Yes, "does the brain compute" is an uninteresting question b/c of course it does. The more interesting Q is to what degree the brain >
-
-
I think there's plenty of theory there, about units of processing, directionality of information flow, connectivity.
-
The people who build them say they have no idea. I think the high level theory is lacking IMHO.
-
Also no theory attached to the brain and yet the model captures human vision computations.
-
But they don't really have "no idea". That's just a turn of phrase. They do understand basic principles of CNN, just not how it all fits >
-
<That's true of all model authors to varying degrees, just more extreme with DNN authors. E.g. my comp neuro. models still produce surprises
-
These people aren't in cogsci. But I have a feeling we use understand and theory differently.
-
Some of the best (in terms of classification accuracy) models are just "add more layers". Eventually this heuristic might not help. Why?
-
That lack of understanding doesn't make then atheoretical, IMO. Noone understands all implications of their own models.
- 1 more reply
New conversation -
-
-
aieee and I got you into another long discussion sorry :)
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.