Machine Learning for High School Students

In this study, we developed a machine learning method for object recognition that can be implemented using knowledge that high school students attain during their normal math and IT classes. We then tailored a two-hour interactive lesson in which the students were divided into groups to implement solutions to six distinct problems required by the method. The solutions were later put together by the teacher into a working web application (HTML + JavaScript). The lesson was taught on three occasions in Romanian schools to students between 13 and 19 years old. The students were excited about the lesson, and the collected data measuring students' intrinsic motivation suggests that the given tasks and the type of instruction were motivating them. The students also found the lesson achievable regardless the level of their previous programming background. The students were even able to suggest viable improvements to the method. The lesson is presented in short in this (17 minute) YouTube video1. Furthermore, we utilized the developed machine learning tool in a workshop with primary school children. Observations from this workshop suggest wider applicability of the tool, as well as further research questions on machine learning in K-12 settings.