On fuzzy neuron models

Abstract

In recent years, an increasing number of researchers have become involved in the subject of fuzzy neural networks in the hope of combining the strengths of fuzzy logic and neural networks and achieving a more powerful tool for fuzzy information processing and for exploring the functioning of human brains. In this paper, an attempt has been made to establish some basic models for fuzzy neurons. First, several possible fuzzy neuron models are proposed. Second, some learning (training) and adaptation mechanisms for the proposed neurons are given. Finally, the possibility of applying non-fuzzy neural networks approaches to fuzzy systems is also described.

Integration of particle swarm optimization-based fuzzy neural network and artificial neural network for supplier selection

“…However, only numerical or crisp data is suitable for the aforementioned fuzzy neural networks (FNNs) and the knowledge of experts is always of linguistic type. Thus, Gupta and Qi [31] presented some models with fuzzy neurons but no learning algorithms within. Thereafter, a number of publications were cited in a survey paper [32], the authors discussed the learning algorithms and applications for FNNs with fuzzy inputs, weights and outputs [33].…”

An Investigation of a Linguistic Perceptron in a Nonlinear Decision Boundary Problem

“…Gupta and Qi [1] introduced basics models for fuzzy neurons. They did not show any calculation or perform any experiments.…”

“…Klir Since, for any of these operations, [A⊗B] α is a closed interval for each α∈(0, 1] and A, B are fuzzy numbers, A ⊗ B is also a fuzzy number. For the collection of linguistic vectors (or type 2 fuzzy set) where each component is non-interactive, Mares [26] has shown that with 0 (the vector of singleton fuzzy number 0), 1 (vector of singleton fuzzy number 1), and componentwise addition and scalar multiplication, this forms a vector space. Also, with appropriate definitions of distance, these spaces exhibit the properties of metric spaces [21,27].…”

Neural activation ratio based fuzzy reasoning

“…Pacr(l,3) and Pacr (3,2) represent the antecedent parts of those rules (define input variables as IN1 and IN2) and pcr(l,3) and pcr(3,2) the respective consequents. Take pcr( 1,3) and pcr(3,2) as 0.1 and 0.9 and defined respectively as "low" and "high"( variable OUT).…”

“…fuzzy parameters instead of crisp weights) takes place with the objective of achieving the traditional properties of flexibility and robustness in the neural network, [3], [4], [8], [lo]. Other architectures where both neural nets and fuzzy systems cooperate, more or less independently, have also been proposed in a hybrid way, [2], [l].…”