In plain words: A technical tutorial walks through the three most common neural network types — feedforward, convolutional, and recurrent — explaining their basic building blocks. It then derives the forward pass and the backpropagation update rules step by step, so readers can see exactly how these networks learn.

Abstract

This note presents in a technical though hopefully pedagogical way the three most common forms of neural network architectures: Feedforward, Convolutional and Recurrent. For each network, their fundamental building blocks are detailed. The forward pass and the update rules for the backpropagation algorithm are then derived in full.

Thomas Epelbaum
arXiv:1709.01412 · stat.ML, cs.LG · submitted Sep 5, 2017 · updated Sep 11, 2017
abstract · pdf · The content of this note can be found on GitHub: https://github.com/tomepel/Technical_Book_DL . If you detect any typo, error, or feel that I forgot to cite an important source, don't hesitate to email: [email protected]. v2: minor typos corrected