Hybrid Phishing Detection Using Machine Learning
Author : Dr. K Karuppasamy, Mohammed Sami A, Kali Dhanush S, Nimesh Y and Poovarasu T
Abstract :
The rapid expansion of digital communication technologies and online service platforms has significantly increased exposure to phishing attacks, a form of cybercrime in which malicious actors use deceptive URLs to manipulate users into revealing confidential information such as login credentials, financial data, and personal records. Traditional phishing detection methods primarily rely on blacklist databases, signature matching, and static rule-based filtering. While these approaches are effective against previously identified threats, they are inadequate for detecting newly generated phishing URLs, polymorphic attacks, and zero-day exploits that continuously evolve to bypass conventional security mechanisms.
To address these challenges, this research proposes a Quantum-Enhanced Intelligent Phishing URL Detection system that integrates machine learning-based classification with quantum-inspired optimization techniques to improve detection accuracy and system adaptability. The proposed framework extracts and analyzes multiple structural, lexical, and behavioral features of URLs, including domain characteristics, URL length patterns, token distribution, and anomaly indicators. These features are processed using an intelligent learning model capable of distinguishing between legitimate and malicious links in real time.
A quantum-inspired optimization approach is incorporated to enhance feature selection and model parameter tuning, enabling improved convergence efficiency and predictive performance. This hybrid methodology reduces false positive rates while maintaining high detection sensitivity, making the system more robust against previously unseen phishing strategies.
Experimental evaluation demonstrates that the proposed framework achieves superior classification accuracy, improved computational efficiency, and enhanced scalability compared to traditional detection approaches. By providing adaptive learning capability and real-time threat identification, the system contributes to the development of secure and resilient cybersecurity infrastructures. The proposed solution supports scalable deployment across modern digital environments and offers an effective defense mechanism against emerging web-based threats.
Keywords :
Phishing Detection, Machine Learning, Quantum-Inspired Optimization, Cyber Security, URL Classification.
