Click any publication to view its abstract, achievements, and full paper link — across IEEE, ScienceDirect, MDPI, arXiv and Wiley.
An attention-integrated CNN architecture for multi-class ophthalmic image classification achieving high precision across fundus, OCT, and external eye imaging modalities — validated with concept-level interpretability reports reviewed by clinicians.
A federated learning framework using attention-based aggregation for microscopic fungal image classification, enabling collaborative training across distributed microscopy labs without centralizing sensitive specimen data.
A triple self-supervised pre-training scheme combining masked image modeling, contrastive learning, and reconstruction tasks to learn transferable representations from unlabeled brain MRI volumes. Fine-tuned on small clinical cohorts for accurate tumor classification on the ND-5 dataset.
A knowledge-based federated learning framework for eye disease diagnosis that integrates domain knowledge priors into local model training and produces faithful explanations aligned with ophthalmological concepts — maintaining data privacy across participating hospitals.
A structured analysis of transformer-based models applied to Bengali text summarization, covering extractive and abstractive paradigms, benchmark datasets, evaluation metrics, and the architectural gaps that still separate Bengali from high-resource languages.
An enhanced BERT-based deep learning model fine-tuned for binary classification of ethical versus non-ethical text content, evaluated on curated datasets of online discourse with adversarial test probes.
An IoT-integrated system for real-time monitoring and preservation of perishable crops, combining edge sensing with energy-efficient control logic evaluated on live agricultural deployments.
BibTeX, preprints, and full text are available for every paper — just ask, or drop us a line if you're starting something we should know about.