ah but neural networks are universal function approximators! (proceeds to ignore the size of network needed for an adequate approximation and/or how much data would be required to train it)