Which statement best describes the training setup in unsupervised learning?

Concept

Learning methods differ by whether training examples include target labels. Supervised learning maps inputs to known targets, while unsupervised learning discovers structure in inputs without target labels.

Application

Here the defining condition is that the output or target is not supplied and the data are unlabelled. The algorithm therefore searches for patterns, groups, or latent structure within the input features.

Contrast

• Known output: an explicit target is available, which describes supervised rather than unsupervised training.

• Labelled data: each input is paired with a target label, again describing supervised training.

• Unknown output and unlabelled data: no target label is supplied, so structure must be learned from the inputs.

• Every step supervised: external target signals guide the updates, which is the opposite learning setup.

Therefore, the applicable description is: Output is unknown and unlabelled.