Abstract: Intelligent Tutoring Systems are effective for improving students' learning outcomes. However, constructing pedagogically effective tutoring systems at scale has been widely recognized as a challenging problem. To address this problem, I explore the use of computational models of apprentice learning--or computer models that learn interactively from worked examples and correctness feedback--for supporting developers in building and testing tutoring systems. In particular, I present the Apprentice Learner Architecture, which makes it possible to generate and test alternative models of apprentice learning. I use this architecture to search for two kinds of models: (1) models that fit human behavior and (2) efficient models. Instructional designers can use human-like models as learner "crash test dummies" to simulate students interacting a tutor. To demonstrate this capability, I show evidence that my initial models correctly predict which of two fractions tutor designs will yield better student performance. Further, I compare simulated data to human data from this tutoring system to test which of my models have more human-like behavior. Next, I explore the use of an efficient model of apprentice learning for supporting non-programmers in authoring tutoring systems. Like humans, apprentice learner models can be taught by domain experts through worked examples and feedback. I present results showing that the time needed to author an Algebra tutor by interactively training an apprentice learner model is less than half the time needed to author a tutor using another state-of-the-art authoring-by-demonstration approach.
BIO: Christopher MacLellan is currently a PhD student at the Human-Computer Interaction Institute at Carnegie Mellon University. His work centers on the development of computational models of learning from demonstrations and feedback. He explores the use of these models for developing pedagogically effective educational technologies, such as intelligent tutoring systems and educational games, at scale. Additionally, he aims to leverage the human learning data collected in these environments to improve our understanding of the mechanisms that underly the flexible and general learning capabilities possessed by humans.