Optimal Teaching: The Inverse Problem of Machine Learning

What if there is a teacher who knows the learning goal and wants to design good training data for a machine learner? In this talk, I formalize the optimal teaching problem. In particular, I will focus on learners who employ Bayesian models. The framework is expressed as an optimization problem over teaching (training) examples that balance the future loss of the learner and the effort of the teacher. This optimization problem is in general hard. In the case where the learner employs conjugate exponential family models, I present an approximate algorithm for finding the optimal teaching set. The algorithm first optimizes the aggregate sufficient statistics, then unpacks them into actual teaching examples. I will give several examples to illustrate this optimal teaching framework.