STREAM'LINING THE LIMITS OF MACHINE LEARNING

Description:

This learning design attempts to apply the STREAM model to part of an introduction to machine learning, which focuses on machine learning hype versus technical limits.

The motivation and rationale is two-fold, first the for the topic itself, second for the use of STREAM:

A: New students of machine learning typically buy into much of the hype presented in the media and elsewhere, and as a result have unrealistic expectations. At the same time it is difficult to realize the limitations of machine learning independently if not specifically told where to look, and in many cases even skilled practitioners are also not aware of the real-world technical limits; and as a results may take on tasks or make promises that are simply technically impossible to fulfill.

B: A previous version of the learning design did not feature out-of-class activities, and as a result many students seemed to have difficulties relating cases covered in lectures to other and new cases (part of the ILOs). The STREAM model affords out-of-class activities to strengthen this aspect, for example by way of moderated forum discussions and Mentimeter assessments for learning.

Learning Design Structure:

1. Four short lectures provides background information, and sets up a number of general questions. The lectures are captured for later access and review.

2. In-class group discussions of specific lead-on questions. Lecturer floats.

3. Out-of-class forum discussion on new case study.

4. Out-of-class Mentimeter assessment for learning and evaluation of process.

5. In-class follow up with examples from forum discussion and Mentimeter results.