A PhD thesis on Semi-supervised learning with Graphs by Xiaojin Zhu. Focuses on creating graphs, based on a mixture of labeled and unlabeled data (as per the semi-supervised learning paradigm) and using processes on these graphs to propagate in rigo