Publications

This study aims to fill the gap of insufficient datasets for cybersecurity in Internet of Medical Things (IoMT) ecosystem through data augmentation and their effects on intrusion detection systems performance. To develop augmented datasets with varying label distributions, the research utilizes four augmentation methods: Rule-Based, Tabular Variational Autoencoder (TVAE), Conditional Tabular Generative Adversarial Network (CT_GAN), and Gaussian Copula to generate synthetic datasets based on existing ECU-IoHT and WUSTL-EHMS datasets.

Presents a comprehensive overview of the Cyber Kill Chain (CKC), highlighting the role Artificial Intelligence plays across each phase in terms of offensive and defensive cybersecurity operations. A comparative analysis of 3 cybersecurity frameworks, with justifications for each, was also examined. Drawing on real-world case studies and recent literature, this study further highlights current challenges with the fusion of AI into cybersecurity operations, ranging from data privacy, adversarial attacks, and AI explainability. The review concludes by advocating for the adaptation of dynamic, AI-driven modelling frameworks that better align with the rapidly evolving cyber threat landscape.

Addresses the model utility vs. privacy tradeoff in privacy-preserving Vertical Federated Learning (VFL) by introducing an improved defense mechanism that combines Adversarial Training (to harden the soft labels) and Differential Privacy (aimed at restoring label privacy) to collectively enhance the robustness of the existing KD defense mechanism with marginal model utility trade-off.

Researchers have created several expert systems over the years to predict heart disease early and assist cardiologists to enhance the diagnosis process. We present a diagnostic system in this paper that utilizes an optimized XGBoost (Extreme Gradient Boosting) classifier to predict heart disease.

The intentional dissemination of false information, known as fake news, aims to manipulate readers into accepting biased or untrue beliefs by altering their interpretation and response to real news. However, identifying fake news is a tedious task, especially on platforms like Twitter where information is rapidly disseminated. This AI for Social Good research leverages Machine Learning algorithms to detect COVID-19 fake news on Twitter.