Research Vision

Advancing reliable AI systems for recognition, vision, and anomaly detection.

I aim to design machine learning and computer vision methods that are reliable, robust, and generalisable. My research sits at the intersection of theory and practice, where I develop principled approaches grounded in statistical learning, kernel methods and deep learning approaches, and translate them into practical solutions for biometrics, anomaly detection, and time-series analysis. 

The unifying theme of my work is to build AI systems that can:

• operate with limited supervision, 

• remain effective under unseen conditions, and 

• adapt to new and evolving challenges such as deepfakes and adversarial attacks. 

Core research interests: one-class classification, kernel-based learning, Deep Learning and sparse modelling.

Research Themes

• Face Presentation Attack Detection

• Deepfake and Manipulation Detection

• Adversarial Defense

• One-Class and Open-Set Recognition

• Ensemble and Sparse Fusion

• Sequential/Time-Series Anomaly Detection

Summary of Recent Research in Media Forensics (Slides)

Selected Publications

• lp-norm constrained one-class classifier fusion (Information Fusion, 2025)

• Large-margin multiple kernel -SVDD using Frank–Wolfe algorithm for novelty detection (Pattern Recognition, 2024)

• One-Class Classification Using ℓp-Norm Multiple Kernel Fisher Null Approach (IEEE Transactions on Image Processing, 2023)

• Robust one-class kernel spectral regression (IEEE Transactions on Neural Networks and Learning Systems, 2021)

• Unknown Face Presentation Attack Detection via Localised Learning of Multiple Kernels (IEEE Transactions on Information Forensics and Security, 2023)

• ℓp-Norm Support Vector Data Description (Pattern Recognition, 2022)

• Matrix-Regularised One-Class Multiple Kernel Learning for Unseen Face Presentation Attack Detection (IEEE Transactions on Information Forensics and Security, 2021)

• Energy Normalization for Pose-Invariant Face Recognition Based on MRF Model Image Matching (IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011)

Key Projects

• Deepfake Detection Using One-Class and Open-Set Recognition

Current Directions

• Semi-supervised deepfake detection

• Adversarial image detection

• Sparsity-induced adaptive classifier fusion

• Anomaly detection in sequential data streams

Collaborations

• Centre for Vision, Speech and Signal Processing CVSSP, University of Surrey, UK

          Topic: Biometrics and Anomaly Detection

• School of Computer Science and Electronic Engineering CSEE, University of Essex, UK 

          Topic: Trustworthy AI