When the software world wants to build stuff that works, it still mostly uses stuff that innately incorporates symbol-manipulation. Can someone please show me an operating system, a word processor, a spreadsheet or scaleable DB that is built with neural networks and no symbols?https://twitter.com/tyrell_turing/status/1003313227852181504 …
-
Symbols work well only in narrow domains were there's few ambiguities. A neural network outperformed a symbolic chess playing system (AlphaZero vs Stockfish). Perhaps you need to show me a symbolic system that can do state-of-the-art image or speech recognition.
for someone like me who advocates hybrid models, this is just a non sequitur; different tools for different jobs
-
-
Hybrid, as in, use NN/GA/ML/DL for converting an image/sound into a symbol so that a handcrafted logic in app s/w can use it as input to trigger the appropriate next step? Or is that something else? My doubt: "Would the NN itself apply the car's brakes too?"
@Grady_Booch -
Much of the best of DL is about signal processing; much of the best about symbolic computation is decision making and manipulation and memory and description and explanation. Thus from a systems engineering POV the two modes of computation can be components of a larger system
- 4 more replies
New conversation -
-
-
Different tools for different jobs means that certain tools are better in performing certain jobs. It does not imply that a different tool can't be used to perform jobs inefficiently. The brain does symbolic processing extremely inefficiently. Just compare it to a computer.
-
This Tweet is unavailable.
- 1 more reply
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.