I said false but in attempting to explain my answer, I realized I am unsure what you mean by a "computational principle". Examples?
-
-
-
I’m hesitant to answer because Twitter has a way of turning these kinds of conversations into epistemological food fights. Basically, I mean math-like explanations that approximately (precisely?) describe why/how the interactions of all the parts give rise to computation.
- Još 2 druga odgovora
Novi razgovor -
-
-
Don’t brains do backprop?
-
Don’t be evil.
Kraj razgovora
Novi razgovor -
-
-
Not even obviously useful, much less requires. Required is understanding better what brains are even doing, so we could hope to build better models
-
So False, then.
Kraj razgovora
Novi razgovor -
-
-
I voted false. Because it is probably impossible. Beyond our well known design principles.
- Još 14 drugih odgovora
Novi razgovor -
-
-
I say true, in the following (slightly weakened) sense: Understanding the computational principles underlying artificial deep neural networks is likely (a) easier than and (b) a useful stepping stone toward understanding the computational principles underlying brains.
-
Understanding deep nets may not technically be a prerequisite for understanding brains, but a claim to understand some species’ brain without having a simulation (which would likely qualify as a sort of deep net) and an understanding of that simulation would not be compelling.
- Još 3 druga odgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.


