Machine Learning Models for Money Laundering Detection in Financial Institutions. A Systematic Literature Review (#1682)

Authors

Soria Quijaite, Juan Jesús

Segura Peña, Lidia Victoria

Loayza Abal, Rodrigo Ireneo

Abstract

Financial crimes in institutions have grown exponentially over the years, detecting credit card fraud in which simple and hybrid machine learning have been used for detection. In the world of financial transactions, the development of predictive models in the detection of financial fraud has become a fundamental element for the success of a secure transaction in banking organizations; in this sense, the study aimed to systematize research with machine learning models in the detection of money laundering in financial organizations, the methodological design used was theoretical systematic review, the search explored two databases following the PRISMA statement (Scopus, Web of Science), 189 articles were found, of which, after the eligibility criteria, 25 were systematized. The results refer that work was done with Support Support Machine Models (SVM), Nearest Neighbors (KNN), Artificial Neural Networks (ANN), decision trees, Random Forests and Naive Bayes, which shows that the best accuracy in obtaining the laundering of assets was obtained by the SVM with an accuracy of 93.45%, in second place the Neural network with 92.14%; in the same way it was observed that Gezer, Ali et al. had the highest citation with 29, followed by Eachempati, Prajwal with 22 citations. It has been further revealed that money laundering affected many organizations engaged in being transactions in virtual form, in which artificial intelligence contributes in its support to detect this computer crime. These findings provide valuable information to improve the detection of financial fraud, highlighting the importance of addressing specific aspects that with the help of artificial intelligence can promote a better machine learning model that allows detecting suspicious transactions.