Deep-layer limit and stability analysis of the basic forward-backward-splitting induced network (II): learning problems

要約

Deep unfolding neural networks derived from iterative optimization schemes and numerical ordinary/partial differential equations (ODEs/PDEs) have attracted much attention in data science over the last decade. Therein, numerous important network architectures were constructed from the basic forward-b…