Credit Card Fraud Detection using Neural Embeddings and Radial Basis Network with a novel hybrid fruitfly-fireworks algorithm
Indu B Singh, N. Aditya, Pranjal Srivastava, Shivanshu Mittal, Taruvar Mittal, Noel Vaibhav Surin · 2023
Financial institutions are highly concerned about credit card fraud due to the growing frequency of deceitful activities linked with credit card transactions. To address this issue, a novel fraud detection system is proposed that combines advanced feature engineering and neural network architecture. The proposed approach. FFFW-RBN, employs a hybrid swarm intelligence model comprising the Fruitfly Optimization Algorithm and Fireworks Algorithm, (FFO-FWA) for feature selection, and the Neural Embeddings Generator in combination with a Shallow Radial Basis Network (RBF) for localized pattern classification. The proposed system achieves a 99.97% accuracy, 0.91 recall, and a precision of 0.94. The experimental results demonstrate that our methodology effectively and practically detects fraudulent credit card transactions. The paper presents a comprehensive evaluation of the system, including various performance metrics and a confusion matrix, highlighting its potential for real-world implementation.