Detection of Phishing Emails using Support Vector Classifier and Gaussian Latent Variable Model

Abstract

The explosive growth of internet technologies has transformed online interactions, but it has also increased security risks. Phishing is a serious problem that involves the surreptitious use technology and social engineering to steal sensitive information and account information from innocent victims. This paper tackles the urgent need for advanced and adaptive methods to identify phishing emails, particularly in light of the fact that phishing assaults are still evolving and continue to represent serious risks to individuals and businesses. By merging the Gaussian Latent Variable Model (GLVM) and the Support Vector Classifier (SVC), we suggest a novel hybrid phishing email detection method. The GLVM improves data comprehension and detection efficiency while the SVC gives strong classification abilities. To evaluate the effectiveness of our proposed approach, we conducted experiments using a comprehensive dataset comprising both legitimate and phishing emails. The strength of our model in correctly detecting phishing emails while limiting false positives and false negatives is highlighted by our method's impressive performance metrics, which show an accuracy of 98.43%, precision of 99.28%, and recall of 99.16%. The obtained performance metrics highlight the possibility for enhancing phishing email detection by combining the SVC and GLVM algorithms. Our method helps in advancing the ongoing efforts to improve cybersecurity measures and protect people and organizations from phishing attempts by utilizing the power of machine learning and dimensionality reduction. This method can be improved upon and expanded upon in future study to address the development of phishing tactics.