Morphological regularization neural networks

Enhanced FMAM Based on Empirical Kernel Map

“…N recent years, a number of different morphological neural network models and applications have emerged [1]- [9]. Morphological neural networks can be characterized as neural networks which calculate a maximum or a minimum of a sum at each node.…”

Morphological perceptrons with competitive learning: Lattice-theoretical framework and constructive learning algorithm

“…Consequently, minimax algebra has granted researchers an easy access to the problem of defining the weights of a new type of neural network. These weights correspond to certain sets called structuring elements in the terminology of MM and afterwards other MNNs have been directly developed on the set-theoretical basis of MM [25,40,46].…”

“…In this paper, we will explain that fuzzy lattice neurocomputing (FLN) models such as fuzzy lattice neural networks (FLNNs) and fuzzy lattice reasoning (FLR) classifiers [35][36][37]47] are closely related to MNNs. Morphological and hybrid morphological/rank/linear neural networks [46] have been successfully applied to a variety of problems such as pattern recognition [37,40], prediction [1,3,70], automatic target recognition [39], handwritten character recognition [46], landmine detection [25], self-localization, and hyperspectral image analysis [30,49,55].…”

Permutation-based finite implicative fuzzy associative memories

“…The morphological perceptrons [49,51,57] and the morphological associative memories (MAMs) [50,47,64] are also examples of MNNs. MNNs have been applied in a variety of problems including classification [51,57], automatic target recognition [23], landmine detection [8], and hyperspectral image analysis [53,52,11]. In few words, MNNs are artificial neural networks whose neurons perform elementary operations of mathematical morphology (MM) such as dilations and erosions [3,15,38,55,56].…”