And unsurprisingly even very simple programs struggle with perturbations. Simple Turing machines for example don’t always recover from simple perturbations. Indicates more complex structures won’t fair any better.
the challenges to deep learning just keep on coming. here’s an important technical paper by Sengupta and Friston showing that accuracy doesn’t always entail robustness. (corroborating some of the issues on my own critical appraisal of #deeplearning.) https://arxiv.org/abs/1804.11313
-
-
-
Don’t think any simple machine is going to be robust across sufficiently complex range of circumstances. That’s why humans mix bottom-up and top-down info; System I & System II (reflex and deliberative), etc
- 1 more reply
New conversation -
-
-
Gary, debunking DL myths is healthy for the ecosystem, and your contribution is helpful, to a degree. But, beyond debunking, do you have anything constructive to offer?
-
Sure, e.g. I make specific suggestions about innateness and aspects of symbolic architectures that I believe need to be incorporated in ML, in both of my January arXiv articles
- 1 more reply
New conversation -
-
-
Brilliant Gary!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Gary, you're my hero.
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.