Frequency domain criteria for cellular neural networks

A dynamical system is called globally asymptotically stable if it has a unique equilibrium point which attracts every trajectory in state space. As a consequence its steady state response is insensitive to initial conditions and then depends only on the input. In this paper some criteria are presented for the global asymptotic stability of cellular neural networks (CNNs), concerning both discrete-time and continuous-time dynamics. The proposed criteria represent necessary and sufficient conditions that can easily be checked by computing the discrete Fourier transform of the template elements. For this reason they have been called frequency domain stability criteria. These criteria provide milder constraints on the template coefficients than required in existing results for general recurrent neural network models. © 1997 by John Wiley & Sons, Ltd.     

Equilibrium analysis of non-symmetric CNNs

Two useful results concerning the equilibrium analysis of non-symmetric cellular neural networks (CNNs) are presented. First a new sufficient condition ensuring the existence of a stable equilibrium point in the total saturation region is given. Then another condition which guarantees the uniqueness and global asymptotic stability of the equilibrium point is obtained.     

External inputs,stable equilibria and complete stability of CNNs

We construct two cellular neural networks (CNNs) of three cells to show that a CNN can have stable equilibria, but is not completely stable and that the complete stability of CNN depends on the choice of external inputs. These phenomena cannot occur for two‐cell CNNs. Copyright © 2003 John Wiley & Sons, Ltd.     

On the dynamic behaviour of two-cell cellular neural networks

A complete study of non-linear differential equations describing two-cell cellular neural networks (CNNs) is presented. the stability properties are investigated and the domains of attraction of the stable fixed points are determined. Also, the conditions for the existence of periodic and homoclinic cycles are stated.     

Cellular neural networks: Theory and circuit design

Cellular neural networks or CNNs are a novel neural network architecture introduced by Chua and Yang which is very general and flexible, has some important properties desirable for design applications and can be efficiently implemented on custom hardware based on analogue VLSI technology. In this paper an abstract normalized definition of cellular neural networks with arbitrary interconnection topology is given. Instead of stability, the property of convergence is found to be of central importance: large classes of convergent CNNs in practice always asymptotically approach some stable equilibrium where each component of the corresponding output is binary-valued. A highly efficient CMOS-compatible CNN circuit architecture is then presented where a basic cell consists of only two fully differential op amps, two capacitors and several MOSFETs, while a variable interconnection weight is realized with only four MOSFETs. Since all these elements are standard components in the current analogue IC technology and since all network functions are implemented directly on the device level, this architecture promises high cell and interconnection densities and extremely high operating speeds.     

Complete stability of feedback CNNs with dynamic memristors and second‐order cells

The paper considers a feedback cellular neural network (CNN) obtained by interconnecting elementary cells with an ideal capacitor and an ideal flux‐controlled memristor. It is supposed that during the analogue computation of the CNN the memristors behave as dynamic elements, so that each dynamic memristor (DM)‐CNN cell is described by a second‐order differential system in the state variables given by the capacitor voltage and the memristor flux. The proposed networks are called DM‐CNNs, that is CNNs using a dynamic (D) memristor (M). After giving a foundation to the DM‐CNN model, the paper establishes a fundamental result on complete stability, that is convergence of solutions toward equilibrium points, when the DM‐CNN has symmetric interconnections. Because of the presence of dynamic memristors, a DM‐CNN displays peculiar and basically different dynamic properties with respect to standard CNNs. First of all a DM‐CNN computes during the time evolution of the memristor fluxes, instead of the capacitor voltages as for a standard CNN. Furthermore, when a steady state is reached, the memristors keep in memory the result of the computation, that is the limiting values of the fluxes, while all memristor currents and voltages, as well as all currents, voltages, and power in the DM‐CNN vanish. Instead, for standard CNNs, currents, voltages, and power do not drop off when a steady state is reached. Copyright © 2016 John Wiley & Sons, Ltd.     

A study on WTA cellular neural networks

This paper addresses a number of basic issues concerning the dynamics of a class of winner‐take‐all cellular neural networks (WTA CNNs) proposed by Seiler and Nossek. The main result is an analytical estimate for the settling time, which shows from a theoretical point of view that such CNNs are well suited for on‐line applications requiring a large number of units, fast processing speed and a relatively high resolution. Other results are the determination of the largest parameter set that guarantee a correct WTA functionality for all initial conditions and the solution of a conjecture made by Seiler and Nossek. These results are proved by means of a new Lyapunov function to analyse the global dynamical behaviour of the WTA CNNs. Copyright © 2001 John Wiley & Sons, Ltd.     

A rigorous design method for binary cellular neural networks

In order to be able to take full advantage of the great application potential that lies in cellular neural networks (CNNs) we need to have successful design and learning techniques as well. In almost any analogic CNN algorithm that performs an image processing task, binary CNNs play an important role. We observed that all binary CNNs reported in the literature, except for a connected component detector, exhibit monotonic dynamics. In the paper we show that the local stability of a monotonic binary CNN represents sufficient condition for its functionality, i.e. convergence of all initial states to the prescribed global stable equilibria. Based on this finding, we propose a rigorous design method, which results in a set of design constraints in the form of linear inequalities. These are obtained from simple local rules similar to that in elementary cellular automata without having to worry about continuous dynamics of a CNN. In the end we utilize our method to design a new CNN template for detecting holes in a 2D object. © 1998 John Wiley Sons, Ltd.     

Preliminary results from an analog implementation of first‐order TDCNN dynamics

This work falls into the category of linear cellular neural network (CNN) implementations. We detail the first investigative attempt on the CMOS analog VLSI implementation of a recently proposed network formalism, which introduces time‐derivative ‘diffusion’ between CNN cells for nonseparable spatiotemporal filtering applications—the temporal‐derivative CNNs (TDCNNs). The reported circuit consists of an array of Gm‐C filters arranged in a regular pattern across space. We show that the state–space coupling between the Gm‐C‐based array elements realizes stable and linear first‐order (temporal) TDCNN dynamics. The implementation is based on linearized operational transconductance amplifiers and Class‐AB current mirrors. Measured results from the investigative prototype chip that confirms the stability and linearity of the realized TDCNN are provided. The prototype chip has been built in the AMS 0.35 µm CMOS technology and occupies a total area of 12.6 mm sq, while consuming 1.2 µW per processing cell. Copyright © 2010 John Wiley & Sons, Ltd.     

CIRCULANT MATRICES AND THE STABILITY OF A CLASS OF CNNs

In this paper we show that feedback matrices of ring CNNs are block circulants; as special cases, for example, feedback matrices of one-dimensional ring CNNs are circulant matrices. Circulants and their close relations the block circulants possess many pleasant properties which allow one to describe their spectrum completely. After deriving the spectrum of the feedback operator, we discuss conditions for a CNN to be contractive, ensuring global asymptotic stability.     

Absolute stability of a class of neural networks with unbounded delay

In this paper, the existence and uniqueness of the equilibrium point and absolute stability of a class of neural networks with partially Lipschitz continue activation functions are investigated. The neural networks contain both variable and unbounded delays. Using the matrix property, the necessary and sufficient condition for the existence and uniqueness of the equilibrium point of the neural networks is obtained. By constructing proper vector Liapunov functions and non‐linear integro‐differential inequalities involving both variable delays and unbounded delay, the sufficient conditions for absolute stability (global asymptotic stability) are obtained. Copyright © 2004 John Wiley & Sons, Ltd.     

Novel stability criteria for bidirectional associative memory neural networks with time delays

In this paper, the bidirectional associative memory (BAM) neural network with axonal signal transmission delay is considered. This model is also referred to as a delayed dynamic BAM model. By combining a number of different Lyapunov functionals with the Razumikhin technique, some sufficient conditions for the existence of a unique equilibrium and global asymptotic stability of the network are derived. These results are fairly general and can be easily verified. Besides, the approach for the analysis allows one to consider several different types of activation functions, including piecewise linear sigmoids with bounded activations as well as C¹‐smooth sigmoids. It is believed that these results are significant and convenient in the design and applications of BAM neural networks. Copyright © 2002 John Wiley & Sons, Ltd.     

Robust stochastic convergence and stability of neutral‐type neural networks with Markovian jump and mixed delays

The robust stochastic convergence and stability in mean square are investigated for a class of uncertain neutral‐type neural networks with both Markovian jump parameters and mixed delays. First, by employing the Lyapunov method and a generalized Halanay‐type inequality for stochastic differential equations, a delay‐dependent condition is derived to guarantee the state variables of the discussed neural networks to be globally uniformly exponentially stochastic convergent to a ball in the state space with a prespecified convergence rate. Next, by applying the Jensen integral inequality and a novel reciprocal convex lemma, a delay‐dependent criterion is developed to achieve the globally robust stochastic stability in mean square. With some parameters being fixed in advance, the proposed conditions are all expressed in terms of LMIs, which can be solved numerically by employing the standard MATLAB LMI toolbox package. Finally, two illustrated examples are given to show the effectiveness and less conservatism of the obtained results over some existing works. Copyright © 2014 John Wiley & Sons, Ltd.     

Long transient oscillations in a class of cooperative cellular neural networks

The real‐time processing capabilities of cellular neural networks (CNNs) are inherently related to the fast convergence time of the solutions toward the asymptotically stable equilibrium points. A typical requirement is that the settling time should not exceed a few (or at most 10) cell time constants. This paper introduces a class of completely stable nonsymmetric cooperative CNN rings whose solutions display unexpectedly long transient oscillations for a wide set of initial conditions and for a wide set of interconnection parameters. Numerical simulations show that the oscillations can easily last hundreds of cycles, and thousands of cell time constants, before settling to a steady state, thus possibly impairing their real‐time processing capabilities. Goal of the paper is also to show, by means of laboratory experiments on a discrete component prototype of the CNN ring, that the long oscillation phenomenon is physically robust with respect to the non‐idealities of the circuit implementation. The experiments show some other peculiar features of the long lasting oscillations as the metamorphosis between different periodic behaviors during the transient. Finally, analytic asymptotic estimates on the duration of the transient oscillations are provided. Copyright © 2013 John Wiley & Sons, Ltd.     

Effects of leakage delay on global asymptotic stability of complex‐valued neural networks with interval time‐varying delays via new complex‐valued Jensen's inequality

This paper investigates the global asymptotic stability analysis for a class of complex‐valued neural networks with leakage delay and interval time‐varying delays. Different from previous literature, some sufficient information on a complex‐valued neuron activation function and interval time‐varying delays has been considered into the record. A suitable Lyapunov‐Krasovskii functional with some delay‐dependent terms is constructed. By applying modern integral inequalities, several sufficient conditions are obtained to guarantee the global asymptotic stability of the addressed system model. All the proposed criteria are formulated in the structure of a complex‐valued linear matrix inequalities technique, which can be checked effortlessly by applying the YALMIP toolbox in MATLAB linear matrix inequality. Finally, two numerical examples with simulation results have been provided to demonstrate the efficiency of the proposed method.     

Nonlinear Lipschitz measure and adaptive control for stability and synchronization in delayed inertial Cohen‐Grossberg–type neural networks

In this paper, without transforming the original inertial neural networks into the first‐order differential equation by some variable substitutions, time‐varying delays are introduced into inertial Cohen‐Grossberg–type networks and the existence, the uniqueness, and the asymptotic stability and synchronisation for the neural networks are investigated. Firstly, the existence of a unique equilibrium point is proved by using nonlinear Lipschitz measure method. Second, by finding a new Lyapunov‐Krasovskii functional, some sufficient conditions are derived to ensure the asymptotic stability, the asymptotic synchronization, and the asymptotic adaptive synchronization. The results of this paper are new and they complete previously known results. We illustrate the effectiveness of the approach through a few examples.     

On the universe of stable cellular neural networks

A cellular neural network (CNN) is a novel analogue circuit architecture with many desirable features. This paper extends previous stability results of CNNs to include classes of strictly sign-symmetric and acyclic templates. We show that most of the 3×3 strictly sign-symmetric templates are stable almost everywhere, with the unknown templates reduced to three classes. We also introduce template graphs and CNN graphs and utilize them to obtain results concerning stability and irreducibility of CNN templates.     

Cellular neural network with trapezoidal activation function

This paper presents a cellular neural network (CNN) scheme employing a new non‐linear activation function, called trapezoidal activation function (TAF). The new CNN structure can classify linearly non‐separable data points and realize Boolean operations (including eXclusive OR) by using only a single‐layer CNN. In order to simplify the stability analysis, a feedback matrix W is defined as a function of the feedback template A and 2D equations are converted to 1D equations. The stability conditions of CNN with TAF are investigated and a sufficient condition for the existence of a unique equilibrium and global asymptotic stability is derived. By processing several examples of synthetic images, the analytically derived stability condition is also confirmed. Copyright © 2005 John Wiley & Sons, Ltd.     

On the separating capability of cellular neural networks

The cellular neural network is able to perform different image-processing tasks depending on the template values, i.e. the network parameters, used. In the case of linear templates the parameter space is divided into different regions by hyperplanes. Every region is associated with a task, such that all points within that region let the cellular neural network perform the desired task. In this paper a lower and an upper bound for the number of regions that can be separated with binary-input cellular neural networks are given, thus answering the question of how many different-tasks such a cellular neural network can perform.