Machine teaching studies the interaction between a teacher and a\nstudent/learner where the teacher selects training examples for the learner to\nlearn a specific task. The typical assumption is that the teacher has perfect\nknowledge of the task---this knowledge comprises knowing the desired learning\ntarget, having the exact task representation used by the learner, and knowing\nthe parameters capturing the learning dynamics of the learner. Inspired by\nreal-world applications of machine teaching in education, we consider the\nsetting where teacher's knowledge is limited and noisy, and the key research\nquestion we study is the following: When does a teacher succeed or fail in\neffectively teaching a learner using its imperfect knowledge? We answer this\nquestion by showing connections to how imperfect knowledge affects the\nteacher's solution of the corresponding machine teaching problem when\nconstructing optimal teaching sets. Our results have important implications for\ndesigning robust teaching algorithms for real-world applications.\n