From Misconceptions to Conceptual Change.
From Misconceptions to Conceptual Change.
复制标题
从误解到观念转变。
DOI：
10.1145/3287324.3287392
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
B. Metz
中科院分区：
文献类型：
--
作者：
Julia Gooding;B. Metz
机器学习(ML)由于其有用的应用和社会影响，已经成为跨学科学生理解的重要话题。同时，关于ML教育的现有研究成果很少，尤其是针对非专业ML教学的研究。本文对教学内容知识在非专业外语教学中的应用进行了探索。通过对非专业ML课程教师的十次访谈，我们询问了学生的成见，以及学生对学习ML有什么容易或难的地方。我们从学生面临的三种先入为主的观念和面临的五种障碍以及教师采取的六种教学策略来确定PCK。这些先入为主的观念被发现更多地关注ML的声誉，而不是其内部运作。学生的障碍包括低估人类在ML中的决定，以及将人类思维与计算机处理混为一谈。ML教学的教学策略包括策略性地选择数据集、手动遍历问题以及针对学生的领域(S)进行定制。当我们考虑这些发现的教训时，我们希望这将作为改善非专业ML教学的第一步。
Machine learning (ML) has become an important topic for students across disciplines to understand because of its useful applications and its societal impacts. At the same time, there is little existing work on ML education, particularly about teaching ML to non-majors. This paper presents an exploration of the pedagogical content knowledge (PCK) for teaching ML to non-majors. Through ten interviews with instructors of ML courses for non-majors, we inquired about student preconceptions as well as what students find easy or difficult about learning ML. We identified PCK in the form of three preconceptions and five barriers faced by students, and six pedagogical tactics adopted by instructors. The preconceptions were found to concern themselves more with ML's reputation rather than its inner workings. Student barriers included underestimating human decision in ML and conflating human thinking with computer processing. Pedagogical tactics for teaching ML included strategically choosing datasets, walking through problems by hand, and customizing to the domain(s) of students. As we consider the lessons from these findings, we hope that this will serve as a first step toward improving the teaching of ML to non-majors.