[ogrisel]

> does not need labeled data

It needed training on 1TB of labeled images in the first place. Arguably it can be used to transfer that knowledge to other tasks with a much smaller amount of labeled samples but still requires supervision.

[blauwbilgorgel]

Google trained a NN on unlabeled Youtube stills. It was able to detect/group/cluster pics of cats without ever seeing a label. This still needs supervision to teach the NN that whatever name it created for this cluster, us humans call this "cats".

If the error rate gets low enough, a NN could start labeling pics.

Finally, recent work has shown that running a dictionary through an image search engine can yield high quality labeled images automatically.

Aside: Thank you for contributing to sklearn. Really feel like I am standing on the shoulders of giants when I use that library.