PSO-ACO based ANN Approach for Credit Card Fraud Detection

Credit card fraud is a major problem for both consumers and institutions in the ever-changing financial environment of today. Credit card fraud detection is a significant challenge in financial security, and a novel approach is proposed to enhance its accuracy. This work intends to detect fraudulent transactions on credit cards by use of PSO along with ACO in combination. The Hybrid PSO-ACO based ANN model uses PSO and ACO to refine the training process, resulting in improved classification performance. PSO optimizes the network's epoch settings and batch processing, while ACO fine-tunes batch selections. Experiments on two credit card transaction datasets show that the Hybrid PSO-ACO based ANN outperforms conventional ANN models and other optimization-based ANN approaches in terms of accuracy, sensitivity and specificity. The proposed model improves generalizing to fresh data, reduces overfitting, and balances minority class data. This work highlights the potential of combining multiple optimization techniques to advance fraud detection capabilities and provides a robust framework for future research