Convolutional Neural network CNN

Human vision normally passes through various layers where each layer does a certain task.
Definitions
Filter: a matrix that is passed over an image to identify certain features
Kernel: filter
Convolutional layers

Each node will act as a filter.
A filter detects very simple things such as an edge or texture.
A matrix of a specified size will then repeatedly look at each part of the image. (convolver)

The product of the filter by each matrix, is then stored on another matrix.

That new image on the right will then get passed to the next layer.
Kernel
- Matrix of weights
- A numerical filters that slides over the pixels of a picture
Application of filters
Each layer will apply one or may filters.
Each neuron is assigned a set of pixels from the main image and applies the filter to it.
Common structure
