Normalisation is a term used in several areas of mathematics and AI and refers to ways to turn an input into a standardised form that is easier to deal with. Examples include:
- In data preparation turning date fields with diferent formats into a timestamp or an ISO standard date.
- In object recognition during image processing transforming the scale and orientation of a detected object so that it always ends up being presented to later stages in the same way.
- In statistical processing scaling variables so that they have the same standard deviation.
- In statistics this may also mean transforming a variable so that it has a tractable form, typically a Normal distribution or uniform distribution. This is also often called pre-whitening.
Used in Chap. 10: page 132; Chap. 12: page 178
Also known as: normalization
Used in glossary entries: image processing, Normal distribution, object recognition, pre-whitening, standard deviation, statistics, uniform distribution