.@Plinz if we unleashed DL on a Mandelbrot set, would it figure out the function? If not, why not?
Great talk btw.
https://media.ccc.de/v/35c3-10030-the_ghost_in_the_machine#t=2092 …
The question is what we mean by "Deep Learning". I think that it stands for "compositional function approximation", so the answer is yes, we can probably set up a configuration that figures it out. If we specify a given fashionable algorithm, I would be surprised.
-
-
Do you think our best current techniques could do it for the function in your talk? Has anyone tried? Would be really interesting to test algos on a randomly selected region and see what insights are obtained for layers above & below. An “ImageNet”-like challenge.pic.twitter.com/6Dr4PLTPhu
-
maybe you already answered, i.e. it’s unlikely a given fashionable algorithm could accomplish this. But in theory there’s no reason why not.
- 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.