Every story and comment from the Hacker News API, in a layout borrowed from 2001.
A comment on Reverse Engineering Self-Supervised Learning. Parent comment
The naive statistical approach fails, the regularisation approach succeeds.
I'd say this does need explanation, but the paper really omits exactly what needs explaining... it isnt so much the introduction of regularisation...
Rather, it's why-it-should-be that our data generating processes encode metadata which can be capture by such regularisation
Phrased like this, i htink the answer is somewhat obviuos
There are no comments on this comment yet.
The naive statistical approach fails, the regularisation approach succeeds.
I'd say this does need explanation, but the paper really omits exactly what needs explaining... it isnt so much the introduction of regularisation...
Rather, it's why-it-should-be that our data generating processes encode metadata which can be capture by such regularisation
Phrased like this, i htink the answer is somewhat obviuos