|
@neuroecology | |||||
|
Does AI think this is a duck or a rabbit? reddit.com/r/dataisbeauti… pic.twitter.com/gnegEPJJbV
|
||||||
|
||||||
|
Julie J. Lee
@synapticlee
|
7. ožu |
|
It depends on the network, @deworrall92 and others have done nice stuff on rotation equivariance e.g. deworrall92.github.io/2017/06/01/har… :)
|
||
|
|
||
|
Adam J Calhoun
@neuroecology
|
8. ožu |
|
still, it looks like its inference roughly corresponds with mine (except when the 'duck' is upside down - need to do a better job with those priors). We certainly aren't rotationally equivariant!
|
||
|
|
||
|
Adrian Hutchins
@OKCUCardGuy
|
7. ožu |
|
|
||
|
Ronnie Hughes
@asenseofplace1
|
9. ožu |
|
It’s been such an intense kind of sociology week @LivUniSoc that here’s Wittgenstein turning up in my Twitter feed first thing on a Saturday morning? Think I’m still dreaming.
|
||
|
|
||
|
Tom Murtha
@tomemurtha
|
9. ožu |
|
The only two words I remember from those who did sociology at @GoldsmithsUoL were phenomenology and ethnomethodology.
|
||
|
|
||
|
Michael Shadlen
@shadlen
|
9. ožu |
|
The answer is simple: Yes!
|
||
|
|
||
|
Alison Niedbalski
@qualiaisreal
|
8. ožu |
|
But what about when the duckrabbits are going at it like bunnies? pic.twitter.com/nAS3243CFw
|
||
|
|
||
|
Rakesh Bhandari
@postdiscipline
|
7. ožu |
|
Dammit, so which one is it? :)
|
||
|
|
||
|
Rakesh Bhandari
@postdiscipline
|
7. ožu |
|
We can't easily say it's both because they're incompatible with each other. But to then say that it is neither a duck nor a rabbit misses something, as either interpretation of the black &white dots is an adequate interpretation. Aristotle's logic replaced by a logic of ambiguity
|
||
|
|
||