Hi, I am Mohamad Hawshar

AI Researcher

What I do

AI Researcher

I'm an AI Researcher, specializing in machine learning and computer vision. Currently an AI Research assistant at LINUM - UQAM, I develop innovative approaches using Vision Transformers for numerous precise brain analysis tasks. I collaborate on scientific papers and curate a vast dataset of brain images for groundbreaking research in biomedical imaging.

My Work

Who I am

Master's in Artifial intelligence student

I'm Mohamad Hawchar, a passionate scientist with a Master's in Computer Science focused on Artificial Intelligence. I love exploring machine learning, computer vision, and big data analytics. Through hands-on projects, I've developed advanced medical image segmentation models and tackled misinformation with machine learning.

Beyond AI, I've actively volunteered in Lebanon for COVID-19 prevention and vaccination. With proficiency in English, French, and Arabic, I value effective communication and seek to make a positive impact through my work. I also find joy in video games, cycling through scenic routes, and embracing the great outdoors through camping adventures.

My Work

Some of my projects

White Matter Segmentation
Implemented an innovative self-supervised approach using Vision Transformers to accurately segment White Matter regions in brain images. This cutting-edge method surpassed traditional unsupervised techniques by 30% and achieved results comparable to supervised methods, with a further fine-tuning improvement of 7%. The project holds significant potential for advancing research in biomedical image processing and could have promising applications in medical diagnostics.
Technologies used: Python, Pytorch, Pandas, Numpy, OpenCV, WandB
Code will be released upon paper acceptance
OBAL: Medical Lab Management System
OBAL is a user-friendly Medical Lab System designed to streamline operations and manage clients and analyses efficiently. With distinct user roles (technician, manager, and secretary), it offers seamless coordination and secure data organization. Developed using Java, Swing, and MySQL, OBAL is a robust solution for medical laboratories, simplifying analysis results and receipt management. This project demonstrates expertise in software development and database integration, aiming to enhance the efficiency of healthcare processes.
Technologies used: Java, Swing, MySQL
Code
Stained Brain Image Dataset
Generated a large-scale dataset of over one million stained brain section images using the Allen Institute API for synthetic data generation and medical backbone training. This valuable resource provides researchers and developers with a diverse and extensive collection of brain images, contributing to the progress of various applications, including medical imaging, deep learning, and computer vision research.
Code will be released upon paper acceptance
Retinal Vessel Segmentation
Developed a highly accurate retinal vessel segmentation model using PyTorch and deep learning techniques. The model's primary goal is to aid medical professionals in diagnosing retinal-related diseases by precisely segmenting retinal blood vessels. Through rigorous testing and optimization, the model achieved an impressive accuracy rate of 81%, making it a promising tool for improving retinal disease diagnosis and treatment.
Technologies used: Python, Pytorch, Pandas, Numpy, OpenCV
Code
Fake Tweet Detection
Created a machine learning model to combat misinformation by identifying fake tweets. Employing a combination of algorithms such as Random Forest, Decision Trees, and Ensemble Learning, the model achieved an outstanding accuracy rate of 95% in distinguishing genuine tweets from fake ones. This project makes significant strides in supporting information integrity and countering the spread of false information on social media platforms.
Technologies used: Python, Scikit-Learn, Pandas, Numpy
Code
Animal Classification
Designed and implemented a deep convolutional neural network model for highly accurate animal classification. The model achieved an exceptional accuracy rate of 94% across six distinct animal classes. This project contributes to the field of image recognition and classification and holds potential for applications in wildlife conservation, zoology, and environmental research.
Technologies used: Python, Keras, TensorFlow, Pandas, Numpy
Code