Four threads chosen for their depth, difficulty, and the difference they can make outside the journal — pursued with seven distinguished international collaborators and two partner labs.
Every paper, every dataset, every review cycle reaches across borders. Our co-authorship network spans four countries and three regions — the work simply doesn't happen any other way.
Each track is rigorous, open, and pursued with international collaborators across academic medicine, engineering, and CS.
Self-supervised pre-training and federated learning for MRI brain tumor, skin cancer, and ocular disease classification — built to generalize beyond the hospital that trained them.
Modern CNN, transformer, and graph-neural architectures for medical and scientific recognition. Evaluated for failure modes — not just accuracy at the median.
Transparent, interpretable decisions for high-stakes clinical settings — faithfulness over plausibility, explanations that survive ablation and clinician review.
Bengali NLP, multilingual ethical content classification, and speech recognition — pushing the under-resourced side of NLP toward parity.
Researchers from Khalifa, AIUB, Woosong, King Saud, UIU and beyond whose mentorship and co-authorship has shaped our published work.
AE · Abu DhabiCo-authored T3SSLNet in IEEE Access — a triple-method self-supervised framework leveraging contrastive learning to improve automated brain tumor diagnosis accuracy and efficiency.
BD · DhakaContinuous research alliance across T3SSLNet, a cervical cancer vision-transformer in Healthcare Analytics, and a monkeypox diagnostic in Healthcare Technology Letters. Many more under review.
BD · DhakaCo-authored the cited Color Sorting Robotic Arm work and is now developing a state-of-the-art skin cancer classification model with comprehensive ablation studies.
KR · DaejeonLandmark benchmarking study on brain tumor diagnosis — currently under active peer review in Nature Scientific Reports, decisively outperforming literature on external, held-out datasets.
SA · RiyadhCo-authored the brain tumor benchmarking study under review at Nature Scientific Reports. His expertise in image optimization and pattern recognition refined the software architecture and generalization.
BD · DhakaCo-authored an industrial IoT paper on perishable crop monitoring in IJES, plus joint work on extremism detection and transformer-based Bengali summarization.
BD · DhakaPrimary co-author across our healthcare portfolio — T3SSLNet, cervical cancer Vision-Transformer, monkeypox detection, kidney ultrasound classification, and the Hi-TGNet hybrid transformer-graph architecture.
Working on something we should know about, or want to co-author a paper across imaging, XAI or Bengali NLP? Get in touch — we read every message.
Cross-lab alliances that harden our methodology, audit our code, and bulletproof our manuscripts before they hit a reviewer's desk.
Intelligent Multimedia Signal & Image Processing Lab — an elite international research hub specialising in AI-driven sensor data processing, deep learning, and explainable computer vision. MSiP serves as our quality-assurance anchor: pre-peer-review evaluations, codebase audits, and architectural validation for our deep learning models heading to top-tier venues.
Advanced Machine Intelligence Research Lab — a research collective at the forefront of AI, Computer Vision, and Large Language Models. Beyond our joint publication footprint, AMIR is our collaborative incubator: pre-submission manuscript pipelines, comprehensive code reviews, algorithmic optimisation, and training-framework stress tests for shared software architectures.
Three doors into the lab — for collaborators, students, and the openly curious.
Co-authorships and joint projects with faculty and global teams — from imaging to ethics to embedded systems.
Hands-on AI research with mentorship from the lab and our partner network. Cohorts open a few times a year.
Idea sharing, inclusive discussion, and curiosity-driven innovation. Drop us a line about anything you're working on.
The collaborator wall above grew one conversation at a time. Start one.