Spatio-Temporal Network-based Bank Transactional Behavior Analysis to Detect Suspicious Activities

Abstract

Money laundering refers to the act of disguising the proceeds of illegal activities as legitimate funds. This is a significant problem as it enables criminals to profit from illegal activities and finance further criminal endeavors. Money laundering is also linked to other crimes, such as drug trafficking, terrorism financing, and corruption. To combat money laundering, governments and financial institutions have implemented various measures, such as Know Your Customer (KYC) regulations, Anti-Money Laundering (AML) laws, and the use of financial intelligence units. However, the existing money laundering system is complex, making it difficult to detect and prevent money laundering activities. Many current systems rely on outdated technology and manual processes, which can be time-consuming and prone to error. Therefore, there is a need for effective detection and prevention systems that can identify suspicious transactions and patterns of behavior. This project aims to prevent and detect money laundering activities by identifying suspicious transactions and monitoring the movement of funds through the financial system. The proposed system uses Long Short-Term Memory (LSTM) to detect and prevent money laundering activities. By analyzing the data in a time-series format, LSTM can identify unusual patterns of transactions and flag them for further investigation. LSTM can also predict future trends in financial data, allowing for the detection and prevention of potential money laundering activities before they occur. The proposed transactional network and behavior analysis system provides a more efficient and accurate method for identifying potential money laundering activities. This system can ultimately lead to a more effective and efficient anti-money laundering system. Keywords: Anti-Money laundering, Long short-term memory, Legitimate funds, Transaction network, behavior analysis, Time series format