Publications

2026

A Classroom Study of LLM-Generated Feedback Intervention in Introductory Programming

Published in IRAISE 2026, Festival of Learning (AIED, EDM & L@S), 2026

A randomized classroom deployment comparing natural-language hints, AI-generated failing test cases and no AI feedback in introductory Python labs, with the ProgFeed dataset of 6,693 submissions from 215 students; natural-language feedback is linked to higher completion and faster convergence to correct solutions.

Cite

Hasnain Heickal and Andrew Lan. "A Classroom Study of LLM-Generated Feedback Intervention in Introductory Programming." IRAISE 2026, Festival of Learning (AIED, EDM & L@S), 2026.

Paper

2025

Learning Code-Edit Embeddings to Model Student Debugging Behavior

Published in International Conference on Artificial Intelligence in Education, 2025

AI in Education LLMs

This work presents an encoder–decoder model that learns code-edit embeddings from consecutive student submissions, using test case outcomes to fine-tune large language models. The approach enables personalized code suggestions that preserve student style while improving correctness, and reveals common debugging patterns through clustering analysis.

Cite

Hasnain Heickal and Andrew Lan. "Learning Code-Edit Embeddings to Model Student Debugging Behavior" International Conference on Artificial Intelligence in Education 2025, 91-98.

PDF

Improving User Engagement and Learning Outcomes in LLM-Based Python Tutor: A Study of PACE

Published in CHI Conference on Human Factors in Computing Systems (Extended Abstracts), 2025

AI in Education LLMs HCI

PACE, a Python tutor built on small language models that gives step-by-step guidance and adaptive feedback, evaluated for engagement and learning outcomes.

Cite

Muhtasim Ibteda Shochcho, Mohammad Ashfaq Ur Rahman, Shadman Rohan, Ashraful Islam, Hasnain Heickal, AKM Mahbubur Rahman, M. Ashraful Amin, and Amin Ahsan Ali. "Improving User Engagement and Learning Outcomes in LLM-Based Python Tutor: A Study of PACE." CHI EA 2025.

Paper

2024

Generating Feedback-Ladders for Logical Errors in Programming using Large Language Models

Published in Educational Data Mining, 2024

AI in Education LLMs

Using large language models (LLMs) to generate a “feedback-ladder,” providing multiple levels of feedback for programming assignments based on student submissions. The approach aims to offer progressively detailed feedback while considering the learning context of students, addressing limitations of existing methods that provide overly direct, one-size-fits-all feedback.

Cite

Hasnain Heickal and Andrew Lan. "Generating Feedback-Ladders for Logical Errors in Programming using Large Language Models." Educational Data Mining 2024, 947-951.

PDF

2021

An Automatic Abstractive Text Summarization System

Published in Dhaka University Journal of Applied Science and Engineering, 2021

NLP & Speech

An abstractive text summarization system using pointer generator network and topical feature extraction.

Cite

Barna, Nasid Habib, and Hasnain Heickal. "An Automatic Abstractive Text Summarization System." Dhaka University Journal of Applied Science and Engineering 6, no. 2 (2021): 39-48.

Paper

2018

A framework for developing graphically programmable low-cost robotics kit for classroom education

Published in International Conference on Education Technology and Computers (ICETC), 2018

AI in Education Robotics

A framework for building a low-cost educational robotics kit programmed through a visual language, so students can apply programming concepts without worrying about syntax.

Cite

Mehedi Hasan Masum, Tanvir Shahriar Rifat, Saifuddin Md Tareeq, and Hasnain Heickal. "A framework for developing graphically programmable low-cost robotics kit for classroom education." ICETC 2018, 22-26.

Paper

End-to-End Speech Synthesis for Bangla with Text Normalization

Published in 2018 5th International Conference on Computational Science/Intelligence and Applied Informatics (CSII), 2018

NLP & Speech

An end-to-end speech synthesis system in Bangla (or Bengali) using minimal training and zero language specific knowledge.

Cite

Pial, Tanzir Islam, Shahreen Salim Aunti, Shabbir Ahmed, and Hasnain Heickal. "End-to-End Speech Synthesis for Bangla with Text Normalization." In 2018 5th International Conference on Computational Science/Intelligence and Applied Informatics (CSII), pp. 66-71. IEEE, 2018.

Paper

Deep Learning Based Bangla Speech-to-Text Conversion

Published in 2018 5th International Conference on Computational Science/Intelligence and Applied Informatics (CSII), 2018

NLP & Speech

A Bangla (or Bengali) Speech-to-Text conversion system using deep recurrent neural networks.

Cite

Tausif, Md Tahsin, Sayontan Chowdhury, Md Shiplu Hawlader, Md Hasanuzzaman, and Hasnain Heickal. "Deep learning based bangla speech-to-text conversion." In 2018 5th International Conference on Computational Science/Intelligence and Applied Informatics (CSII), pp. 49-54. IEEE, 2018.

PDF

Segmentation of Heterogeneous Documents into Homogeneous Components using Morphological Operations

Published in IEEE/ACIS International Conference on Computer and Information Science (ICIS), 2018

Computer Vision

A language-independent document segmentation system that splits printed pages into halftones and graphics, tables with their cells, and text blocks.

Cite

Nasid Habib Barna, Tisa Islam Erana, Shabbir Ahmed, and Hasnain Heickal. "Segmentation of Heterogeneous Documents into Homogeneous Components using Morphological Operations." ICIS 2018, 513-518.

Paper

2017

User Authentication from Mouse Movement Data Using Multiple Classifiers

Published in International Conference on Machine Learning and Computing (ICMLC), 2017

Security

User authentication from mouse movement trajectories, using twelve movement features with SVM, k-nearest-neighbour and naive Bayes classifiers.

Cite

Masud Karim, Hasnain Heickal, and Md. Hasanuzzaman. "User Authentication from Mouse Movement Data Using Multiple Classifiers." ICMLC 2017, 122-127.

Paper

2016

SeqDev: An Algorithm for Constructing Genetic Elements Using Comparative Assembly

Published in Plant Tissue Culture and Biotechnology, 2016

Bioinformatics

SeqDev builds contigs from raw reads using reference sequences, combining a weighted-frequency consensus (BlastAssemb) with suffix arrays and gap-filling search; demonstrated on Pokkali rice promoter sequences.

Cite

Tanzilur Rahman, Hasnain Heickal, Sumaiya Tabrejee, M. M. K. Chowdhury, S. M. Sarwar, and Mohammad Shoyaib. "SeqDev: An Algorithm for Constructing Genetic Elements Using Comparative Assembly." Plant Tissue Culture and Biotechnology 26, no. 1 (2016): 105-121.

Paper

2015

Computer Vision-Based Real-Time 3D Gesture Recognition Using Depth Image

Published in International Journal of Image and Graphics, 2015

Computer Vision HCI

Real-time 3D gesture recognition from depth images by tracking head, neck, shoulder, arm, hand and leg joints with a Kinect sensor, with centre-of-gravity correction and joint-position normalization.

Cite

Hasnain Heickal, Tao Zhang, and Md. Hasanuzzaman. "Computer Vision-Based Real-Time 3D Gesture Recognition Using Depth Image." International Journal of Image and Graphics 15, no. 1 (2015): 1550004.

Paper

2013

Real-time 3D full body motion gesture recognition

Published in IEEE International Conference on Robotics and Biomimetics (ROBIO), 2013

Computer Vision Robotics HCI

Full-body gesture recognition with a Kinect, trained on 12 cricket-umpire gestures; reaches 98.11% accuracy with a neural-network classifier.

Cite

Hasnain Heickal, Tao Zhang, and Md. Hasanuzzaman. "Real-time 3D full body motion gesture recognition." IEEE ROBIO 2013, 798-803.

Paper