One hot encoding is a method used for converting categorical features into numerical features with binary values (0 or 1) in which only one category in the feature set is 1 and the rest are 0

This technique is often used when dealing with text or categorical data. For example if we're labelling a colour as either red, green or blue we could encode the labels for each colour as for red, for green and for blue.

This type of encoding is useful when using the categorical cross-entropy loss function as it allows for a simplified and faster version to be used.