#learning-paradigms

Question 1: Select all that apply

In which of these setups is the model trained self-supervised, with targets produced automatically from unlabeled data? Select all that apply.

Parts of each unlabeled photo are hidden, and the model learns to fill them back in1
The model sees the first frames of unlabeled video clips and learns to predict the next frame2
Customers are grouped into segments by how similar their purchase histories are3
A robot arm learns to grasp objects from a reward it receives after each successful attempt4