Developing a short undergraduate introduction to online machine learning

With the growing interest in data science, machine learning is becoming an important part of a well-rounded computer science curriculum. Although more and more schools are starting to teach machine learning and data science to undergraduates, these usually cover only supervised and unsupervised learning, causing students to graduate with a limited understanding of other important areas of machine learning. In this paper, we focus on one particularly exciting area, online machine learning. This area offers a fundamentally different perspective on learning, and has interesting theoretical underpinnings in addition to a large range of practical applications. Even though the fundamentals of this area are easy to understand, it is rarely taught and is not in the ACM/IEEE CS2013 curriculum recommendations. We present a short (roughly one hour) example unit which introduces online machine learning to undergraduates, and can be included at any point in an AI, data science, or machine learning course. (For supplementary material related to this unit see: http://grail.cs.washington.edu/projects/olteaching/.) A small pilot study with 13 student responses suggests that students find this unit engaging and highly valuable.