Do people in ML consider large, opaque models interesting from an academic/theoretical standpoint? What do we learn from them?
@mc Why wouldn't it be interesting? Major fields of science exist entirely to study opaque things, like biology and macroeconomics
@julesh @mc
https://mobile.twitter.com/ch402/status/1533164918886703104?lang=en
“The elegance of ML is the elegance of biology, not the elegance of math or physics. Simple gradient descent creates mind-boggling structure and behavior, just…
a Schelling point for those who seek one