Research

Video Anomaly Detection & Intelligent Surveillance

My core research area focuses on video anomaly detection and intelligent surveillance, addressing challenges such as weak supervision, rare event learning, temporal ambiguity, and real-world deployment constraints. I design hybrid deep architectures that integrate convolutional networks, temporal modeling, attention mechanisms, and transformers to reliably detect abnormal events in large-scale surveillance environments.

Representative Contributions

• Weakly-supervised and semi-supervised anomaly recognition frameworks

• CNN–LSTM and CNN–Transformer hybrid architectures

• Temporal attention and memory-augmented anomaly modeling

• Benchmark evaluation on large-scale surveillance datasets

• Selected Publications

• Multiple peer-reviewed articles in leading journals and conferences such as Pattern Recognition, Knowledge-Based Systems, and IEEE-indexed venues. (Complete list available on Google Scholar)

UAV-Based Vision & Remote Sensing

I work on AI-driven vision systems for UAVs and remote sensing platforms, focusing on event detection, anomaly recognition, and intelligent scene understanding from aerial imagery and video. My research emphasizes lightweight and efficient deep learning models that are suitable for real-time inference and deployment under resource-constrained environments.

This work targets applications such as disaster monitoring, public safety, infrastructure inspection, and intelligent geospatial analysis.

Representative Contributions

• UAV-based video event recognition and anomaly detection

• Lightweight deep learning architectures for aerial and edge-based platforms

• Transformer-based and attention-driven models for remote sensing analysis

• Real-time inference under computational and communication constraints

Selected Publications

• UAV-assisted real-time disaster detection using optimized transformer-based architectures

• Research articles on deep learning and transformer models for aerial video understanding

• Peer-reviewed studies addressing intelligent surveillance and event recognition from UAV platforms

Energy Informatics  

My research in Energy Informatics focuses on the application of machine learning and deep learning to short-term energy forecasting and electricity consumption modeling in residential and smart-building environments. I develop hybrid deep learning strategies that integrate convolutional feature extraction with temporal sequence modeling to improve forecasting accuracy under real-world operating conditions, including changing consumption patterns and disruptive events such as the COVID-19 pandemic.

This research supports intelligent energy management, demand-side forecasting, and data-driven planning for smart grids and building energy systems.

Representative Contributions

• Hybrid deep learning models for short-term residential load forecasting

• CNN–GRU and related temporal modeling architectures for energy prediction

• Data-driven modeling of electricity consumption under behavioral and societal shifts

• Benchmarking and analysis of forecasting datasets and research directions

Selected Publications

• A novel CNN-GRU-based hybrid approach for short-term residential load forecasting

• Electrical energy prediction in residential buildings for short-term horizons using hybrid deep learning strategy

• Modelling electricity consumption during the COVID-19 pandemic: Datasets, models, results and a research agenda

Federated Learning, Split Learning & Privacy-Preserving AI 

I actively research federated learning (FL) and split federated learning (SFL) for distributed intelligence, with emphasis on privacy, communication efficiency, robustness, and adversarial resilience. My work proposes novel aggregation strategies, client-aware learning mechanisms, and knowledge distillation-based federation to enable collaborative learning without sharing raw data.

Representative Contributions

• Split federated learning for video and medical data

• Client-aware and adaptive aggregation strategies

• Privacy-enhanced training under non-IID data distributions

• Federated anomaly detection and edge intelligence

Selected Publications

Published and ongoing work in IEEE journals and international conferences addressing FL-based surveillance, healthcare, and IoT systems.

Transformer-Based Vision Models

My research focuses on transformer architectures for computer vision, with particular emphasis on spatial–temporal modeling for video understanding and anomaly recognition. I develop hybrid frameworks that integrate vision transformers with convolutional and temporal learning modules to capture both long-range contextual dependencies and fine-grained spatial–temporal features. These models are designed for complex vision tasks in surveillance environments, where subtle temporal patterns and contextual reasoning are critical.

A key theme of this work is the fusion of attention mechanisms with dual-stream and reservoir-based architectures to enhance robustness, interpretability, and generalization in real-world video analysis.

Representative Contributions

• CNN–Transformer hybrid architectures for video anomaly detection

• Dual-stream transformer-enhanced frameworks for human activity recognition

• Vision transformer attention mechanisms for spatial–temporal anomaly modeling

• Integration of transformers with temporal and reservoir-based networks

Selected Publications

• TDS-Net: Transformer enhanced dual-stream network for video anomaly detection

• Shots segmentation-based optimized dual-stream framework for robust human activity recognition in surveillance video

• Vision transformer attention with multi-reservoir echo state network for anomaly recognition

Medical Image Analysis & Healthcare AI

My research in healthcare AI focuses on the development of robust and explainable deep learning models for medical image analysis, including classification, segmentation, and disease progression modeling. I emphasize cross-dataset generalization, interpretability, and clinical relevance, particularly when working with limited, heterogeneous, or noisy medical data.

A key aspect of this work is the design of data-driven frameworks that support early diagnosis, disease monitoring, and assisted healthcare decision-making through advanced machine learning and deep learning techniques.

Representative Contributions

• Deep learning models for disease detection, segmentation, and diagnosis

• Multi-view and multi-modal medical image analysis frameworks

• Explainable AI techniques for medical decision support

• Robust modeling under limited data and heterogeneous imaging conditions

Selected Publications

• Early progression detection from MCI to AD using multi-view MRI for enhanced assisted living

• An adaptive filtering technique for segmentation of tuberculosis in microscopic images

• Peer-reviewed journal articles addressing AI-driven medical image analysis and diagnostic systems

Bioinformatics and Computational Genomics

This research direction focuses on the application of machine learning and deep learning techniques to problems in genomics and biomedical data analysis. My work emphasizes sequence-based learning, temporal modeling, and multi-stage deep architectures for extracting meaningful patterns from high-dimensional biological data. A key contribution of this research is the development of machine learning models for splicing site prediction in the human genome, where sequence characteristics are learned to accurately identify functional genomic regions. In addition, I have worked on deep learning–based classification of COVID-19 variants, leveraging multi-stage temporal convolutional networks to model complex biological signatures for reliable variant discrimination.

Representative Contributions

• Machine learning–based prediction of splicing sites in the human genome

• Deep learning architectures for sequence and temporal biological data

• Multi-stage temporal convolutional networks for COVID-19 variant classification

• Robust modeling of biomedical data using data-driven AI approaches

Selected Publications

• Peer-reviewed journal articles on genomic sequence analysis using machine learning

• Research on deep learning frameworks for COVID-19 variant classification