Enhancing skin lesion classification using a Tri-Path Attention Stacked Ensemble architecture with Cohen's Kappa Proportioned Averaging
Hasan, Efat, Bhuiyan & Mansia · Digital Health, 2026
94.44% accuracyHAM10000Hi, I am
Machine learning researcher and full-stack developer
Passionate about medical image classification
CSE graduate from IUBAT, Dhaka. I train deep learning models that read medical images and speech, and build web apps that run real inventory and billing.
My work focuses on medical image classification and speech emotion recognition. My bachelor's thesis, supervised by Anwar Hossain Efat, proposed attention-based stacked ensembles for skin lesion classification on HAM10000.
Hasan, Efat, Bhuiyan & Mansia · Digital Health, 2026
94.44% accuracyHAM10000Bhuiyan, Efat, Hasan & Mansia · Digital Health, 2026
93.96% accuracyHAM10000Bhuiyan, Imran, Rony, Anni & Hasan · Springer, presented at BIM 2025, Dhaka International University
LSTMSpeechSeven-class classifier trained on 10,015 dermoscopic images. Combines EfficientNet backbones with triplet attention and fuses model layers using MCC-weighted and Cohen's Kappa averaging. Grad-CAM shows which part of the lesion drives each prediction.
Admins manage products and categories and view billing data. Customers browse the catalog and place orders. Role-based login separates the two, and orders use Eloquent transactions.
Database-backed library system with user profiles, search filters for book availability and rental history.
Personal expense manager with user login, create/read/update/delete for daily spending and monthly reports.
A daily task manager built with Laravel and styled with Tailwind CSS.
A to-do web app built with plain HTML, CSS and JavaScript.
A live clock web page that shows the current time and date.
Built the inventory management and smart billing system with Laravel and MySQL. Rated excellent for performance and conduct. See the certificate.
Taught students one-to-one, explaining lessons clearly and patiently.
IUBAT – International University of Business Agriculture and Technology
CGPA: 3.52 / 4.00
Thesis on hierarchical attention stacked ensembles for skin lesion classification, supervised by Anwar Hossain Efat. Practicum project: an inventory management and smart billing system with Laravel.
Student ID 22103114
Copy the student ID, open the verification page and paste it there.
Armed Police Battalion Public School & College
GPA: 5.00 / 5.00
Completed higher secondary studies in the Science group with a perfect GPA of 5.00. Alongside academics, I took part in debate competitions, science fairs and sports events, which strengthened my teamwork, leadership and ability to adapt to new challenges.
Activities: Debate team
Daudpur High School
GPA: 5.00 / 5.00
Completed secondary education in the Science group with a perfect GPA of 5.00. As a member of the Student Cabinet I helped organise school events and represent student interests. Volleyball and football built discipline and teamwork, and debating sharpened my critical thinking and public speaking.
Activities: Student Cabinet, volleyball team, football team, debating
English proficiency, CEFR B2
Overall band 6.0
Listening 7.0 · Reading 5.5 · Writing 6.0 · Speaking 6.0
Two versions of my CV. The Academic CV is for research and PhD or Master's applications, and the Europass CV follows the standard European format.
Research interests, publications, projects, skills and references.
Download PDFEducation, work experience, certifications and language skills in the Europass format.
Download PDF