Trained and deployed an hybrid pretrained EfficientNetB6 and Bidirectional LSTM model to predict whether the lesion in given image is benign or malignant. Dataset used was obtained from The International Skin Imaging Collaboration (ISIC) 2020 Challenge dataset.
Implemented gradient boosting regression model and neural network to predict vehicles' miles per gallon (MPG) values, which is the primary measurement of a car's fuel efficiency. Auto MPG dataset was used which consists of a description of many automobiles in the late 1970s and early 1980s.
Explored stroke prediction dataset and implemented a regularizing gradient boosting model to predict the likelihood of a patient having a stroke or not considering input parameters like gender, age, various diseases, and smoking status. The model was deployed using Streamlit built web app deployed on Heroku.
Built a simple web app that classifies different Toyota brands using Streamlit. Images for different Toyota models were gotten using Bing search API. After image cleaning and pre-processing, a pre-trained Resnet101 model was implemented classify the images using FastAI.
Created and deployed a web app that predicts whether an individual would take up a health insurance policy or not leveraging a machine learning classification model. Factors that most likely influence taking up a health insurance policy by an individual were also investigated. Dataset used was the Individual Recode section of the 2018 Nigerian Demographic and Health Survey DHS dataset.
Through hands-on learning and class challenges/assignments, I'm training twelve primary school students (Grades 4-6) in programming using Scratch. Completed a catch-an-apple Scratch project with the students, and we are working on more projects.
I volunteered as a group tutor for PWSAfrica, a 2-week workshop aiming to empower scientists in Africa with computer programming skills. Handled a break out group during coding sessions where I helped participants practise what was learnt, from introduction to Python to Machine Learning. My group worked on a Book Recommender project at the end of the workshop.
Volunteered as a tutor for the AI Invasion program, a 5-day program to introduce participants to core concepts of Machine Learning, and hands-on practice of solving Machine Learning problems using Python. Introduced participants to Machine Learning hackathons and we participated in a 4-day Kaggle hackathon.
Abstract: Current rapid development in Artificial Intelligence (AI) provides a vast selection of high-quality tools to solve complex problems in more efficient ways than before. As a consequence, many fields of science and engineering are starting to explore AI tools, especially Deep Learning (DL) models for computer vision, audio and video understanding, speech recognition and decision making. In this project, we studied a type of deep learning model - Convolutional Neural Networks (CNNs), starting with a basic Machine Learning algorithm - Logistic Regression, moving on to building blocks of Neural Networks and method of training and optimising neural network models. We then implemented three different CNN architectures; a base model, VGG16 and ResNext-50, to classify different plant diseases using healthy and diseased leaf images. The best performing model was ResNext-50 with 98% accuracy.