Novel Complex Hopfield Neural Networks: Convergence Theorems

EasyChair Preprint 15492

Abstract

In  this  research  paper, a  novel  proof  of  convergence  theorem  associated  with  a  complex  valued  neural  network  based  on  complex  signum  function  is   proved.  Also,  two  novel  Complex  Valued  Neural  Networks (CVNNs) are   proposed. One  of  them  is  based  on  magnitude  and  phase  quantization  using  Ceiling  type  activation function  operating  on  rectangular coordinate  representation  of  complex net   contribution.  The  other  CVNN  is  also  based  magnitude  and  phase  quantization  using  Ceiling  type  activation  function  operating  on  polar  coordinate  representation  of  the  complex  net  contribution.  The  converence  theorems  associated  with  such  novel  CVNNs  are  also  proved.

Keyphrases: Complex Hopfield Neural Network, Magnitude Quantization, Phase Quantization, convergence theorem, stable states

@booklet{EasyChair:15492,
  author    = {Rama Murthy Garimella},
  title     = {Novel  Complex  Hopfield   Neural  Networks: Convergence  Theorems},
  howpublished = {EasyChair Preprint 15492},
  year      = {EasyChair, 2024}}