Recurrent Internal Symmetry Networks by Backpropagation in Wallpaper Image Segmentation

Abstract

Internal Symmetry Networks are a new developed class of Cellular Neural Network inspired by the phenomenon of internal symmetry in quantum physics. The structures of the nets are based on irreducible group representations. Recurrent cycles are extended to the nets very recently. Dynamic recurrent cycles are proved to be effective by many experiments. In this paper, train them by backpropagation with noise to perform wallpaper image processing tasks.