An Empirical Investigation of Phishing Detection Through Machine Learning Methods

An Empirical Investigation of Phishing Detection Through Machine Learning Methods

Pandya Himani1), Dr. Khyati Zalawadia2), Dr. Harish Prajapati3)

1) Department of Computer Engineering, Parul Institute of Engineering and Technology; Parul University; Vadodara; Gujarat e-mail: himanipandya31@gmail.com

2) Department of Computer Engineering, Parul Institute of Engineering and Technology; Parul University; Vadodara; Gujarat e-mail: Khyati.Zalawadia29490@paruluniversity.ac.in

3) Department of Computer Engineering, Parul Institute of Engineering and Technology; Parul University; Vadodara; Gujarat e-mail: harish.prajapati35068@paruluniversity.ac.in

          Because of the increasing likelihood of deceptive attacks meant to steal sensitive data, phishing threat detection is essential to cybersecurity. The application of machine learning techniques to identify phishing attempts is described in this synopsis. To detect phishing traits, the procedure involves removing aspects from websites, email content, and URLs. To increase accuracy, methods such as OCR, NLP, and behavioural analysis are used.For classification, a variety of machine learning techniques are utilized, such as supervised learning (Logistic Regression, Decision Trees, SVMs), deep learning (CNNs, RNNs, Transformers), and ensemble approaches (GBM, XGBoost). By spotting odd patterns, unsupervised learning techniques like anomaly detection and clustering increase security.During implementation, data must be gathered and pre-processed, features must be engineered, models must be trained, and a real-time detection system that learns from fresh data must be put into place.

Keywords: Machine learning, Cyber-attack Phishing detection, RNN