Representation Learning

about | blog | config | notes | github

1. What is it?

A paradigm of Machine Learning that focuses on learning representations of data. Representation Learning's goal is to essentially learn simpler features from data that have meaning. These features/representations can be used to make/solve complex decisions/problems.

2. Example

Suppose you have a dataset of dogs and cats and you wish to distinguish them. A Representation Learning algorithm would learn simple features which may answer simpler questions like "is there a paw?", "does it have pointy ears?", etc. These simpler features can be used then make the decision of whether the picture is a dog or a cat.

Created: 2021-11-13

Emacs 26.1 (Org mode 9.5)