Publications

Proposes a privacy-first machine unlearning approach for medical imaging, allowing TB detection models to forget patient data on demand while preserving diagnostic performance.

FIDES addresses increasingly sophisticated financial fraud using Federated Deep Learning with CV-QKD-secured gradient transmission across non-IID client data. FedDyn aggregation achieves 97.74% accuracy with >250,000 bits/sec key rate and >98% secure key ratio, providing information-theoretically strong privacy guarantees for inter-institution model training.

Combines explainable AI techniques with quantum cryptographic primitives for classifying and interpreting network attacks in smart home IoT environments.

Develops a deep learning pipeline with integrated SHAP-based explainability for detecting false data injection attacks in autonomous vehicle sensor networks, providing both accuracy and interpretable model decisions.

Introduces a continuous-variable quantum key distribution framework for securing V2X communications in autonomous vehicle networks against eavesdropping and quantum-level adversarial attacks.

Presents a quantum-assisted edge computing framework for real-time anomaly detection and secure data processing in IoT-enabled healthcare environments.

Q-PhishNet is a phishing-detection framework for IoT networks that combines deep learning, explainable AI, machine unlearning, and quantum cryptography. It uses deep learning for phishing URL classification and LIME for interpreting predictions. To support data revocation and counter poisoned data, it integrates three machine-unlearning techniques: KL Adversarial Unlearning, Fisher Regularization, and Reverse Distillation. The BBM92 quantum key distribution protocol secures data transmission during the unlearning process.