Semi-supervised assumptions

As explained in the previous section, semi-supervised learning is not guaranteed to improve a supervised model. A wrong choice could lead to a dramatic worsening in performance; however, it's possible to state some fundamental assumptions which are required for semi-supervised learning to work properly. They are not always mathematically proven theorems, but rather empirical observations that justify the choice of an approach otherwise completely arbitrary.