Abstract :

The Rapid Evolution Of Phishing Attacks Poses A Serious Challenge To Conventional Cybersecurity Defenses, Particularly Systems That Rely On Static Rules Or Excessive Feature Dependencies. This Work Presents An Enhanced Phishing Detection Framework That Emphasizes Feature Efficiency, Adaptive Learning, And Holistic Performance Evaluation. The Proposed Approach Integrates An Optimized Feature Selection Strategy With Multiple Deep Learning Architectures To Accurately Distinguish Phishing URLs From Legitimate Ones While Minimizing Computational Overhead. Instead Of Relying On A Large Feature Pool, The System Systematically Identifies A Compact And Highly Discriminative Subset Of URL-based Attributes Using Permutation-based Importance Analysis, Ensuring Both Robustness And Real-time Applicability. To Strengthen Detection Capability, Multiple Learning Models—including Feedforward Neural Networks, Deep Neural Networks, Wide-and-Deep Architectures, And TabNet—are Trained And Evaluated Using Stratified Crossvalidation. A Novel Anti-phishing Score Is Introduced To Provide A Comprehensive Assessment By Jointly Considering Accuracy, False Positive Rate, True Positive Rate, And Testing Time, Thereby Addressing Limitations Of Single-metric Evaluations. Experimental Results Demonstrate That The Proposed Framework Achieves Superior Detection Performance With Reduced Latency, Making It Suitable For Deployment In Practical Environments. Validation On An Independent Dataset Further Confirms The Generalization Ability Of The Model Against Evolving Phishing Patterns. Overall, This Research Contributes A Scalable, Efficient, And Performance-aware Phishing Detection Mechanism That Enhances Cybersecurity Defenses Against Modern Web-based Threats. Index Terms— Phishing Detection, Cybersecurity, Feature Selection, Deep Learning, Feedforward Neural Networks, URL-based Analysis, Permutation Importance, Antiphishing Score, Real-time Threat Detection, Machine Learning Optimization.

Published:

20-2-2026

Issue:

Vol. 26 No. 2 (2026)

Page Nos:

70-74

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite