Abstract
This article presents a summary of the doctoral dissertation of the author on the topic of semi-supervised learning for predicting structured outputs.