Shift invariance (Spatial invariance)

From Computer Science Wiki

Exclamation.png This is student work which has not yet been approved as correct by the instructor

Case study notes[1]

Introduction[edit]

Shift Invariance simply refers to the 'invariance' that a CNN has to recognising images. It allows the CNN to detect features/objects even if it does not look exactly like the images in it's training period. Shift invariance covers 'small' differences, such as movements shifts of a couple of pixels.

 <ref> https://stats.stackexchange.com/questions/121703/what-does-shift-invariant-mean-in-convolutional-neural-network </ref>
 

How does it work or a deeper look[edit]

  • Due to pooling/max pooling it is acceptable that shift invariance only covers such small changes. This is because pooling already strips the image away of it's useless features, and gives a compressed version of the input. As such if there is 50 pixels of white space to the left in one version of the input, and 70 pixels in another version, the image that gets processed would be the same, as pooling would get rid of the white space

Examples[edit]

Please include some example of how your concept is actually used. Your example must include WHERE it is used, and WHAT IS BENEFIT of it being used.

Pictures, diagrams[edit]

Pictures and diagrams go a LONG way to helping someone understand a topic. Especially if your topic is a little abstract or complex. Using a picture or diagram is a two part process:

  1. upload a file
  2. use the file on a wiki page

External links[edit]

  • It would be helpful
  • to include many links
  • to other internet resources
  • to help fellow students
  • Please make sure the content is good
  • and don't link to a google search results, please

References[edit]