Over-parameterization for regularization is really old. The pseudoinverse min-no... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		pishpash on Feb 10, 2022 \| parent \| context \| favorite \| on: Computer scientists prove why bigger neural networ... Over-parameterization for regularization is really old. The pseudoinverse min-norm solution for under-determined linear systems even has that flavor.

tomrod on Feb 10, 2022 [–]

Sure, but that's identification approaches in econometrics and matrix analysis contexts. Using that for neural networks is new-ish in the zeitgeist, which did not exist in the 1990s as it does today.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact