3 ms·
But this is true of neural nets as well, so I'm not sure why the author was trying to contrast them. Essentially it is a question of do you want a system that w
by gaius 10y ago
But this is true of neural nets as well, so I'm not sure why the author was trying to contrast them. Essentially it is a question of do you want a system that works well but that no-one understands or one that works not so well but is understandable? But if you meta it, it's the same.
- backpropaganda 10y agoI don't think even this dichotomy is real. Deep learning methods are both performant and elegant.
- baq 10y agoThey're elegant only if the sentence 'consider a model with a billion features' doesn't make you feel disgusted, I'm afraid.
- backpropaganda 10y agoI sure hope you're not similarly disgusted by "consider a computer with a billion transistors". Computer Science is full of abstraction and complexity. Denying it to machine learning seems conservative and naive at best.
- trevyn 10y agoelegant, adjective: (of a scientific theory or solution to a problem) pleasingly ingenious and simple. One easily-described approach that is state-of-the-art in a massive number of domains? That's pretty f'ing elegant.