The paper is called “Credit Card Fraud Detection in e-Commerce: An Outlier Detection Approach”, and it not only recognises behaviour related to transactions and payments, but it also flags activities that fall outside of the expected norm.
The researchers leveraged techniques used to identify groups of similar objects in a dataset, with different parameters. Every data point was assigned to a cluster in each training run from which a mathematical representation (vector) was produced. This constitutes “fingerprints” of the data point that could be combined into a unique signature representation of it. Moreover, to generate a signature that represented “good behaviour”, the team combined the per-data point vectors and weighed them by the size of the respective cluster, arriving at a single score between 0 and 1.
The approach is different than the conventional AI fraud detection, as it didn’t require prior knowledge of outliers or inliers. The underlying algorithm was both scalable and general in nature. Also, it could be applied to virtually any clustering problem, including those in the medical domains.
The team sourced data science platform Kaggle’s publicly available credit card database to test their method. The database contains 284,807 samples of credit card transactions made in September 2013 by European cardholders in two days, out of which 492 are fraudulent. After a total of 10 runs, the algorithm was able to identify 40% of fraud cases. It flagged 29 legitimate transactions out ofhundreds of thousands of data points at play.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now