This document outlines a dissertation defense for a Ph.D. in machine learning with incomplete information. The defense covers machine learning concepts, applications, and the student's work in active learning and budgeted learning. For active learning, the student's algorithms focus on selecting labelers and estimating ground truths rather than just selecting instances. For budgeted learning, the student develops algorithms that exploit dependencies between features in Bayesian networks to learn more accurate classifiers within a given budget.