It would be neat to keep a list around a list of "simple problems where neural nets can't learn a generalizable solution, even with lots of training data". There are many of them, and they could serve as a benchmark for future progress.https://twitter.com/GaryMarcus/status/957685727969660929 …
-
Show this thread
-
The gist of it is this: neural nets do *pattern recognition*, which achieves *local generalization* (which works great for supervised perception). But many simple problems require some (small) amount of abstract modeling, which modern neural nets can't learn
6 replies 62 retweets 232 likesShow this thread -
Replying to @fchollet
Can Capsule Networks form part of a solution to address the need for abstract modeling?
1 reply 0 retweets 0 likes
Replying to @DoomscrollPro
As far as I can tell, no. Capsules address limitations of CNNs but do not address the overarching limitations of deep learning.
11:26 AM - 28 Jan 2018
0 replies
0 retweets
6 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.