Exploring the Effectiveness of Machine Learning Models in Detecting Anomalous Transactions

Abstract

Money laundering and illicit financial flows facilitate criminal operations and undermine economic stability. Cryptocurrencies present regulatory challenges due to their anonymity and decentralized nature. Anomalous transactions refer to financial transactions that deviate from established patterns, indicating potential fraud, errors, or unusual behavior. This paper reviews machine learning techniques for detecting anomalous cryptocurrency transactions from an anti-money laundering/counter-terrorist financing (AML/CFT) perspective. A real-world Bitcoin transaction dataset is analyzed for our study. The paper assesses how well various machine learning models perform in detecting anomalous transactions. Detecting these anomalies is important in preventing fraud in areas like banking, e-commerce, and financial services.