Companies in the financial industry are among those who are implementing
their operations online as a result of the internet's rapid growth in usage. Because of the
enormous economic damages that arise from financial fraud, it is becoming an
important concern as financial frauds are becoming more common and sophisticated
globally. An economic fraud detection system (FDS) must be able to identify risks such
as unusual assaults and unauthorized entry. The last few decades have seen a
widespread application of data mining and machine learning (ML) methods to address
this problem. These techniques still require improvement, though, to handle big data
quickly and recognize unidentified trends in attacks. The importance of data safety and
evaluation systems for big data has changed recently due to the enormous volume of
data and its continuous growth. Big data is defined as information that is difficult to
handle, store, and evaluate using standard software tools and databases. Big databases
exhibit considerable quantity, speed, and diversity, necessitating the development of
novel methods for handling them. Thus, employing blockchain and a deep learningbased (DL) approach for FDS is suggested in this work for credit card frauds. The
objective of this framework is to improve the accuracy of identification in the context
of big data in a private-permissioned blockchain network, while also improving the
existing detection methods. A present DL algorithm called the Auto-encoder algorithm and a few other ML algorithms are contrasted with the outcomes of the suggested
framework's evaluation using a real database of credit card frauds. According to the
study findings, the LSTM performed flawlessly, achieving 99.9% accuracy in roughly
a minute.
Keywords: Big data, Blockchain, Credit card frauds, Deep learning, Fraud detection system.
