News for hackers. Stuff that matters.
Re:The truth about deep learning by ktRolster on Wednesday June 01, 2016 @10:26PM (#11819939)
>Neural nets are universal function approximators. There isnt something they can't learn.

Teach one to sort an arbitrary list. A universal function approximator is not a Turing machine.

Re:The truth about deep learning by riyadparvez on Wednesday June 01, 2016 @10:55PM (#11820057)
There is a new research going on in Neural Turing Machine which, in theory, can exactly do that. Maybe we have to wait some time to see NNs can sort numbers.
Re:The truth about deep learning by eva1984 on Thursday June 02, 2016 @12:19AM (#11820338)
On a high level, sort() is a seq2seq function, I think throw LSTM/RNN will yield some interesting results to this problem.
Re:The truth about deep learning by gradys on Thursday June 02, 2016 @12:44AM (#11820424)
Sibling comments have mentioned Neural Turing Machines, but even just regular recurrent neural networks are technically Turing complete[1].

[1] - http://lipas.uwasa.fi/stes/step96/step96/hyotyniemi1/

Re:The truth about deep learning by semisight on Wednesday June 01, 2016 @10:46PM (#11820022)
A Neural Turing Machine can learn programs.
Re:The truth about deep learning by return0 on Wednesday June 01, 2016 @10:31PM (#11819963)
ok i meant some function
Re:The truth about deep learning by umanwizard on Wednesday June 01, 2016 @10:52PM (#11820039)
sort is a function (in the mathematical sense)
Re:The truth about deep learning by return0 on Wednesday June 01, 2016 @11:28PM (#11820169)
then if it 's a continuous function it can be approximated
Re:The truth about deep learning by harryjo on Wednesday June 01, 2016 @11:39PM (#11820200)
and if it's not a continuous function?
Re:The truth about deep learning by return0 on Wednesday June 01, 2016 @11:47PM (#11820227)
https://en.wikipedia.org/wiki/Universal_approximation_theore...

It's proved for a subset of functions. Maybe you can prove it for more.