The Artificial Neuron. 1 credit, 1 contact hour

Catalog description

Prerequisites: MATH 111 and CS 100 with a grade C or better. This course builds and understands one artificial neuron completely - the canonical computational unit underlying modern neural networks. Starting from raw data and ending with a working multiclass classifier, students implement every piece by hand: the linear score, the sigmoid, the loss function, the gradient, and the training loop. Rather than surveying many algorithms at a surface level, the course goes deep on one.