Highly related: Bayesian Concept Learning (pdf).
Yeah, it’s also closely related to minimum-message-length prediction.
But this isn’t about learning from examples. Though man, the way I presented this article sure does seem like it’s about learning from examples. Whoops, bad pedagogy.
It’s about how, when we prove the number of smooth structures on R^2, we as humans automatically use that information to make predictions about the number of smooth structures on R^3. The question was—is there a basis for that that we can formalize?
Highly related: Bayesian Concept Learning (pdf).
Yeah, it’s also closely related to minimum-message-length prediction.
But this isn’t about learning from examples. Though man, the way I presented this article sure does seem like it’s about learning from examples. Whoops, bad pedagogy.
It’s about how, when we prove the number of smooth structures on R^2, we as humans automatically use that information to make predictions about the number of smooth structures on R^3. The question was—is there a basis for that that we can formalize?