http://zackmdavis.net/blog/2016/09/concerns-ii/ "you yourself admit that your model won't assign literally all of its probability mass to the exact outcome?!"
"category boundaries" were just a visual metaphor for talking about beliefs? There's a little naive Bayes model in my head with "blueness" and "eggness" observation nodes hooked up to a central "blegg" category-membership node, such that I can use observations to update my beliefs about category-membership, and use my beliefs about category-membership to predict observations. The set of things I'll classify as a blegg with probability greater than p is conveniently visualized as an area with a boundary in blueness–eggness space, but the beliefs are the important thing.
http://zackmdavis.net/blog/2016/09/concerns-ii/ "you yourself admit that your model won't assign literally all of its probability mass to the exact outcome?!"
"category boundaries" were just a visual metaphor for talking about beliefs? There's a little naive Bayes model in my head with "blueness" and "eggness" observation nodes hooked up to a central "blegg" category-membership node, such that I can use observations to update my beliefs about category-membership, and use my beliefs about category-membership to predict observations. The set of things I'll classify as a blegg with probability greater than p is conveniently visualized as an area with a boundary in blueness–eggness space, but the beliefs are the important thing.