Zero-shot Learning is an extreme form of transfer learning, meaning the model can directly make predictions in the second setting without labeled examples. Also known as Zero-data learning.

In zero-shot learning scenarios, consider three random variables- the traditional inputs x, the traditional outputs or targets y, and an additional random variable describing the task, T. If we have a training set containing unsupervised examples of objects that live in the same space as T, we may be able to infer the meaning of unseen instances of T.