PhishNetVAE Cybersecurity Approach: An Integrated Variational Autoencoder and Deep Neural Network Approach for Enhancing Cybersecurity Strategies by Detecting Phishing Attacks

Phishing attacks continue to rise in parallel with the growing demand for online services, posing a serious threat to individuals, organizations, and governments.To address this challenge, we propose a novel PhishNetVAE cybersecurity approach that combines Variational Autoencoder (VAE) and Deep Neural Network (DNN) in a unified framework for more robust phishing detection.The key idea behind PhishNetVAE is to leverage the generative capabilities of the VAE for extracting latent representations of URL patterns, which are then fed into a DNN classifier trained to distinguish phishing URLs from benign ones.This two-stage architecture harnesses the unsupervised feature learning of the VAE alongside the supervised discrimination of the DNN, ensuring that subtle phishing indicators are captured and classified accurately.To evaluate the effectiveness of PhishNetVAE, we conduct experiments using the ISCX-URL2016 dataset.The results indicate that the proposed method achieves a detection accuracy of 99.80% with a false positive rate of 0.18%, surpassing several recent approaches under comparable conditions.By highlighting the synergy between variational autoencoding for high-quality feature representation and deep neural classification, PhishNetVAE offers a novel and highly accurate cybersecurity solution for detecting phishing cyberattacks in diverse and rapidly evolving threat landscapes.