My disconnect with your argument for the need of "symbol manipulation" is that you need it to be innate machinery and not something that is simulated by a neural network. RNNs are known to be Turing complete. What's the argument that symbol manipulation cannot be done by DL?
if machine translation systems could *answer questions* about what they translated without symbol-manipulation, I would be genuinely impressed. instead they sometimes can't even keep track of coreference from one sentence to the next.https://twitter.com/intuitmachine/status/952843100443041793 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
What do you then say about the latest news from Alibaba and Microsoft: http://www.zdnet.com/article/alibaba-neural-network-defeats-human-in-global-reading-test/ … ? Are symbol manipulation programs better than Deep Learning programs? If false, then why even need symbol manipulation at all?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.