Im currently working on semi-supervised learning problems, and when Im confused of the idea that semi-supervised learning can use unlabeled samples to make predictions,
For example this picture, it aims to predict if the middle point is positive or negative, and it observes the other unlabeled samples(those gray ones) and label it with positive with high confidence.
How does it work? From my point of view, I don’t know why the middle point can be positive.