@halhod this is one of the two critical intellectual/scientific questions that was distorted by the media blitz and paper framing. (the other is innateness, viz the solving part was portrayed as learned, though actually innate.)
Show this thread
Dear #deeplearning fans,
The critical question, as I have said, over and over again, is not whether deep learning is useful, it's whether or not we need to move towards hybrids.
You can't keep portraying hybrids as pure deep learning systems, and claim to have won the argumenthttps://twitter.com/GaryMarcus/status/1184460444037025792 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Your appeals to "hybrid" tend to suggest there are only two approaches to integrate; and that the other major approach is something like "symbolic reasoning" in the gofai tradition. Is that a fair summary of your position?
-
No, but it's a very good question. I've pressed particularly for integrating symbol-manipulation, but human intelligence likely weaves together multiple threads, and I suspect that AGI will need to, as well. (Also, I am pressing for symbol-manipulation, but not GOFAI per se.)
- 1 more reply
New conversation -
-
-
Dear Gary Markus, 1. trying to win arguments on how to name things is not a viable scientific objective. 2. Deep networks with inductive biases are still deep networks. Else we would have called LeNet 'Neural network with shift invariance from signal processing'.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
“Search and learning are the two most important classes of techniques for utilizing massive amounts of computation in AI research.” http://www.incompleteideas.net/IncIdeas/BitterLesson.html …
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.