In case anybody is in #Cyprus and interested in my work, I'm giving a talk at @RISECyprus on the 16th of July at 10am on: What do deep neural networks know?
More here:https://dx.doi.org/10.1101/626374
I mean it's back-prop so there are top-down influences from the output layer, but I assume you mean if there were top-down influences from a more categorical-like structure like stuff like a label for mammals?
-
-
Those doing such work seem find that indeed you get even more human-like categorisation/similarities, e.g., https://papers.nips.cc/paper/7780-hyperbolic-neural-networks.pdf …
- 1 more reply
New conversation -
-
-
the paper mentions not modeling task-related influences, that's mostly what i'm thinking of :)
-
Ah, attention! Yes, there is work on that too being done in lab. I'm not directly involved but keep your eyes peeled for their preprint.
-
cooool :D excellent work here, those figs show exciting results!
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.