Active learning of introductory machine learning

This paper describes a computer-based training program for active learning of Agent Technology, Expert Systems, Neural Networks and Case-Based Reasoning by undergraduate students using a simple agent framework. While many Machine Learning (ML) and Artificial Intelligence (AI) courses teach ML and AT concepts by means of programming assignments, these assignments have usually no connection to how the student will apply the newly obtained knowledge to previously unseen, real-world problems. The pedagogy that we adopted here is computer-based active learning: teams of students are presented with well-defined assignments aimed at building intelligent agents for person identification and recognition of facial expressions and emotions from video recordings of their faces. Classroom experience indicates that the students found the specified programming assignments highly motivating. Objective evaluation studies suggest that students learn much more effectively when a contextualized, collaborative, constructive, and reflective approach is used than when an orthodox, objectivist approach to teaching ML and AI techniques is used alone.