Advanced Multimodal Artificial Intelligence System for Transforming Digital Banking Operations

Rapid development in the field of digital banking systems has brought to light a lot of problems related to security, fraud detection, and effective finance management. The currently available approaches are based on the use of the existing solutions, which cannot address the modern trends in the field of cyber threats and fraudulent attacks. The purpose of this project is to develop a reliable transaction system with the aid of authentication and machine learning-based fraud detection. PIN identification and facial recognition will be used to ensure the authentication of the user before executing any transaction. To run the experiment, 50,000 synthetic transactions characterized by the value of each transaction, its time interval, and transaction frequency will be created. Anomaly detection is performed using an isolation forest machine learning algorithm through Scikit-learn to detect abnormal transactions without having to label them. The overall architecture of the system comprises a user module, an admin module, and MySQL database backend for storage and monitoring purposes.