Md Akmol Masud

I’m Akmol, an incoming graduate student in Electrical and Computer Engineering at Queen’s University (Kingston, Canada).

My research focuses on federated learning under constraints and split learning, making collaborative machine learning work on real devices with limited bandwidth and non-IID distributions of data.

At Jahangirnagar University (B.Sc., ICT), I spent most of my time on biomedical ML and other tangential topics to sharpen my skills. With that foundation, I’m now investing my time fully in FL and wireless networks.

If you have any discussion, idea, or just want a conversation, feel free to reach out. Always happy to connect.

news

Jun 22, 2026	Paper accepted in JMIR AI 🧬
Jun 01, 2026	SQU accepted at IEEE QPAIN 2026 ⚛️
Apr 15, 2026	HealthNLP_Retrievers at ArchEHR-QA 2026 (CL4Health @ LREC) 🏥
Apr 08, 2026	MosQNet-SA is out in PLOS ONE 🦟
Apr 01, 2026	DREAM is published in Biomedical Signal Processing and Control 🩺 — we detect sleep apnea from a single-lead ECG (the kind you can get from a wearable), with built-in explainability so clinicians can see exactly which signal segments triggered the alert. A meaningful step toward accessible, low-cost sleep diagnostics. [ScienceDirect]

selected publications

JMIR AI
Clinical Validity in Synthetic EHR Generation: Best-of-N Sampling in Tabular GANs

Md Akmol Masud and Mahmud Hasan

JMIR AI, 2026

Accepted; in production

Abs DOI Bib HTML

We introduce a three-regime taxonomy for evaluating clinical validity in synthetic EHR generation, governed by the generator’s valid-support mass (p_valid). Best-of-N sampling—generating multiple candidates and retaining the clinically closest one—shifts generators toward safer regimes. Benchmarked across WGAN-GP and CTGAN on multiple clinical datasets with a rigorous multi-metric privacy audit including distance-to-closest-record and proximity concentration analysis.
@article{masud2026jmirai, author = {Masud, Md Akmol and Hasan, Mahmud}, title = {Clinical Validity in Synthetic {EHR} Generation: Best-of-N Sampling in Tabular {GAN}s}, journal = {JMIR AI}, year = {2026}, doi = {10.2196/90590}, url = {https://doi.org/10.2196/90590}, note = {Accepted; in production}, }
TMLR
Quantum Rationale-Aware Graph Contrastive Learning for Jet Discrimination

Md Abrar Jahin, Md Akmol Masud, M. F. Mridha, and 1 more author

Transactions on Machine Learning Research, 2026

Abs arXiv Bib HTML

We propose a quantum rationale-aware graph contrastive learning framework for jet discrimination in high-energy physics, combining graph neural networks with quantum computing to improve classification performance.
@article{masud2026quantumcontrastive, author = {Jahin, Md Abrar and Masud, Md Akmol and Mridha, M. F. and Dey, Nilanjan}, title = {Quantum Rationale-Aware Graph Contrastive Learning for Jet Discrimination}, journal = {Transactions on Machine Learning Research}, year = {2026}, url = {https://arxiv.org/abs/2411.01642}, }
arXiv
Stabilizing Federated Learning under Extreme Heterogeneity with HeteRo-Select

Md Akmol Masud, Md Abrar Jahin, and Mahmud Hasan

2025

Abs arXiv Bib HTML

HeteRo-Select is a client selection framework for federated learning under extreme statistical and system heterogeneity, featuring variance-aware scheduling and heterogeneity-aware aggregation strategies.
@misc{masud2025heteroselect, author = {Masud, Md Akmol and Jahin, Md Abrar and Hasan, Mahmud}, title = {Stabilizing Federated Learning under Extreme Heterogeneity with HeteRo-Select}, year = {2025}, url = {https://arxiv.org/abs/2508.06692}, }

PLOS ONE

MosQNet-SA: Explainable Convolutional-Attention Network for Mosquito Classification with Application as a RESTful API for Dengue and Malaria Risk Mapping

Md Akmol Masud, Sanjida Akter, Nadia Sultana, and 4 more authors

PLOS ONE, 2026

Abs DOI Bib HTML

@article{masud2026mosquito,
  author = {Masud, Md Akmol and Akter, Sanjida and Sultana, Nadia and Islam, Mohammad Shahidul and Abu Yousuf, Mohammed and Noori, Farzan M. and Uddin, Md Zia},
  title = {MosQNet-SA: Explainable Convolutional-Attention Network for Mosquito Classification with Application as a RESTful API for Dengue and Malaria Risk Mapping},
  journal = {PLOS ONE},
  year = {2026},
  volume = {21},
  number = {4},
  pages = {1--30},
  doi = {10.1371/journal.pone.0344970},
  url = {https://doi.org/10.1371/journal.pone.0344970},
}