Paper Key : IRJ************986
Author: Muskan Naik,Abha Choubey,Prageet Bajpai
Date Published: 16 Oct 2023
Abstract
The detection of cheaters can be tough because to their unexpected nature and the absence of identifiable samples, rendering them difficult to identify. Fraudulent individuals endeavor to enhance their own gains by using prior advancements in a manner that detrimentally affects others. The individuals in question did not achieve success in successfully completing many detour protection exams, leading to a significant financial deficit amounting to millions of dollars. One approach to monitor fraudulent transactions is the utilization of statistical mining methods, aimed at analyzing unexpected physical transactions and identifying them as instances of deceit. The objective of this study is to conduct a comparative analysis of various artificial intelligence (AI) methodologies, including OK-closest neighbor (KNN), abnormal timberland, and backing vector machines (SVM). Additionally, this research aims to develop a comprehensive comprehension of AI techniques such as auto encoders, convolutional neural networks (CNN), restricted Boltzmann machines (RBM), and deep belief networks. The topic of discussion is exchanges, specifically referring to the concept of decentralized business networks (DBN). This application will utilize datasets from the European Union (EU), Australia, and Germany. There exist three distinct assessment measures that might potentially be employed. Three often used evaluation metrics in machine learning are the Area Under the Receiver Operating Characteristic Curve (AUC), the Matthews Correlation Coefficient (MCC), and the Cost of Error. This study presents an examination of the diverse targeting strategies employed
DOI LINK : 10.56726/IRJMETS45256 https://www.doi.org/10.56726/IRJMETS45256