Experimental Performance Evaluation of an Enhanced Deep Neural Network Model for Phishing Intrusion Detection in Web-Enabled Financial Information Systems

Abstract

This study proposes an enhanced Deep Neural Network (Deep-NN) model for phishing intrusion detection in web-enabled financial information systems. The approach combines Principal Component Analysis (PCA) for feature optimization with both supervised and unsupervised learning techniques to improve detection accuracy and efficiency. Data pre-processing and dimensionality reduction were implemented in Python (Spyder IDE), while MATLAB was used for model training, validation, and testing. Performance was evaluated using regression (R), mean square error (MSE), and Jaccard similarity index. Results show a 90% classification precision with a 10% error rate, outperforming existing models in the literature. The findings demonstrate the model’s potential to strengthen cyber defense mechanisms in financial systems through robust and adaptive phishing detection.