Why do you think physicists have created a monoculture? Because no one told them "this technique isn't unique to quantum mechanics"? It really is a bit odd for a physicist to claim that since pre-quantum era physicists already used it.
-
-
-
I meant that “AI” has mostly narrowed to fiddling with backpropagation hyperparameters. Sometimes just empirically, but if there’s insight to be had, it’s from thinking about the energy surface and stuff. There’s value in this, but there’s a lot of problems it doesn’t address.
- 2 more replies
New conversation -
-
-
I will have to disagree. Many of the concepts in physics such as non-linear dynamics, phase transitions and criticality, renormalization, complex systems etc aren't in the vocabulary of deep learning research.
-
Most papers that are submitted with physics inspired approaches are rejected from conferences. There is indeed a mono culture, but it's not due to a bias in physics. Maybe physics envy.
- 17 more replies
New conversation -
-
-
Most of the crankiest people I know about the culture of AI are physicists-turned-AI people. I doubt that's a reason for a monoculture; seems quite the reverse, they're usually the ones saying "hey, fiddling with hyperparameters is the wrong thing to be doing..."
-
Yes, and "let's look at the energy surface" leads to insights that vaguely-motivated tweaking can't. (Such tweaking being an approach CS-educated folks are liable to fall into.) But it doesn't lead to the insights that understanding mechanism-domain interactions can.
- 6 more replies
New conversation -
-
-
I partially agree with the latter point (former physicist checking in) - but - I don't think that's a reasonable inference you can draw from the article. I think it has more to do with the lack of mathematical training of the journalists that cover this stuff.
-
Yes, I agree. I was making the "monoculture" point on the basis of my take on the direction of the field overall in the past ~5 years, rather than drawing on the article specifically.
End of conversation
New conversation -
-
-
Seems like you could also argue that calling SVD a "quantum mechanical" technique is a sign that *not enough* physicists have gone into AI yet.
-
Pretty sure it's just cheap journalism.
End of conversation
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.