![Szilard [Deeper than Deep Learning]](https://pbs.twimg.com/profile_images/1016878602653085696/9RaALXe__normal.jpg)
Marcus has offered nothing to solve the problems of deep learning other than, "let's add symbolic AI to it." How lame is that. The problems with DL cannot be solved by modifying it. DL must be discarded altogether. Both Marcus and DL will be a footnote in the history of AI.
Deep learning, LSD, & Hallucinations. (Kidding about the LSD, not the hallucinations, which are serious problem.) https://arxiv.org/pdf/1809.02156.pdf …
-
-
-
Here's a problem with DL that Marcus has no clue how to solve in spite of his incessant pontificating. Unlike DNNs, the brain can instantly see a complex object it has never seen before, i.e., with no prior representations of it in memory. Can your GOFAI crap solve this?
- 1 more reply
New conversation -
-
-
(trolls aside) Gary, I think reported ‘hallucination’ is not a bad sign per se, it shows that there is a context associated to the perceived data. Humans also jump from association to association (like in Dennett’s Sweet Dreams book cover).
-
The key is being able to handle each of these associations independently: compositional representation of knowledge. We remove the information not applicable to a context. Dynamic inhibitory signals are nowhere to be seen in current ML models.
End of conversation
New conversation -
-
-
This Tweet is unavailable.
- 3 more replies
-
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.