Mohammad Junayed Hasan's repositories
occupational-stress-ml
This repository contains code, datasets, and analysis for AI-driven occupational stress detection using machine learning, deep learning, and NLP. It includes feature selection, explainable AI, synthetic data generation, and model validation for workplace safety applications. 🚀
Quantum-Machine-Learning-Qiskit-PyTorch
This repository contains codes and tutorials for quantum machine learning using PyTorch and Qiskit. It covers topics such as qiskit basics, deep learning fundamentals, and hybrid quantum-classical models. It also demonstrates how to use TorchQuantum, a PyTorch-based framework for quantum neural networks.
valentines_blossoming_flower
Create a romantic surprise! 💖 This interactive Valentine’s webpage asks Will You Be My Valentine? with fun animations, a blossoming flower effect, and playful responses. Fully customizable and easy to host on GitHub Pages. 🌸✨ Try it now!
Quantum-Machine-Learning
This repository implements knowledge distillation from classical to quantum neural networks for image classification. It includes experiments on MNIST and FashionMNIST datasets, demonstrating improved accuracy in quantum models. Code for classical teachers and quantum students, with baseline and distilled versions, is provided.
Deep-Learning-PyTorch
This repo contains jupyter notebooks that show how to use PyTorch for various deep learning tasks, such as image classification, object detection, NLP, and more. The notebooks cover data loading, model building, training, evaluation, and advanced techniques.
spontaneous-smile-recognition
A deep learning framework for distinguishing spontaneous from posed smiles in videos. It uses a multi-task approach, leveraging Duchenne Marker features through transformer networks. The model simultaneously predicts smile types and D-Markers, achieving state-of-the-art results on major datasets without D-Marker computation during inference.
Adult-Income-Prediction-Machine-Learning
This project analyzes the Adult Income Dataset to predict individual income levels based on census data. It showcases data preprocessing, visualization, and machine learning techniques. The workflow includes data cleaning, transformation, feature engineering, and predictive modeling, providing insights into income factors.
alzheimers-disease-detection
Implementation of an ensemble approach for Alzheimer's disease detection using MRI images. Combines EfficientNet-B2 and VGG16 models with feature concatenation. Utilizes ADASYN for data balancing. Includes individual model training, ensemble implementation, and pre-trained weights. Based on the paper by Qin et al. (2023).
Life-Satisfaction-Machine-Learning
This repo contains code for predicting life satisfaction using machine learning and explainable AI, as published in Heliyon. It includes a Jupyter Notebook with data processing, model building, and result visualization using Python libraries. The analysis uses the SHILD dataset to explore factors influencing life satisfaction.
Clinical-Language-Model-Distillation-Pruning-Quantization
OptimCLM enhances clinical decision-making and management by optimizing clinical language models for patient outcome predictions, achieving up to 22.88× compression and 28.7× speedup with minimal performance loss, using techniques like knowledge distillation, pruning, and quantization on MIMIC-III data.
CrazyCrawler
A website which works as a search engine and enables users to scrape and crawl data from websites. Built using Django framework with HTML, CSS, JS and mySQL database. Android app built using Android Studio.
Help-The-Needy
An online donation system to help the poor people at remote areas of Bangladesh.
LLM-Aggregated-Ensemble-Clinical-Outcomes
Implementation of a novel method using aggregated ensembles of large language models to analyze long clinical texts. Addresses input length limitations and data source diversity in clinical outcome prediction. Includes experiments on mortality prediction and length of stay estimation using MIMIC-III data.
word-alignment-techniques-in-machine-translation
A collection of word alignment models for machine translation, including Dice's Coefficient, IBM Model 1, and a Bidirectional model with grow-diag-final-and heuristic. Implements training, alignment, and evaluation tools. Demonstrates progression from baseline to advanced techniques, with significant improvements in Alignment Error Rate (AER).
KD-LLM-Clinical-Outcomes
Codes for the research article "Distilling the Knowledge of Clinical Outcome Predictions in Large Language Models for Resource Constrained Healthcare Systems"
auto-annotate-ner
This repository contains code for automatically annotating Bengali health text data using Large Language Models (LLMs). It contains different implementations, starting with a baseline in-context learning, then using chain-of-thought prompting, scoring mechanisms, and human-in-the-loop workflows one by one.
BengaliClinicalMT
Official repository for "Extrinsic Evaluation of Machine Translation Quality via Downstream Tasks for Low-Resource Bengali Clinical Texts." Includes datasets, preprocessing pipelines, translation models, code for evaluation, and clinical outcome prediction tasks.
junayed-hasan.github.io
Personal academic-industry portfolio website
life-trajectory-embeddings
Cloud-native system that transforms biographical data into high-dimensional vector embeddings, enabling semantic search across a wide range of life trajectories. Built with Vertex AI, BigQuery, and FastAPI - explore how career paths cluster in embedding space and compare your own trajectory against historical figures.
llendify
LLendify leverages large language models (LLMs) to analyze bank statements and provide instant loan eligibility insights. Upload your bank statement PDF to receive AI-powered loan recommendation, analysis of income patterns, spending habits, and creditworthiness. Get comprehensive financial assessments and loan recommendations in seconds.
Machine-Translation-Decoders
This repository contains Python implementation of machine translation decoders. It includes multiple decoding algorithms: a monotone decoder, a non-monotonic decoder with phrase reordering, and an advanced decoder using beam search with future cost estimation. It aims to compare these algorithms in terms of translation quality and efficiency.
Machine_Learning
CSE445 Machine Learning Resources
nlp-adv-hw
This repository contains code for profiling the computational complexity, memory usage, and wall clock time of self-attention mechanisms as a function of input sequence length.
smile-recognition-fusion
An automatic, efficient, and effective framework for spontaneous smile classification combining handcrafted D-Marker features with transformer-based automatic features using diverse feature fusion techniques. Achieves state-of-the-art results across benchmark smile datasets.