These are the slides and notebooks, of the course Easy Nets, taught at the School of Molecular and Theoretical Biology the summer of 2021 (held on-line due to COVID).
Neural Networks are the current most-widely used machine learning tool to model real life problems. They can be used to solve regression problems, such as predicting whether stock price of a company will increase tomorrow, or classification problems, such as recognizing faces in an image. With knowledge of multivariate calculus and optimization, students learnt and understood the mathematics behind these structures in simple examples. In the last two session, we held a team competition to see who was able to build the most efficient neural network to solve some of the common classification datasets: MNIST, CIFAR10, SVHN...