M. G. Sarwar Murshed
About News Publications Teaching Awards Team Open Source Talks
Photo of Dr. Md Golam Sarwar Murshed

M. G. Sarwar Murshed

Assistant Professor
Computer Science
University of Wisconsin–Green Bay
Green Bay, Wisconsin, USA
Lab
GBAI Lab
Research
Efficient & Trustworthy AI, Biometrics, Cybersecurity, Edge AI

I develop efficient and robust deep learning systems that bridge state-of-the-art AI performance with real-world deployment constraints. Recent projects include CRFSEG (97.17% fingerprint matching accuracy) and EdgeLite (92.37% accuracy on edge devices with 13% parameter reduction).

Efficient & Trustworthy AI, built for real-world deployment.

My research focuses broadly on developing algorithms for efficient and scalable machine learning systems. I am particularly interested in hardware-friendly deep learning algorithms and its applications in biometrics, vision, language, and speech using the edge computing paradigm. My current work focuses on (i) resource-aware deep learning and edge AI for computer vision, (ii) trustworthy AI for biometrics and cybersecurity with interpretability, and (iii) adapting foundation models (LLMs) for resource-constrained settings.
Efficient & Resource-Aware Deep Learning Biometrics & Security Edge AI for Vision Efficient Foundation Models
CV Publications Teaching

Who I Am

Quick snapshot

About

Md Golam Sarwar Murshed is an Assistant Professor of Computer Science in the Resch School of Engineering at the University of Wisconsin–Green Bay (UWGB), where he founded and directs the Green Bay AI Lab (GBAI Lab). He earned his M.S. and Ph.D. from Clarkson University, NY, with research spanning efficient, robust, and explainable deep learning for real-world deployment across biometrics, computer vision, robotics, and cybersecurity. His work includes CRFSEG (Clarkson Rotated Fingerprint Segmentation), a deep learning system for accurately localizing and classifying rotated slap fingerprints (97.17% fingerprint matching accuracy), as well as EdgeLite, which achieves 92.37% accuracy on edge devices with a 13% parameter reduction. He has also introduced the Oriented Regional Proposal Network (O-RPN) to improve high-precision detection of oriented objects. His research has accumulated 725+ citations (as of Jan 2026) and has been supported by organizations including NSF-CITeR, Verizon, Badger Technologies, and WiSys. Before joining UWGB, he spent five years as a Lead Software Engineer at Samsung R&D, bringing strong industry experience in building scalable software systems that informs his focus on deployable, trustworthy AI.

At a glance

Position
Assistant Professor (CS)
Affiliation
UW–Green Bay
Background
Ph.D., Clarkson University
Industry
Samsung R&D (5 yrs)
Citations
725+ (Google Scholar)
Funding
NSF-CITeR · Verizon · WiSys
Publications Open-source software
Efficient deep learning Edge AI Biometrics Explainable AI Cybersecurity Foundation Models

Highlights

Selected themes
Placeholder: Efficient biometrics systems

Efficient object detection

Oriented object detection, Deep Learning, Fingerprint segmentation (contact-based and contactless).

AI Deep learning Computer vision
Efficient biometrics systems

Biometrics at scale

Fingerprint segmentation and recognition (contact-based and contactless), including age-invariant segmentation, spoof synthesis, and template security.

Fingerprints Contactless Template security
Placeholder: Edge AI

Edge AI & deployment

Resource-aware model design and deployment for vision, robotics, and infrastructure (hazard detection, pavement distress), including EdgeLite-style efficiency.

Edge deployment Compression ROS
Placeholder: Trustworthy AI

Explainable & trustworthy AI

Robustness and interpretability for high-stakes recognition and security (e.g., face, fingerprint recognition), improved localization.

XAI Robustness Security

News

View all

Selected Publications

A few recent items · Full list
2025 · Journal
Conditional Synthetic Live and Spoof Fingerprint Generation
IET Biometrics
DOI arXiv
2025 · Conference
Deep Learning-Based Approaches for Contactless Fingerprints Segmentation and Extraction
BIOSIG 2025
DOI arXiv
2024 · Journal
Deep Age-Invariant Fingerprint Segmentation System
IEEE T-BIOM
DOI arXiv
2021 · Journal
Machine Learning at the Network Edge: A Survey
ACM Computing Surveys
DOI arXiv

Teaching

Course pages

UW–Green Bay

Assistant Professor (Aug 2023–Present)

  • Artificial Intelligence
  • Advanced Object-Oriented Design
  • Cloud Computing (course developed)
  • Discrete Mathematics
  • Introduction to Computing & Internet
  • Web Programming

Clarkson University

Teaching Assistant (Aug 2018–Dec 2019)

  • EE 262: Intro to OOP & Software Design
  • EE 260/360: Embedded Systems / Microprocessors

Contact

Let’s connect

Email is the fastest way to reach me: murshedm (at) uwgb (dot) edu.

UW–Green Bay Google Scholar GitHub