4 ms·
Why couldn't mathematical proofs/models have predicted or revealed this to be the case back then?
by Solvency 3y ago
Why couldn't mathematical proofs/models have predicted or revealed this to be the case back then?
- actionfromafar 3y agoHow??
- nurbsnn 3y agoOn the contrary, there was a mathematical proof that one-hidden-layer neural network with nonlinearity is enough to represent any function. Using more than 1 hidden seemed a waste. https://en.wikipedia.org/wiki/Universal_approximation_theorem https://en.wikipedia.org/wiki/Universal_approximation_theore...