Search

We propose a class of Clifford-valued neutral-type fuzzy cellular neural network models with leakage delays on time scales, give the definition and basic properties of Clifford-valued almost automorphic functions on time scales, and prove a composition theorem of Clifford-valued almost automorphic functions on time scales, and on this basis, we study the existence and global exponential stability of almost automorphic solutions of this class of neural networks. The research in this paper can unify the research of real-valued neural networks, complex-valued neural networks and quaternion-valued neural networks, as well as the research of continuous-time neural networks and discrete-time neural networks. Finally, an example illustrates the feasibility of the results obtained.

In this paper, we report a novel memristor-based cellular neural network (CNN) without equilibrium points. Dynamical behaviors of the memristor-based CNN are investigated by simulation analysis. The results indicate that the system owns complicated nonlinear phenomena, such as hidden attractors, coexisting attractors, and initial boosting behaviors of position and amplitude. Furthermore, both heterogeneous multistability and homogenous multistability are found in the CNN. Finally, Multisim circuit simulations are performed to prove the chaotic characteristics and multistability of the system.

—In this paper, a popular dynamic neural network structure called Cellular Neural Network (CNN) is employed as a channel equalizer in digital communications. It is shown that, this nonlinear system is capable of suppressing the effect of intersymbol interference (ISI) and the noise at the channel. The architecture is a small-scaled, simple neural network containing only 25 neurons (cells) with a neighborhood of r = 2 , thus including only 51 weight coefficients. Furthermore, a special technique called repetitive codes in equalization process is also applied to the mentioned CNN based system to show that the two-dimensional structure of CNN is capable of processing such signals, where performance improvement is observed. Simulations are carried out to compare the proposed structures with minimum mean square error (MMSE) and multilayer perceptron (MLP) based equalizers.

The purpose of this paper is to study the stability of traveling wave solutions for nonlinear cellular neural networks (CNN) with distributed delays. First, we establish two kinds of comparison theorems for the Cauchy problem of the CNN model. With the help of the comparison theorems, we can show that all solutions of the Cauchy problem for the CNN model converge exponentially to the traveling wave solutions when the initial perturbations around the traveling wave solutions belonging to suitable weighted spaces.

Corner detection represents one of the most important steps to identify features in images. Due to their powerful local processing capabilities, Cellular Nonlinear/Neural Networks (CNN) are commonly utilized in image processing applications such as image edge detection, image encoding and image hole filling. CNN perform well for locating corner features in binary images. However, their use in grayscale images has not been considered due to their design difficulties. In this chapter, a corner detector based on CNN for grayscale images is presented. In the approach, the original processing scheme of the CNN is modified to include a nonlinear operation for increasing the contrast of the local information in the image. With this adaptation, the final CNN parameters that allow the appropriate detection of corner points are estimated through the Differential evolution algorithm by using standard training images. Different test images have been used to evaluate the performance of the presented corner detector. Its results are also compared with popular corner methods from the literature. Computational simulations demonstrate that the presented CNN approach presents competitive results in comparison with other algorithms in terms of accuracy and robustness.

Recent papers in the literature introduced a class of neural networks (NNs) with memristors, named dynamic-memristor (DM) NNs, such that the analog processing takes place in the charge–flux domain, instead of the typical current–voltage domain as it happens for Hopfield NNs and standard cellular NNs. One key advantage is that, when a steady state is reached, all currents, voltages, and power of a DM-NN drop off, whereas the memristors act as nonvolatile memories that store the processing result. Previous work in the literature addressed multistability of DM-NNs, i.e., convergence of solutions in the presence of multiple asymptotically stable equilibrium points (EPs). The goal of this paper is to study a basically different dynamical property of DM-NNs, namely, to thoroughly investigate the fundamental issue of global asymptotic stability (GAS) of the unique EP of a DM-NN in the general case of nonsymmetric neuron interconnections. A basic result on GAS of DM-NNs is established using Lyapunov method and the concept of Lyapunov diagonally stable matrices. On this basis, some relevant classes of nonsymmetric DM-NNs enjoying the property of GAS are highlighted.

The nonlinear dynamic system modeling based on the input/output relationship results from solving the approximation problem. One can distinguish two large classes: polynomials and neural networks. The different types of neural networks draw attention. The discrete time feedforward cellular neural network is suggested for filtering non-Gaussian noise, as well as the example of nonlinear filters modeling to cancel the impulse noise is represented.

Neural networks in which communication works only among the neighboring units are called cellular neural networks (CNNs). These are used in analyzing 3D surfaces, image processing, modeling biological vision, and reducing nonvisual problems of geometric maps and sensory-motor organs. Topological indices (TIs) are mathematical models of the (molecular) networks or structures which are presented in the form of numerical values, constitutional formulas, or numerical functions. These models predict the various chemical or structural properties of the under-study networks. We now consider analogous graph invariants, based on the second connection number of vertices, called Zagreb connection indices. The main objective of this paper is to compute these connection indices for the cellular neural networks (CNNs). In order to find their efficiency, a comparison among the obtained indices of CNN is also performed in the form of numerical tables and 3D plots.

The effect of non-homogeneities in cell parameters for a Cellular Neural Network (CNN) architecture consisting of linear cells locally connected within a specified neighborhood and able to exhibit pattern formation is investigated. The results are an extension for another architecture of Turing pattern formation in nonhomogeneous two-grid CNN's and are based on a continuity theorem about the dependence of polynomial roots on their coefficients.

The scope of this chapter is limited with the connection structures of the biologically inspired neural networks. Many biological networks are constructed with both regular and random connections between neurons. Bioinspired systems should mimic this mixed topology of biological networks, while the artificial system is still realizable. In this chapter, we focus on cellular neural networks, which is a bioinspired network with many analog realizations. Cellular neural network models are investigated under the existence of random connections additional to its regular connections.

We characterize the attractor-based expressive power of several models of recurrent neural networks.The deterministic rational-weighted networks are Muller Turing equivalent.The deterministic real-weighted and evolving networks recognize the class of B C ( ź 2 0 ) neural ω languages.The nondeterministic rational and real networks recognize the class of Σ 1 1 neural ω-languages. We provide a characterization of the expressive powers of several models of deterministic and nondeterministic first-order recurrent neural networks according to their attractor dynamics. The expressive power of neural nets is expressed as the topological complexity of their underlying neural ω-languages, and refers to the ability of the networks to perform more or less complicated classification tasks via the manifestation of specific attractor dynamics. In this context, we prove that most neural models under consideration are strictly more powerful than Muller Turing machines. These results provide new insights into the computational capabilities of recurrent neural networks.

A cellular neural network (CNN) with a bipolar stepwise activation function is considered. A comparative analysis of CNN learning algorithms on a given set of binary reference images for various degrees of noise (inversion of randomly selected pixels) and various cell neighborhood sizes is performed. For CNN training a local projection method, which provides much higher noisy images quality filtering than the classical local perceptron learning algorithm, is proposed.

In this paper we study travelling wave solutions of the viscoelastic Burgers’ equation. RTD-based Cellular Neural Networks (CNN) model of this equation is presented. Using such realization new wave profiles of the travelling wave solutions are obtained.

Synchronization and control are two very important aspects of any dynamical systems. Among various kinds of nonlinear systems, competitive neural network holds a very important place due to its application in diverse fields. The model is general enough to include, as subclass, the most famous neural network models such as competitive neural networks, cellular neural networks and Hopfield neural networks. In this paper, the problem of feedback controller design to guarantee synchronization for competitive neural networks with time-varying delays is investigated. The goal of this work is to derive an existent criterion of the controller for the exponential synchronization between drive and response neutral-type competitive neural networks with time-varying delays. The method used in this brief is based on feedback control gain matrix by using the Lyapunov stability theory. The synchronization conditions are given in terms of LMIs. To the best of our knowledge, the results presented here are novel and generalize some previous results. Some numerical simulations are also represented graphically to validate the effectiveness and advantages of our theoretical results.

Recent years have seen a surge of interest in deep neural networks fueled by their successful applications in numerous image processing and computer vision tasks. However, such applications typically come with huge computational loads. In this article, we explore the possibility of using small neural networks to denoise images. In particular, we present SSaNN (Self-Similarity and Neural Networks), a denoising algorithm which combines the strength of BM3D on large-scale structured patterns with that of neural networks on small-scale texture content. This algorithm is able to produce a better overall recovery than both BM3D and small neural networks.

The article is devoted to the analysis of neural networks consisting of generalized neural elements. The first part of the article proposes a new neural network model — a modified network of generalized neural elements (MGNE-network). This network developes the model of generalized neural element, whose formal description contains some flaws. In the model of the MGNE-network these drawbacks are overcome. A neural network is introduced all at once, without preliminary description of the model of a single neural element and method of such elements interaction. The description of neural network mathematical model is simplified and makes it relatively easy to construct on its basis a simulation model to conduct numerical experiments. The model of the MGNE-network is universal, uniting properties of networks consisting of neurons-oscillators and neurons-detectors. In the second part of the article we prove the equivalence of the dynamics of the two considered neural networks: the network, consisting of classical generalized neural elements, and MGNE-network. We introduce the definition of equivalence in the functioning of the generalized neural element and the MGNE-network consisting of a single element. Then we introduce the definition of the equivalence of the dynamics of the two neural networks in general. It is determined the correlation of different parameters of the two considered neural network models. We discuss the issue of matching the initial conditions of the two considered neural network models. We prove the theorem about the equivalence of the dynamics of the two considered neural networks. This theorem allows us to apply all previously obtained results for the networks, consisting of classical generalized neural elements, to the MGNE-network.

In this paper, we are concerned with the existence, uniqueness and global asymptotic stability of equilibrium point for a class of complex-valued Cohen-Grossberg neural networks with time delays. By using homeomorphism theory, new matrix inequality techniques and inequality techniques, several LMI-based sufficient conditions on the existence, uniqueness and global asymptotic stability of equilibrium point for above complex-valued Cohen-Grossberg neural networks with two classes of complex-valued activation functions, behaved functions and amplification functions are obtained. So far, only the stabilities of complex-valued recurrent neural networks, Hopfield neural networks and Cellular neural networks have been studied. Hence, it is for the first place that the stability of complex-valued Cohen-Grossberg neural networks is discussed in this paper.

Being two famous neural networks, the error back-propagation (BP) algorithm based neural networks (i.e., BP-type neural networks, BPNNs) and Hopfield-type neural networks (HNNs) have been proposed, developed, and investigated extensively for scientific research and engineering applications. They are different from each other in a great deal, in terms of network architecture, physical meaning and training pattern. In this paper of literature-review type, we present in a relatively complete and creative manner the common natures of learning between BP-type and Hopfield-type neural networks for solving various (mathematical) problems. Specifically, comparing the BPNN with the HNN for the same problem-solving task, e.g., matrix inversion as well as function approximation, we show that the BPNN weight-updating formula and the HNN state-transition equation turn out to be essentially the same. Such interesting phenomena promise that, given a neural-network model for a specific problem solving, its potential dual neural-network model can thus be developed.

A major problem encountered by researchers of dynamic neural networks is the computational complexity increasing the learning time. In this paper the parallel realization of the RTRN and the Elman networks are discussed. Both networks are examples of dynamic neural networks. Inherent parallelism of dynamic neural networks has been employed to accelerate the learning process. The proposed solution is based on a highly parallel three dimensional architecture to speed up the learning performance. The presented structures are suitable for efficient parallel realization in digital hardware or vector processors.

This paper studies the opportunities of energy-accuracy tradeoff in cellular neural network (CNN). Algorithmic characteristics of CNN is coupled with hardware-induced error distribution of a digital CNN cell to evaluate energy-accuracy tradeoff for simple image processing tasks as well as a complex application. The analysis shows that errors modulate the cell dynamics and propagate through the network degrading the output quality and increasing the convergence time. The error propagation is determined by the task being performed by the CNN, specifically, the strength of the feedback template. Controlling precision is observed to be a more effective approach for energy-accuracy tradeoff in CNN than voltage over scaling.

The cellular neural network (CNN) is one of the classic artificial neural networks. During the past decades, the one-dimensional CNN models and their applications have not yet been paid enough enthusiasm too. For this reason, this paper proposes a simplified one-dimensional CNN model and then designs a pairwise network using this model to demonstrate its applicability. This pairwise CNN consists of two parallel one-dimensional CNNs, a fixed master and a movable slave. Using this pairwise CNN, an algorithm is developed to perform the global alignment of two DNA sequences. In this algorithm, the slave moves forward step by step, and the cell states of the master are computed in the meanwhile. Based on all the states obtained in all time steps, a state selection array is generated then a global alignment path is determined from this array. Under the guidance of the alignment path, two DNA sequences are globally aligned by inserting blank spaces in the appropriate positions of these two sequences. Experiments on aligning the DNA sequences from the publicly available databases of the NCBI with this method are carried out in this paper and compared with the other two methods. Through evaluating computation time and similarity, these experiments prove that the proposed one-dimensional CNN model is effective, and the alignment algorithm based on a pairwise CNN of the model is efficient, obtaining higher similarity with less computation time than the other two.

WOS: 000440049800011, In this study, a Lorenz-like system based on cellular neural networks (CNNs) is proposed. Complex butterfly chaotic attractors of Lorenz-like systems are constructed using a three-cell CNN. The proposed system makes the CNN imitate quadratic system dynamics without altering its output nonlinearity. Experimental and numerical analysis results are presented in order to verify the convenience of the system., Erciyes University [FDK-2016-6757], This work is supported by the Research Fund of Erciyes University (Project Code: FDK-2016-6757).

In this paper, the technique of biomedical image is presented by employing cellular neural networks (CNN) and linear matrix inequality (LMI). The main objective is to obtain the templates. Based on Cellular neural networks (CNN) is a high-speed parallel processor, this paper proposes a optimized templates for biomedical image edge detection.

Enhancing energy/resource efficiency of neural networks is critical to support on-chip neural image processing at Internet-of-Things edge devices. This paper presents recent technology advancements towards energy-efficient neural image processing. 3D integration of image sensor and neural network improves power-efficiency with programmability and scalability. Computation energy of feedforward and recurrent neural networks is reduced by dynamic control of approximation, and storage demand is reduced by image-based adaptive weight compression. Emerging devices such as tunnel FET and Resistive Random Access Memory are utilized to achieve higher computation efficiency than CMOS-based designs.

BACKGROUND: Realistic and real-time modeling and simulation of soft tissue deformation is a fundamental research issue in the field of surgical simulation. OBJECTIVE: In this paper, a novel cellular neural network approach is presented for model-ing and simulation of soft tissue deformation by combining neural dynamics of cellular neural network with ChainMail mechanism. METHOD: The proposed method formulates the problem of elastic deformation into cellular neural network activities to avoid the complex computation of elasticity. The local position adjustments of ChainMail are incorporated into the cellular neural network as the local con-nectivity of cells, through which the dynamic behaviors of soft tissue deformation are trans-formed into the neural dynamics of cellular neural network. RESULTS: Experiments demonstrate that the proposed neural network approach is capable of modeling the soft tissues' nonlinear deformation and typical mechanical behaviors. CONCLUSIONS: The proposed method not only improves ChainMail's linear deformation with the nonlinear characteristics of neural dynamics but also enables the cellular neural net-work to follow the principle of continuum mechanics to simulate soft tissue deformation.

The synthesis of nonlinear impulse noise filters in the form of feedforward cellular neural network and two-layer perseptron network are considered. Neural filters are used for cancelling the “salt and pepper”-type interference from half-tone images. It is shown that the filter in the form of cellular neural network surpasses two-layer perseptron filter in the filtration accuracy. Neural filters ensure a higher level of impulse noise suppression as compared with median filters.

Neural networks have been treated as “black boxes” for the majority of the machine learning community. The difficulty in making sense of neural networks lies in the complex topology of the hidden layers. Although there have been works in the literature aimed at demystifying the way neural networks operate, making sense of the hidden layer still remains a challenge. In this work, we propose a way to derive physical meaning from the hidden layer by mapping our neural network to the topology of a Bayesian network. Using this mapping, we enhance the probabilities of the Bayesian network resulting in a hybrid model that outperforms both the Bayesian and neural networks in the task of traffic accident prediction. Our analysis suggests that a neural network can estimate the node probabilities of a Bayesian network if mapped accordingly.

VSF-Network is a neural network model that learns dynamical patterns. It is hybrid neural network combining a chaotic neural network and a hierarchical neural network. The hierarchical neural network part is used for pattern learning. The chaotic neural network part monitors behavior of neurons in the middle layer of the hierarchical neural network. In this paper, two theoretical backgrounds of VSF-Network are introduced. An incremental learning moel using chaotic neural networks is introduced. The monitoring by chaotic neural network is based on the clusters of synchronized oscillators. Using the monitoring results, redundant neurons in the hierarchical neural network are found and they are used for learning of new patters. The second background is about the pattern recognition by combining learned patterns. The mechanism about recognition of combined learned patterns is explained by subspace selection in linear space. Through an experiment, its ability for the incremental learning and the pattern recognition are shown, and the factors influencing learning of VSF-Network are also shown.

Recurrent neural networks are represented as non-linear models of dynamic systems. This kind of neural networks is divided into two groups, which are globally and locally recurrent neural networks. Some types are distinguished among globally recurrent networks. The major approximation properties and features of every distinguished type are emphasized. The represented analysis is useful for choosing the neural network structure a priori (prior to its training or constructing the mathematical model of the nonlinear system).

Artificial neural networks (ANNs) have been applied in many areas successfully because of their ability to learn, ease of implementation and fast real-time operation. In this research, there are proposed two algorithms. The first is cellular neural network (CNN) with noise level estimation. While the second is modify cellular neural network with noise level estimation. The proposed CNN modification is by adding the Rossler chaos to the CNN fed. Noise level algorithm were used to image noise removal approach in order to get a good image denoising processing with high quality image visual and statistical measures. The results of the proposed system show that the combination of chaos CNN with noise level estimation gives acceptable PSNR and RMSE with a best quality visual vision and small computational time.

In the current paper, a neural network method to solve sixth-order differential equations and their boundary conditions has been presented. The idea this method incorporates is to integrate knowledge about the differential equation and its boundary conditions into neural networks and the training sets. Neural networks are being used incessantly to solve all kinds of problems hailing a wide range of disciplines. Several examples are given to illustrate the efficiency and implementation of the Neural Network method.

The stochastic neural networks can be considered as an expansion of cellular neural networks and Hopfield neural networks. In real world, the neural networks are prone to be instable due to time delay and external disturbance. In this paper, we consider the asymptotic stability for the stochastic neural networks with multiple time-varying delays. By employing a Lyapunov-Krasovskii function, a sufficient condition which guarantees the asymptotic stability of the state trajectory in the mean square is obtained. The criteria proposed can be verified readily by utilising the linear matrix inequality toolbox in Matlab, and no parameters to be tuned. In the end, two numerical examples are provided to demonstrate the effectiveness of the proposed method.

This paper addresses the function projective synchronization problem between 3-cell Quantum cellular neural network hyperchaotic system and 6th-order cellular neural network system in the presence of unknown system parameters. Quantum-CNN systems produce nanoscale chaotic controllers under certain conditions. Based on Lyapunov stability theory the adaptive control law and the parameter update law are proposed to make the states of the 3-cell Quantum-CNN hyperchaotic system and 6th-order cellular neural network system asymptotically synchronized. Numerical simulations are presented to demonstrate the effectiveness of the proposed synchronization schemes.

In this paper we provide an up-to-date survey on the study of the complexity of the mosaic solutions on neural network equations. Three types of equations, namely, cellular neural networks (CNNs), multi-layer CNN (MCNNs) and inhomogeneous CNNs (ICNNs) are discuss herein. Such topic strong related to the learning algorithm and training process on neural network equations. Each neural network produces different mosaic solution space, and each mosaic solution space induces an different symbolic dynamics. To understand the complexity (spatial entropy) of the mosaic solution space for a given neural network equation, we need to identify which the underlying symbolic space is, then using the established knowledge of symbolic dynamical systems to compute its spatial entropy. Recently there has been substantial progress in this field. This paper is a comprehensive survey of this field. It provides a summary of the interesting results in this field. It is our hope that the paper will provide a good overview of major results and techniques, and a friendly entry point for anyone who is interested in studying problems in this field.

In this paper, the problem of exponential synchronization for a class of chaotic neural networks which covers the Hopfield neural networks and cellular neural networks with reaction-diffusion terms and time-varying delays is investigated. A feedback control gain matrix is derived to achieve the state synchronization of two identical neural networks with reaction-diffusion terms, and the synchronization condition can be verified if a certain Hamiltonian matrix with no eigenvalue on the imaginary axis.

In this paper, infrared image edge detection algorithm based on the combination of the cellular neural networks (CNN) and distributed genetic algorithm (DGA) is proposed. The CNN template is used to train the network with distributed genetic algorithm (CNN_DGA). The experimental results show: the edge of the infrared image was extracted accurately with CNN_DGA edge detection algorithm; furthermore, the noise of the infrared image was reduced greatly. Compared with the edge detection algorithms based on cellular neural networks with template trained by particle swarm optimization, parameters’ search range and convergence speed are greatly improved.

In this study, a revised Group Method of Data Handling (GMDH)-type neural network self-selecting functions is applied to the computer aided image diagnosis (CAD) of lung cancer. The GMDH-type neural network algorithm has an ability of self-selecting optimum neural network architecture from three neural network architectures, such as sigmoid function neural network, radial basis function neural network and polynomial neural network. The GMDH-type neural network also has abilities of self-selecting the number of layers, the number of neurons in hidden layers and useful input variables. This algorithm is applied to CAD of lung cancers, and it is shown that this algorithm is useful for the CAD, and is very easy to apply to practical complex problems because optimum neural network architecture is automatically organized.

Neuromorphic hardware using simplified elements and thin-film semiconductor devices as synapse elements is proposed. It is assumed that amorphous metal-oxide semiconductor devices are used for the synapse elements, and the characteristic degradation is utilized for the learning rule named modified Hebbian learning. First, we explain an architecture and operation of a Hopfield neural network. Next, we model the electrical characteristic of the thin-film semiconductor devices and simulate the letter recognition by the neural network. Particularly in this presentation, we show a degradation map. On the other hand, we also explain an architecture and operation of a cellular neural network, model the thin-film semiconductor devices, and simulate the letter recognition. Particularly in this presentation, we evaluate connection schemes. It is found that the cellular neural network has higher performance when it has diagonal connections. Moreover, we compare the Hopfield and cellular neural networks. It is found that the Hopfield neural network has higher performance, although the cellular neural network has a simple structure.

As an important tool to study practical problems of biology, engineering and image processing, the cellular neural networks (CNNs) has caused more and more attention. Some interesting results about the existence of solution for cellular neural networks have been obtained. In this paper, by means of iterative analysis, the existence of periodic solution and the uniform stability of the equilibrium point of impulsive Hopfield cellular neural networks with distributed delays are considered. Some new results are obtained.

A fractional-order two-neuron Hopfield neural network with delay is proposed based on the classic well-known Hopfield neural networks, and further, the complex dynamical behaviors of such a network are investigated. A great variety of interesting dynamical phenomena, including single-periodic, multiple-periodic, and chaotic motions, are found to exist. The existence of chaotic attractors is verified by the bifurcation diagram and phase portraits as well.

This paper presents a method of using Cellular Neural Network (CNN) for solving hydraulic Navier-Stokes equations, which are a set of three partial differential equations with 3 functional variables and each function has three variables for time and space. The paper has 5 parts: part 1 is the introduction; Part 2 discusses the Navier-Stokes for flow through narrow slots; Part 3 analyzes and finds templates for designing CNN architecture; Part 4 provides some solutions for optimizing resources and speeding up computing; Part 5 sets up CNN chip for implementing and computing results; the last part is conclusion and developing trends.

The algorithm of logical or operation for binary image is less. A kind of Cellular Neural Networks (CNN) template that can realize logic or of binary image was studied and a new algorithm was proposed. By setting binary image logic or algorithms, and analyzes LOGOR CNN template robustness, proposed a theorem, finally gives a strict mathematical proof. As long as the actual use of the template parameters conform to theorem is given in the parameter range, CNN can realize logic or operation of binary image. Experimental results verify the effectiveness of LOGOR CNN and the feasibility of the design robustness theorem in practical application.

Edge is one of the basic characteristics of image. Edge detection is a very important step in image analysis, and the cellular neural network is a method that is very effective in edge detection. This article is based on cellular neural networks (cellular neural network, CNN), researching the algorithm of CNN's neighborhood radius equal 2 about the process of image edge detection, expounds the key steps in the process of algorithm realization, and prove the stability of the algorithm. The result based on the neighborhood radius equal 2 of CNN algorithm compare with the CNN's neighborhood radius equal to 1 algorithm and classical algorithm (prewit, cannyt, sobel, etc.), and we can analyze and compares the advantage and disadvantage of several kinds of algorithm on the performance, the accuracy of the quantitative comparison of the test results. The experimental results show that the CNN template algorithm of edge detection based on the neighborhood radius equal 2 results are more significantly, and able to high-speed parallel computing in hardware implementation, can achieve real-time image processing.

An unsolved problem in neuroevolution (NE) is to evolve artificial neural networks (ANN) that can store and use information to change their behavior online. While plastic neural networks have shown promise in this context, they have difficulties retaining information over longer periods of time and integrating new information without losing previously acquired skills. Here we build on recent work by Graves et al. [5] who extended the capabilities of an ANN by combining it with an external memory bank trained through gradient descent. In this paper, we introduce an evolvable version of their Neural Turing Machine (NTM) and show that such an approach greatly simplifies the neural model, generalizes better, and does not require accessing the entire memory content at each time-step. The Evolvable Neural Turing Machine (ENTM) is able to solve a simple copy tasks and for the first time, the continuous version of the double T-Maze, a complex reinforcement-like learning problem. In the T-Maze learning task the agent uses the memory bank to display adaptive behavior that normally requires a plastic ANN, thereby suggesting a complementary and effective mechanism for adaptive behavior in NE.

This paper proposes a novel method using deep spatial-temporal neural networks based on deep Convolutional Neural Network (CNN) for multimedia event detection. To sufficiently take advantage of the motion and appearance information of events from videos, our networks contain two branches: a temporal neural network and a spatial neural network. The temporal neural network captures motion information by Recurrent Neural Networks with the mutation of gated recurrent unit. The spatial neural network catches object information by using the deep CNN, to encode the CNN features as a bag of semantics with more discriminative representations. Both the temporal and spatial features are beneficial for event detection in a fully coupled way. Finally, we employ the generalized multiple kernel learning method to effectively fuse these two types of heterogeneous and complementary features for action recognition. Experiments on TRECVID MEDTest 14 dataset show that our method achieves better performance than the state of the art.

The capacity of recurrent neural networks to learn complex sequential patterns is improving. Recent developments such as Clockwork RNN, Stack RNN, Memory networks and Neural Turing Machine all aim to increase long-term memory capacity of recurrent neural networks. In this study, we investigate properties of Neural Turing Machine, compare it with ensembles of Stack RNN on artificial benchmarks and applied it to learn human mobility patterns. We show, that Neural Turing Machine based predictor outperformed not only n-gram based prediction, but also neighborhood based predictor, that was designed to solve this particular problem. Our models will be deployed in anti-drug police department to predict mobility of suspects.

Convolutional Neural Networks have achieved great performance in computer vision and pattern recognition applications. In this work, we propose the use of Convolutional Neural Networks in order to improve the handwritten mathematical symbol classification rate. We trained a Convolutional Neural Networks to classifying mathematical symbols into a proper class using Competition on Recognition of Online Handwritten Mathematical Expressions (CROHME) 2014 dataset. The results show that Convolutional Neural Networks outperforms the previous works in terms of accuracy.

Neurocomputing defined is a computational process based on the way the brain works, and the process of primitive neural emulation in software or hardware. Neurocomputing associated with neural network. Characteristics philosophically first neural network consists of learning. generalization, abstraction, applicability and both technically consists of nodes, characteristics, topology, learning rules. Based on a review of the literature observations by the author, many research publications about neurocomputing application made by several researchers. This paper introduces the field of neural networks by presenting a framework for designing ANN (neuron model, network structure and learning algorithm), and then proceed with a review of some of the most famous types of neural networks. Some examples of applications of all networks will be presented in this paper.

It is well known that the prototype patterns in associative memories can be represented by stable equilibrium points of cellular neural networks (CNNs). Therefore, the stability of equilibrium points of CNNs is critical in associative memories based on CNNs. In this paper, some criteria about the stability of CNNs are established. In fact, these criteria give some constraint conditions for the relationship of parameters of CNNs. Compared with the previous works, our results relax the conservatism of the relationship of parameters and extend the range of the values of parameters. Two design procedures on the parameters of CNNs are given to achieve associative memories under our criteria. Finally, an example is given to verify the theoretical results and design procedures.

In previous paper, the number of equilibrium points of a cellular neural network (CNN) was studied. However, the region of different values of equilibrium points of every cell in the CNN is not investigated. Therefore, in this paper, the regions of the number of different values of equilibrium points in cellular neural networks (CNNs) with thresholding activation function are obtained by the relationship between parameters of CNNs. The influence of the values of connection weights of CNNs on the number of equilibrium points for a cell is found. Some sufficient conditions are obtained by using the relationship among connection weights. Depending on these sufficient conditions, inputs and outputs of a CNN, the regions of the values of parameters can be obtained. Finally, some numerical simulations are presented to support the effectiveness of the theoretical analysis.