I think we will achieve gradually stronger AI in the future by blending pattern recognition ("deep learning etc.") with symbolic AI ("programming etc.") in increasingly automated ways. Details: https://blog.keras.io/the-future-of-deep-learning.html … The best applied AI systems today are still hybrid systems.https://twitter.com/GaryMarcus/status/1071438612690960384 …
-
-
There's an interesting discussion to be had about where to the put the boundary between programs (modular, sample-efficient learning, independently trained, maybe even interpretable) and standard deep nets (none of above, but resilient to noise, no domain knowledge required).
-
We also found that combining symbolic & neural approaches works quite well for very challenging search problemshttps://www.nature.com/articles/nature25978 …
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.