The goal of this experiment is to train a neural network on an imbalanced dataset where one class is underrepresented compared to the other. The training process includes the use of synthetic data, logging of key metrics using Weights & Biases and visualisation such as precision-recall curves and confusion matrices.