Publication Metrics
Publications
Journals & Conferences
MDPI, Elsevier, IEEEAmplifying the imaging power of ground-based telescopes with space-based telescopes data and generative AI
A novel software that enhances ground-telescope imagery by utilizing contextual information from space-based telescopes using a custom-trained generative AI model.
AI-Driven Cybersecurity Threats: A Survey of Emerging Risks and Defensive Strategies
This work analyzes emerging AI-driven cybersecurity threats and defenses, presenting a taxonomy of risks and highlighting the need for explainable and regulation-aware security solutions.
CornViT: A Multi-Stage Convolutional Vision Transformer Framework for Hierarchical Corn Kernel Analysis
A multi-stage framework for automated corn kernel quality assessment, achieving up to 94.11% accuracy across purity, shape, and embryo orientation tasks. Deployed as a lightweight web app, offering a practical solution for seed certification.
Galaxy Image Simplification Using Generative AI
Modern digital sky surveys acquire images of billions of galaxies requiring effective automation. This paper introduces a novel approach using generative AI to simplify galaxy images.
Multimodal Detection of Fake Reviews using BERT and ResNet-50
A multimodal fake review detection model using BERT for text and ResNet-50 for images, achieving a 0.934 F1-score. It effectively detects cross-modal inconsistencies, enhancing trust in online reviews.
FedOnco-Bench: A Reproducible Benchmark for Privacy-Aware Federated Tumor Segmentation with Synthetic CT Data
A benchmark for evaluating privacy-aware federated learning on synthetic oncologic CT scans. It highlights the trade-off between privacy and performance across FL methods, offering a standardized platform for medical image segmentation research.
Explainable Deep Learning in Medical Imaging: Brain Tumor and Pneumonia Detection
An explainable deep learning framework using ResNet50 and DenseNet121 to detect brain tumors and pneumonia from MRI and X-ray images. DenseNet121 achieved higher accuracy and more focused Grad-CAM visualizations, enhancing both performance and interpretability for clinical use.
Wavelet-based GAN Fingerprint Detection using ResNet50
A wavelet-based method using DWT and ResNet50 to detect StyleGAN-generated images, achieving up to 95.1% accuracy with Daubechies filters. Results show wavelet-domain artifacts offer strong cues for distinguishing GAN images, outperforming traditional spatial-domain models.
Fourier-Based GAN Fingerprint Detection Using ResNet50
Combines frequency-domain analysis with deep learning to distinguish StyleGAN-generated images from real ones. By applying a 2D Discrete Fourier Transform and training a ResNet50 model, it reveals spectral fingerprints, achieving 92.8% accuracy and an AUC of 0.95.
The Evolution of Search Engines: From Keyword Matching to AI-Powered Understanding
Search engines have undergone a dramatic transformation from simple keyword-based tools to intelligent, AI-powered systems. This paper explores their evolution, impact and key innovations like LLMs and semantic search.
Multimodal Deep Learning: A Survey of Models, Fusion Strategies, Applications, and Research Challenges
Multimodal DL is enabling models to process and reason across diverse data types like text, images, audio, and video. This survey explores foundational methods, state-of-the-art architectures, and real-world applications.
Towards a New Era of Sustainable Agriculture: AI Applications and Case Studies in Crop Management
AI is driving a digital transformation in agriculture, enabling smarter, more sustainable crop management practices. This review highlights key AI applications across yield prediction, precision farming, pest control, and irrigation, supported by real-world case studies and deployment insights.
Transforming Medicine With Intelligence: How AI Is Reshaping the Role of Doctors and the Future of Clinical Practice
AI is reshaping clinical practice by enhancing diagnosis, treatment, and care. This review highlights recent advances (2022-2025) and emphasizes human-AI collaboration in healthcare.
Agentic AI-The Rise of Autonomous Intelligent Agents in the Era of LLMs
Agentic AI enables systems to autonomously pursue goals using LLMs for reasoning. This survey covers its foundations, architectures, key frameworks, applications, challenges, and future prospects.
The Impact of Artificial Intelligence on Traditional Art Forms: A Disruption or Enhancement?
AI is reshaping clinical practice by enhancing diagnosis, treatment, and care. This review highlights recent advances (2022-2025) and emphasizes human-AI collaboration in healthcare.
Data-Centric AI: A Systematic Review of Methods, Challenges, and Future Directions
As model architecture improvements plateau, data-centric AI (DCAI) shifts focus to enhancing datasets for better performance. This paper reviews recent advances in DCAI methods, applications, challenges, and future directions toward more reliable and fair AI systems.
Explainable Identification of Similarities Between Entities for Discovery in Large Text
This research addresses limitations in existing AI tools by developing an n-gram analysis framework that provides precise and explainable insights into textual similarities.
Identifying Bias in Deep Neural Networks Using Image Transforms
CNNs often work as "black boxes" leading to hidden biases. This paper proposes techniques for identifying dataset biases using image transforms including Fourier and wavelet transforms.
Identifying bias in CNN image classification using image scrambling and transforms
CNNs achieve high accuracy in vision tasks but often rely on hidden, biased background features. This thesis proposes two methods to detect and separate contextual information from noise, improving interpretability across various datasets.