Argument mixes two aspects. First, and correctly, brain is super complex and can not be demixed. Second, and disagreeable, no representation. Complexity is not by itself a reason against representation. It is, however, a good argument against the misleading hubel+Wiesel ideas.https://twitter.com/GunnarBlohm/status/1028982136282533888 …
-
-
But what can we say meaningfully about micro-causality? We have the full micro-causality of, say, resnet, or an RNN, but can we really say much about how it computes? What kind of insights will we obtain?
-
We can absolutely talk about computation (i.e. information transformations) that come about in recurrent networks, as a function of their connectivity and dynamics. They are mechanistic, causal descriptions of information transformation.
End of conversation
New conversation -
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.