3 ms·
They are arguing, that NNs are essentially polynomials, because they can approximate each other. With the same right you can argue, that NNs are eassentially fo
by beyondCritics 8y ago
They are arguing, that NNs are essentially polynomials, because they can approximate each other. With the same right you can argue, that NNs are eassentially fourier series, since under very mild conditions they approximate each other as well. Therefore this is not a valid argument. Is this some kind of joke? I have checked the date, but that gives no clue.
- geezerjay 8y ago> With the same right you can argue, that NNs are eassentially fourier series, since under very mild conditions they approximate each other as well. Actually that's very wrong. Some classes of NN (step function, ReLU, etc) do define piecewise polynomial functions, albeit in a contrived way when compared with simply generating a spline. So the functional space is exactly the same.