Matrix shapes and calculations behind standard artificial neural networks

This educational tool developed at the University of Melbourne aims to give students a better understanding of the Python code for artificial neural networks and the matrix shapes and calculations behind that code. For each step through the artificial neural network (both the forward pass and back propagation) the corresponding code is given. Students are asked to write down for each step the dimensions of the matrix calculations, before checking their answers (correct answers are also given in this document).

The document is a PDF version of this elearning made in Articulate Rise. Please get in touch with the author if you would like to embed the elearning in your learning management system.