This theorem is a very important one in the neural nets theory.
Indeed, in 1989, George Cybenko proved that a single hidden layer feed forward neural network can approximate any continuous and multivariate function.
Moreover, he also proved that a failure with such neural nets would be caused by bad weights or learning rate, etc.
This theorem helps [...]
Artificial Neural Networks are considered as an obscure Artificial Intelligence technique.
However, they’re very used in many domains :
Automated pilot for planes
Pattern recognition
Medical diagnosis
Anything related to classification
More generally, we can approach any function that complies to some conditions (see the Cybenko Theorem)
I’ve written an article about feed-forward artificial neural networks on my French personal websiite on [...]
