Mr. Pramod Salunkhe
Engineering
September 2022
Frauds are said to be dynamic and devoid of patterns, making them challenging to detect. Fraudsters profit from recent technological improvements. Security precautions were overcome, causing a loss of millions of dollars. Using data mining techniques to monitor and spot unusual behavior is one way to track fraudulent transactions. Transactions. This article compares deep learning methods including auto encoders, convolutional neural networks, restricted Boltzmann machines, and deep belief networks to k-nearest neighbor (KNN), random forest, and support vector machines (SVM) (DBN). This study will make use of the European (EU), Australian, and German databases. The Area Under the ROC Curve (AUC), Matthews Correlation Coefficient (MCC), and Cost of Failure are the three assessment metrics that would be used.
274- 285
© Airo Journals 2021
