My background knowledge: Basically, supervised learning is based on labeled data. Using the labeled data, the machine can study and determine results for unlabeled data. To do that, for example, if we handle picture issue, manpower is essentially needed to cut raw photo, label on the photos, and scan on the server for fundamental labeled data.

I know it sounds weird, but i'm just curious if there are any algorithms/system to make a label automatically for supervised learning.


Supervised learning needs labeled data.

Unsupervised learning finds useful representations of the data from unlabeled data.

Semi-supervised learning uses both, labeled and unlabeled data.

If you could generate labels without anything, why should you train a classifier?

| cite | improve this answer | |
  • $\begingroup$ Thanks for your comment. i have an additional question. how about FaceID? it doesn't need any labeled data. It just automatically take a photo and crop man's face and assign identification. Based on the id(it can be the labeled data automatically itself.), service(machine) can get smarter. In this case, what am i missing? i am confused. $\endgroup$ – Sean.G Feb 26 '18 at 6:44
  • $\begingroup$ "how about FaceID? it doesn't need any labeled data." - wrong. It for sure had a lot of labeled data during development. And when you use it, it also gets at least one labeled image of you. It belongs to the semi-supervised group. $\endgroup$ – Martin Thoma Feb 26 '18 at 6:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy