ML approach that focuses on learning useful representations of data without explicitly contrasting positive examples against negative examples.
Generality: 525
