SHARABLE DEVICE-AWARE PHISHING ATTACK DETECTION IN CLOUD ENVIRONMENTS USING VAE WITH CLUSTERING MECHANISM AND SVM

Garikipati, Venkat; Ubagaram, Charles; Dyavani, Narsing Rao; Jayaprakasam, Bhagath Singh; Mandala, Rohith Reddy; Ogunmola, Gabriel Ayodeji

doi:10.22059/ijms.2025.392673.677516

SHARABLE DEVICE-AWARE PHISHING ATTACK DETECTION IN CLOUD ENVIRONMENTS USING VAE WITH CLUSTERING MECHANISM AND SVM

Articles in Press

Document Type : SI: BDDEP-2026

Authors

Venkat Garikipati ¹

Charles Ubagaram ²

Narsing Rao Dyavani ³

Bhagath Singh Jayaprakasam ⁴

Rohith Reddy Mandala ⁵

Gabriel Ayodeji Ogunmola ⁶

¹ Innosoft, Sacramento

² Tata Consultancy Services, Ohio

³ Uber Technologies Inc

⁴ Cognizant Technology Solution

⁵ Tekzone Systems Inc

⁶ Associate Professor, Department of Economic Theory, Faculty of Economics, Tashkent State University of Economics, Uzbekistan.

10.22059/ijms.2025.392673.677516

Abstract

Phishing attacks pose a significant risk to cybersecurity and are especially troublesome in cloud-based settings as the risk is heightened due to shared and multi-device access. To counteract this, the current paper presents a Sharable Device-Aware Phishing Attack Detection system that integrates Variational Autoencoder (VAE) with a Clustering Mechanism and Support Vector Machine (SVM) to make the detection of phishing attacks more effective. The VAE is employed to perform feature extraction and to support the unsupervised learning of phishing behavior by giving the clustering mechanism the ability to group threats in a way that classification can be done with the SVM model being the one that classifies phishing cases accurately afterwards. The presented model is compared on a dataset of phishing behaviors harvested from cloud-based IoT environments, showcasing high performance in terms of detection accuracy, recall, and F1-score. Results show 99.5% accuracy, 99% precision, 98.5% recall, and 99.95% F1-score, better than the current phishing detection algorithms. The combination of device-aware learning with more sophisticated machine learning concepts offers an effective, scalable, and flexible phishing detection algorithm for cloud security.

Keywords

Phishing Attack Detection

Sharable Devices

Device-Aware Detection

Variational Autoencoders (VAE)

Clustering Mechanism

Subjects