Performance evaluation of a fraud detection system based artificial immune system on the cloud

Fraud is defined as the unlawful and intentional misrepresentation which can lead to actual or potential disadvantage to another individual or group. Fraud detection is a topic applicable to many industries including banking and financial sectors, insurance, law enforcement, and more. Credit card fraud is a major problem in the financial industry 111. Therefore, real time fraud detection is a vital issue. Indeed, we implement the progress of Negative Selection Algorithm an anomaly detection approach in AIS on the cloud. As a result of our experiments, in serial NSA the time of training phase is around 23800s whereas parallel NSA training phase time is 78s. Also, in parallel algorithm the detection rate increased around %50 compared to the serial algorithm. But we concluded false positive rate a little raised that compared to increase the detection rate is almost negligible. Tests are done with real data sets, and all executions are run using mapreduce and apache hadoop.