Phishing Detection via Hybrid Machine Learning: Leveraging URL characteristics for Enhance Security

Phishing Detection via Hybrid Machine Learning: Leveraging URL characteristics for Enhance Security

Harshad Sonule, Siddhant Vibhute, Yash Wankhade, Dhiraj Varma

Prof. Pratima Bharati

Harshadsonule0703@gmail.com, siddhantvibhute123@gmail.com,

yashwankhade145@gmail.com, dhirajv474@gmail.com

Abstract
Phishing is becoming among the most common types of cybercrime, with hackers using ever-more-advanced techniques to trick victims into divulging private and financial data. Blacklist-based techniques and other conventional phishing detection tools frequently fail to detect emerging threats. By combining cutting-edge machine learning techniques with extracted features from URLs, this research suggests a sophisticated hybrid machine learning method for identifying phishing URLs.

The technology efficiently separates phishing URLs from authentic ones by examining important characteristics including URL length, character composition, domain reputation, and the usage of dubious keywords. To increase the accuracy of classification, we combine deep learning approaches like Convolutional Neural Networks (CNN) with supervised training methods like Random Forest and Support Vector Machines.

The technology efficiently separates phishing URLs from authentic ones by examining important characteristics including URL length, character composition, domain reputation, and the usage of dubious keywords. To improve the accuracy of classification, we combine deep learning algorithms like Convolutional Neural Networks (CNN) using supervised method of learning like Random Forest and Support Vector Machines.

Keywords
Phishing prevention, real-time detection, feature extraction, cybersecurity, deep learning, Random Forest, SVM, Convolutional Neural Networks, URL analysis, hybrid machine learning, and phishing detection.