Describing face outlines with uncertainty
I am helping prepare an adult learning session about automated face categories. The goal is to ask better questions about classifications without ranking anyone’s appearance. A topic reference I have saved for discussion is what is my face shape because the wording makes the classification question explicit. I have not evaluated its output. Before looking at a label, learners could list the proportions they think a method uses. A second discussion could examine the information missing from one photograph. No participant would need to upload a personal image; a discussion of example descriptions would be sufficient. Each group could explain which evidence would change its initial interpretation. We could also compare two written descriptions of the same outline and ask whether their labels mean the same thing. This would make the assumptions available for discussion without requiring a supposedly correct personal result. What questions would make this a useful lesson in evaluating claims rather than another appearance quiz?
頁:
[1]