People are way more spoof-proof, at least in certain perceptual tasks. Good find @GaryMarcus ! Will reference this piece in my upcoming panel in 5 minutes. Thx for the Just-in-time case study on the limits of #DL.https://twitter.com/GaryMarcus/status/1042877884627542016 …
-
-
Replying to @hjameswilson @GaryMarcus
Caution: This is not an inherent limitation of DL. I believe it can be fixed fairly easily.
1 reply 0 retweets 0 likes -
Replying to @tdietterich @hjameswilson
you think spoofing can be fixed easily? or just this particular version? there have been at least a couple dozen independent versions of spoofing demonstrated.
1 reply 0 retweets 0 likes
Don't forget that human perception maps every single pattern to the same global model (a single dynamic 3space), which is a constraint that is absent in most current image recognizers
8:24 PM - 20 Sep 2018
from Cambridge, MA
0 replies
0 retweets
2 likes
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.