Convolutional Neural Networks: Step by Step

Goal

implement convolutional (CONV) and pooling (POOL) layers in numpy, including both forward propagation and (optionally) backward propagation.

File Description

• .ipynb file is the solution of Week 1 program assignment 1
  • Convolution+model+-+Step+by+Step+-+v2.ipynb
• .html file is the html version of .ipynb file.
  • Convolution+model+-+Step+by+Step+-+v2.html
• .py file
  • Convolution+model+-+Step+by+Step+-+v2.py
• file
  • Convolution+model+-+Step+by+Step+-+v2.md

Snapshot

• Recommend read .ipynb via nbviewer
• computer view. open .html file via brower for quick look.
• brower view. Convolution+model+-+Step+by+Step+-+v2.md

Implementation

• Convolution functions, including:
  • Zero Padding
  • Convolve window
  • Convolution forward
  • Convolution backward (optional)
• Pooling functions, including:
  • Pooling forward
  • Create mask
  • Distribute value
  • Pooling backward (optional)