Delay-dependent exponential stability of cellular neural networks with time-varying delays

The global exponential stability of cellular neural networks (CNNs) with time-varying delays is analyzed. Two new sufficient conditions ensuring global exponential stability for delayed CNNs are obtained. The conditions presented here are related to the size of delay. The stability results improve the earlier publications. Two examples are given to demonstrate the effectiveness of the obtained results.     

非对称细胞神经网络的稳定性分析

This paper describes the problem of stability for one-dimensional Cellular Neural Networks(CNNs). A sufficient condition is presented to ensure complete stability for a class of special CNN's with nonsymmetric templates, where the parameter in the output function is greater than or equal to zero. The main method is analysising the property of the equilibrium point of the CNNs system.     

Decentralized feedback control of uncertain cellular neural networks subject to time-varying delays and dead-zone input

This paper investigates the stabilization problem for uncertain cellular neural networks (CNNs) subject to time-varying delays and dead-zone input. On the basis of Lyapunov stability theory, a memoryless decentralized feedback control law is derived for guaranteeing global exponential stability of the system. The main results illustrate that the derived control law does not impose restriction on the derivative of the time-varying delays and can be applied to stabilizing the uncertain CNNs with time-varying delays and dead-zone input. An illustrative example is given to justify the validity and feasibility of the proposed control scheme.     

Existence and global stability of a periodic solution for a cellular neural network

Purpose of this study is to investigate the dynamical properties of Chua and Yang cellular neural networks (CNNs). Based on the continuation theorem of coincidence degree theory, a novel sufficient condition with respect to the existence of periodic solution for CNNs is derived. Moreover, a generalized Lyapunov–Krasovskii functional is designed to guarantee the global stability of the existed periodic solution. An illustrative example is given to verify the effectiveness and correctness of the proposed method, furthermore, random disturbance is added in the numerical simulation in order to verify the robustness of the proposed approach.     

Edge detection of noisy images based on cellular neural networks

This paper studies a technique employing both cellular neural networks (CNNs) and linear matrix inequality (LMI) for edge detection of noisy images. Our main work focuses on training templates of noise reduction and edge detection CNNs. Based on the Lyapunov stability theorem, we derive a criterion for global asymptotical stability of a unique equilibrium of the noise reduction CNN. Then we design an approach to train edge detection templates, and this approach can detect the edge precisely and efficiently, i.e., by only one iteration. Finally, we illustrate performance of the proposed methodology from the aspect of peak signal to noise ratio (PSNR) through computer simulations. Moreover, some comparisons are also given to prove that our method outperforms classical operators in gray image edge detection.     

Stability in cellular neural networks with a piecewise constant argument

In this paper, by using the concept of differential equations with piecewise constant arguments of generalized type [1], [2], [3] and [4], a model of cellular neural networks (CNNs) [5] and [6] is developed. The Lyapunov-Razumikhin technique is applied to find sufficient conditions for the uniform asymptotic stability of equilibria. Global exponential stability is investigated by means of Lyapunov functions. An example with numerical simulations is worked out to illustrate the results.     

A topological approach to the existence of solutions for nonlinear differential equations with piecewise constant argument

In this paper, we investigate qualitative behavior of nonlinear differential equations with piecewise constant argument (PCA). A topological approach of Wa?ewski-type which gives sufficient conditions to guarantee that the graph of at least one solution stays in a given domain is formulated. Moreover, our results are also suitable for a class of general system of discrete equations. By using a regular polyfacial set, we apply our developed topological approach to cellular neural networks (CNNs) with PCA. Some new results are attained to reveal dynamic behavior of CNNs with PCA and discrete-time CNNs. Finally, an illustrative example of CNNs with PCA shows usefulness and effectiveness of our results.     

Existence and global exponential stability of almost periodic solution for CNNs with variable coefficients and delays

By using exponential dichotomy and the Banach fixed-point theory and combining Yang inequality, some sufficient conditions are derived ensuring existence and global exponential stability of almost periodic solution for CNNs with variable coefficients and delays. Without assuming the boundedness of signal function, these results obtained here can be expected to have highly important significance in designs and applications of the networks. We extend and improve previously known results.     

Improved global exponential stability criteria of cellular neural networks with time-varying delays

In this paper, the existence and uniqueness of the equilibrium point and stability of the cellular neural networks (CNNs) with time-varying delays are analyzed and proved. Several global exponential stability conditions of the neural networks are obtained by the delay differential inequality and matrix measures approach. The obtained results are extensions of the earlier literature. The approach used in this paper is also suitable for delayed Hopfield neural networks and delayed bi-directional associative memory neural networks whose activation functions are often nondifferentiable or unbounded. Two simulation examples in comparison to previous results in literature are shown to check the theory in this paper.     

The Effect of Clone Template Parameters on the Spreading Speeds in CNNs

The aim of this paper is to investigate the effect of clone template parameters on the spreading speeds in cellular neural networks(CNNs). According to the property analysis of spreading speeds of monotone semiflows developed by Yu and Zhang [{\it European Journal of Applied Mathematics}, {\bf 31} (2020), 369-384], we investigate the sign of spreading speeds, continuity and limit cases with no propagation phenomena for CNNs with general output functions where each cell interacts with its 2-neighborhood cell.     

Sinusoidal modulation control method in a chaotic neural network

Chaotic neural networks (CNNs) have chaotic dynamic associative memory properties: The memory states appear non-periodically, and cannot be converged to a stored pattern. Thus, it is necessary to control chaos in a CNN in order to recognize associative memory. In this paper, a novel control method, the sinusoidal modulation control method, has been proposed to control chaos in a CNN. In this method, a sinusoidal wave simplified from brain waves is used as a control signal to modulate a parameter of the CNN. The simulation results demonstrate the effectiveness of this control method. The controlled CNN can be applied to information processing. Moreover, the method provides a way to associate brain waves by controlling CNNs.     

Universality of deep convolutional neural networks

Deep learning has been widely applied and brought breakthroughs in speech recognition, computer vision, and many other domains. Deep neural network architectures and computational issues have been well studied in machine learning. But there lacks a theoretical foundation for understanding the approximation or generalization ability of deep learning methods generated by the network architectures such as deep convolutional neural networks. Here we show that a deep convolutional neural network (CNN) is universal, meaning that it can be used to approximate any continuous function to an arbitrary accuracy when the depth of the neural network is large enough. This answers an open question in learning theory. Our quantitative estimate, given tightly in terms of the number of free parameters to be computed, verifies the efficiency of deep CNNs in dealing with large dimensional data. Our study also demonstrates the role of convolutions in deep CNNs.     

Stochastic exponential synchronization for delayed neural networks with semi-Markovian switchings: Saturated heterogeneous sampling communication

By implementing heterogeneous sampling communication mechanism, this article addresses the exponential synchronization issue of drive–response chaotic neural networks (CNNs) with interval time-varying delays by simultaneously taking into account the semi-Markovian switchings and saturating actuators. More specifically, a semi-Markovian jumping model whose transition rates (TRs) are not constant but depends on the sojourn time (ST) is introduced to characterize the stochastic changing among the interaction of CNNs, which makes the NNs model under consideration more suitable for some actual circumstances. More particularly, we assume that the sampling intervals are heterogeneous and time-varying, which may be more practical in real-life applications than homogeneous sampling policy. Additionally, by introducing some new terms, one novel time-dependent Lyapunov–Krasovskii function (LKF) is ingeniously constructed, which can fully capture the characteristic information of heterogeneous sampling pattern. Benefitting from the introduced relaxed free-weighting matrices (FWM) and resorting to the formed LKF, some sampling-interval-dependent sufficient conditions for controller design of the resulting semi-MJNNs error system are established and expressed by linear matrix inequalities (LMIs). These LMIs-based constraints can be effectively checked by utilizing the available software packages. Therein, the developed synchronization criteria dependent on both the lower and upper bounds of sampling periods, and the available information about the actual sampling pattern is fully considered. Ultimately, two numerical examples are provided to demonstrate the feasibility and practicability of our theoretical findings.     

Real-time tomographic reconstructor based on convolutional neural networks for solar observation

Solar observation is the branch of astronomy devoted to the study of the Sun. When the light wavefront that comes from the Sun penetrates the atmosphere, it suffers some distortions caused by optically turbulent layers that change the wavefront's shape and morphology. Therefore, in order to obtain a good-quality image, it is necessary to correct the induced error. This is done by applying adaptive optics (AO) techniques. In the case of the present research, it is performed with the help of a Single Conjugate Adaptive Optics System (SCAO). The reconstruction technique proposed in this research is a SCAO based on convolutional neural networks (CNNs). This research develops and assesses a real-time tomographic reconstructor based on CNN, able to correct the error introduced by the atmosphere in the light wavefront received from the Sun. The CNN was trained and validated using data from the Durham AO Simulation Platform as input information. This platform incorporates certain solar functionalities that have been employed in the present research, allowing us to simulate a solar telescope. The normalized errors obtained for both ReLu and Leaky ReLu kernels were promising, without showing statistically significant differences among kernels in the value of RMSE volts of the deformable mirror commands. When different kernel dimensions are compared, statistically significant differences are found, showing that RMSE volts of the deformable mirror commands are lower for 3 × 3 kernels when compared with those of dimensions 5 × 5 and 7 × 7. As far as the authors know, this is the first time that an AO system based on CNN has been developed for solar telescopes.     

On the stability properties of linear dynamic time-varying unforced systems involving switches between parameterizations from topologic considerations via graph theory

This paper deals with the stability of linear time-varying systems involving switches through time between different parameterizations of a dynamic linear time-varying system. Graph theory is used to describe the combinations of possible switches of the various sets of parameterizations which ensure the stability of the configurations. Each graph vertex is associated with a particular parameterization while edges (arcs) are associated with switches between the graphs (directed graphs or digraphs). An axiomatic framework is first established concerned with previously known stability results from systems theory related to the achievement of stability when switches between several parameterizations of a dynamic system take place. The axiomatic context is then used to obtain stability results mainly based on the topology of the links between the various configurations associated with a state-trajectory as well as on the nature of the vertices related to the stability of the various isolated parameterizations.     

Stability of the Radau IA and Lobatto IIIC methods for neutral delay differential system

Numerical stability is considered for several Runge–Kutta methods to systems of neutral delay differential equations. The linear stability analysis is adopted to the system. Adapted with the equistage interpolation process as well as the continuous extension, the Runge–Kutta methods are shown to have the numerical stability similar to the analytical asymptotic stability with arbitrary stepsize, when certain assumptions hold for the logarithmic matrix norm on the coefficient matrices of the NDDE system.     

A Stability Analysis of Growth Matrix in Co-operating Enterprises

On the basis of the growth-rate matrix in co-operating enterprises, this study deals with growth stability as well as the possibility of interpolation and extrapolation of a dynamic growth system. This study also suggests an index of growth stability when a new enterprise is introduced: disequilibrium occurs initially, and stability is gradually reached with the passage of time.     

Stabilization of switched linear systems with time-delay in detection of switching signal

A feedback stabilization problem for switched linear systems with time-delay in detection of switching signal is formulated. First, online state feedback controller design method for asymptotic stability and exponential stability is given. Then, offline state feedback controller design method for asymptotic stability and exponential stability is given as well.     

Denoising Convolutional Neural Network with Energy-Based Attention for Image Enhancement

In the realm of image denoising, the use of convolutional neural networks (CNNs) has lately gained traction. Several activities involve the utilization of excellent-clarity pictures and recordings. Images were captured in a wide variety of illumination circumstances, which means that not all of them are of the highest quality. Low-light photography suffers from a decline in perceived image quality because of the restricted dynamic range of the pixel values. Therefore, it is vital to enhance the appearance of images. Maximum texture retention is achieved by the structural similarity index-loss-based method. The suggested discrete wavelet transform (DWT)-self attention (SA)-Denoising convolutional neural networks (DnCNNs) make use of state-of-the-art techniques for image denoising like energy band analysis, very deep architecture, learning algorithms, dense-sparse-dense training, and regularization approaches. DnCNN is intended to remove the hidden layers" latent, yielding a pure picture. After a degraded input sample has had its relevant energy features retrieved using DWT, the perfect image enhancement is achieved thanks to the incorporation of the self-attention mechanism. Second, a hierarchical-branch network is formed by combining the suggested network with the denoising CNN and additional loss in order to reduce the reliance on the amount of noisy data in multi-modal picture analysis and make the problem of image enhancement more tractable. In the end, DWT-SA-DnCNN"s self-learning qualities are used to improve image quality by obtaining features including undesirable noisy data, edge factor, texture, uniform and non-uniform areas, smoothness, and object structure. Simulation results show that our hybrid DWT-SA-DnCNN-based contrast enhancement strategy outperforms state-of-the-art methods.     

Stability on a cone in terms of two measures for impulsive differential equations with “supremum”

The stability of nonlinear impulsive differential equations with “supremum” is studied. A special type of stability, combining two different measures and a dot product, is defined. The definition is a generalization of several types of stability known in the literature. Razumikhin’s method as well as a comparison method for scalar impulsive ordinary differential equations have been employed.