since they are limited liable to the transaction made. The existing enactment and card holder safety guidelines as well as insurance scheme in many countries look after the interest of card owner. However, the most affected the merchants, who, in most situations do not have any evidence to dispute the card holder claim of misused card information. Merchants, ends up bearing all the loses due to charge back, shipping cost of goods, card issuer fees, and charge as well their own administrative cost (Quah & Sriganesh, 2008). Excessive fraudulent cases involving the same merchant can drive away customer cause card issuer banks to withdraw services and also result in loss of reputation and good will. Card issuer bank have to bear the administrative cost, infrastructure cost of setting up the required software and hardware facilities to combat fraud. They also incur costs through transactions delays.

There are many types of techniques exists that can detect such types of frauds. Mainly three broad categories that exist in anomaly detection are (Wu & Banzhaf, 2010): Supervised Technique based on classification that is performed by the help of two steps termed as learning and classification. Semi-supervised Technique is performed by constructing a model of normal behavior and then tests the likelihood test instances. Unsupervised Technique is performed on unlabeled data set by assuming that majority of test cases are under normal behavior (Zhu, 2005).

Complete Article List

Search this Journal:
Volume 15: 1 Issue (2023): Forthcoming, Available for Pre-Order
Volume 14: 4 Issues (2022): 3 Released, 1 Forthcoming
Volume 13: 4 Issues (2021)
Volume 12: 4 Issues (2020)
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing