Publications & Research
Exploring the intersection of cybersecurity, machine learning, and explainable AI. Documenting frameworks designed to make security models both highly accurate and human-interpretable.
Research Papers
Explainable Machine Learning Framework for Phishing Website Detection
Raghav Dadhich, et al.
Phishing campaigns continue to evolve, exploiting sophisticated evasion techniques. Modern machine learning classifiers achieve high detection rates but operate as black boxes, making security analysts hesitant to trust their decisions. This research introduces a transparent, explainable machine learning (XAI) framework designed to identify phishing domains. By coupling robust tree-based models with interpretability modules (SHAP and LIME), we show the specific feature contributions (such as URL entropy, SSL validity, and domain age) that trigger detection, enabling security teams to verify predictions in real-time.
Research Interests
Explainable AI (XAI)
Developing tools to make machine learning models fully transparent, auditable, and interpretable.
Threat Detection
Applying advanced architectures to detect network intrusions, web-based threats, and phishing activities.
Model Robustness
Analyzing cybersecurity classifiers against adversarial perturbations and transferability attacks.