Learning quickly from very few labeled samples is a fundamental attribute
that separates machines and humans in the era of deep representation learning.
unsupervised few-shot learning (U-FSL) aspires to bridge this gap by discarding
the reliance on annotations at training time. Intrigu