Advanced Machine Learning Models for Anti-Money Laundering (AML): Improving Detection Accuracy and Efficiency

Financial institutions and economic governance are greatly endangered by the transnational issue of money laundering. Typical AML solutions fail to identify chronic money laundering techniques because of the rigid IT rules-based mechanisms. The purpose of this study is to determine how AML systems might increase detection accuracy and efficiency through improved ML models. The Elliptic dataset, which contains 203,769 Bitcoin transactions with 166 characteristics, is used to assess sophisticated ML models for AML detection in this work. Such phases are data pre-processing, the decision of which features are most informative and model validation through F1-score, recall, accuracy and precision. After comparing different models, the best result was obtained by Multi-Layer Perceptron (MLP) with 98% accuracy, 99.5% recall, and 98.4% F1-score, which proves this model’s ability to differentiate between illicit transactions correctly and avoid false positives. However, effective methods are still needed to solve certain challenges like the dataset imbalance and how such an approach can be implemented in real-time.