Tutorial

This is a CNN tutorial. This tutorial contains some minimal explanations of the types of what CNNs are and some discussions on the types of layers and so on. This is not detailed enough to be a complete guide, but just, so that it could be used as a recap. The tutorial is more implementation centric. Whenever possible code snippets are provided and comparisons made between theano and tensorflow to help migrants.

The tutorial is organized in such a way that the reader should be able to go article-by-article by clicking the next button at the end of each article. Although there is some ordering in the tutorial, one may skip sections that they are aware of and simple navigate to the section of interest as each one is written independently of the others. Have fun!

• Introduction
• Dot-Product Layers
  • Implementation
• Convolutional Layers
  • Implementation
• Pooling Layers
  • Implementation
• Softmax Layer
  • Implementation
• Reshaping layers
• CNN Architecture Philosophies
• Normalization Layer
• Dropout Layers
  • Implementation
• Network
  • Dataset
    • Fashion-MNIST
  • Network Architecture
  • Cooking the network
• Trainer
• Run and Outputs
  • Tensorboard
    • Scalars
    • Images
    • Graphs
• Additional Notes
  • Stochastic Gradient Descent
  • Off-the-shelf Downloadable networks
  • Distillation from downloaded networks
• Bibiliography