Cellular Neural Network Structure Research Articles

Initial state‐dependent implementation of logic gates with memristive neurons

AbstractThis study introduces a simple memristor cellular neural network structure, a minimalist configuration with only two cells, designed to concurrently address two logic problems. The unique attribute of this system lies in its adaptability, where the nature of the implemented logic gate, be it AND, OR, and XOR, is determined exclusively by the initial states of the memristors. The memristors' state, alterable through current flow, allows for dynamic manipulation, enabling the setting of initial conditions and consequently, a change in the circuit's functionality. To optimize the parameters of this dynamic system, contemporary machine learning techniques are employed, specifically gradient descent optimization. Through a case study, the potential of leveraging intricate circuit dynamics is exemplified to expand the spectrum of problems solvable with a defined number of neurons. This work not only underscores the significance of adaptability in logical circuits but also demonstrates the efficacy of memristive elements in enhancing problem‐solving capabilities.

Building cellular neural network templates with a hardware friendly learning algorithm

Abstract A general solution for the construction of Cellular Neural Network (CNN) weights (cloning template) with Random Weight Change (RWC) algorithm is proposed. A target image for each input image is prepared via a sketch or any other kind of image processing technique for learning of Cellular Neural Network templates. A vector of randomly generated small values is added to the original weights and tested upon the input-target image pair. As a result, if the learning error decreases, the weight is taken for learning in the next iteration and updated using the same vector of random values. Otherwise, a new random vector for updating the weights is regenerated. One of the strong benefits of the proposed weight learning method is the simplicity of its learning algorithm and hence a simpler hardware architecture. Moreover the proposed method provides a unified solution to the problem of learning CNN templates without having to modify the original CNN structure and is applicable for all types of CNNs and input images. Successful learning of templates for various image processing tasks using different CNN structures are also demonstrated in this paper.

RETRACTED ARTICLE: A new method for skull stripping in brain MRI using multistable cellular neural networks

This study proposes a new method on “detecting brain region in MRI data”. This task is generally named as “skull stripping” in the literature. The algorithm is developed by using the cellular neural networks (CNNs) and multistable CNN structures. It also includes a contrast enhancement and noise reduction algorithm. The algorithm is named as multistable cellular neural network on MRI for skull stripping (mCNN-MRI-SS). Three different case studies are performed for measuring the success of the algorithm. Also a fourth case study is performed to evaluate the supporting algorithm, the CEULICA. First two evaluations are performed by using well-known MIDAS-NAMIC and Brainweb databases, which are properly organized Talairach-compatible databases. The third database was obtained from the research and application hospital of Necmettin Erbakan University Meram Faculty of Medicine. These MRI data were not Talairach-compatible and less sampled. The algorithm achieved 0.595 Jaccard, 0.744 Dice, 0.0344 TPF and 0.383 TNF mean values with the Brainweb T1-weighted images and 0.837 Jaccard, 0.898 Dice, 0.0124 TPF and 0.1511 TNF mean values with the MIDAS-NAMIC T2-weighted images. The algorithm achieved 0.8297 Jaccard, 0.9012 Dice, 0.0951 TPF and 0.1225 TNF mean values and achieved with the obtained data the best values among the other algorithms. As a result, it can be claimed that algorithm performs best with the non-Talairach-compatible MRI data due to its nature of performing at cellular level.

Detection of microcalcification in digitized mammograms with multistable cellular neural networks using a new image enhancement method: automated lesion intensity enhancer (ALIE)

Microcalcification detection is a very important issue in early diagnosis of breast cancer. Generally physicians use mammogram images for this task; however, sometimes analyzing these images become a hard task because of problems in images such as high brightness values, dense tissues, noise, and insufficient contrast level. In this paper, we present a novel technique for the task of microcalcification detection. This technique consists of three steps. The first step is focused on removing pectoral muscle and unnecessary parts from the mammogram images by using cellular neural networks (CNNs), which makes this a novel process. In the second step, we present a novel image enhancement technique focused on enhancing lesion intensities called the automated lesion intensity enhancer (ALIE). In the third step, we use a special CNN structure, named multistable CNNs. After applying the combination of these methods on the MIAS database, we achieve 82.0% accuracy, 90.9% sensitivity, and 52.2% specificity values.

Modular Cellular Neural Network Structure for Wave-Computing-Based Image Processing

This paper introduces the modular cellular neural network (CNN), which is a new CNN structure constructed from nine one-layer modules with intercellular interactions between different modules. The new network is suitable for implementing many image processing operations. Inputting an image into the modules results in nine outputs. The topographic characteristic of the cell interactions allows the outputs to introduce new properties for image processing tasks. The stability of the system is proven and the performance is evaluated in several image processing applications. Experiment results on texture segmentation show the power of the proposed structure. The performance of the structure in a real edge detection application using the Berkeley dataset BSDS300 is also evaluated.

A Novel Cloning Template Designing Method by Using an Artificial Bee Colony Algorithm for Edge Detection of CNN Based Imaging Sensors

Cellular Neural Networks (CNNs) have been widely used recently in applications such as edge detection, noise reduction and object detection, which are among the main computer imaging processes. They can also be realized as hardware based imaging sensors. The fact that hardware CNN models produce robust and effective results has attracted the attention of researchers using these structures within image sensors. Realization of desired CNN behavior such as edge detection can be achieved by correctly setting a cloning template without changing the structure of the CNN. To achieve different behaviors effectively, designing a cloning template is one of the most important research topics in this field. In this study, the edge detecting process that is used as a preliminary process for segmentation, identification and coding applications is conducted by using CNN structures. In order to design the cloning template of goal-oriented CNN architecture, an Artificial Bee Colony (ABC) algorithm which is inspired from the foraging behavior of honeybees is used and the performance analysis of ABC for this application is examined with multiple runs. The CNN template generated by the ABC algorithm is tested by using artificial and real test images. The results are subjectively and quantitatively compared with well-known classical edge detection methods, and other CNN based edge detector cloning templates available in the imaging literature. The results show that the proposed method is more successful than other methods.

Implementation of multi-scroll Chua’s circuit by hybrid single electron transistor and metal oxide semiconductor structure

Based on the structure of cellular neural network, multi-scroll Chua’s circuit is implemented by the nanoelectronic device of hybrid single electron transistor and metal oxide semiconductor (SETMOS) structure with its negative differential resistance characteristic. The basic dynamical properties, including phase portrait, bifurcation diagram, Lyapunov exponent spectrum, Poincaré mapping and power spectrum are studied by theoretic analysis and numerical simulation. The validity and the feasibility of three-order Chua’s circuit with four scrolls are further confirmed by the circuit simulation experiment. Finally, the results show that the negative differential resistance characteristic of SETMOS determines complex dynamical behaviors of multi-scroll Chua’s circuit. Also, the designed circuit has simple structure and is easy to realize.

Noise Reduction in CMOS Image Sensor Using Cellular Neural Networks with a Genetic Algorithm

In this paper, Cellular Neural Networks using genetic algorithm (GA-CNNs) are designed for CMOS image noise reduction. Cellular Neural Networks (CNNs) could be an efficient way to apply to the image processing technique, since CNNs have high-speed parallel signal processing characteristics. Adaptive CNNs structure is designed for the reduction of Photon Shot Noise (PSN) changed according to the average number of photons, and the design of templates for adaptive CNNs is based on the genetic algorithm using real numbers. These templates are optimized to suppress PSN in corrupted images. The simulation results show that the adaptive GA-CNNs more efficiently reduce PSN than do the other noise reduction methods and can be used as a high-quality and low-cost noise reduction filter for PSN. The proposed method is designed for real-time implementation. Therefore, it can be used as a noise reduction filter for many commercial applications. The simulation results also show the feasibility to design the CNNs template for a variety of problems based on the statistical image model.

THE IDENTIFICATION OF COUPLED MAP LATTICE MODELS FOR AUTONOMOUS CELLULAR NEURAL NETWORK PATTERNS

The identification problem for spatiotemporal patterns which are generated by autonomous Cellular Neural Networks (CNN) is investigated in this paper. The application of traditional identification algorithms to these special spatiotemporal systems can produce poor models due to the inherent piecewise nonlinear structure of CNN. To solve this problem, a new type of Coupled Map Lattice model with output constraints and corresponding identification algorithms are proposed in the present study. Numerical examples show that the identified CML models have good prediction capabilities even over the long term, and the main dynamics of the original patterns appears to be well represented.

Study of chaos based on single electron device

Chua′s circuit in structure of cellular neural networks is realized by the hybrid device of single electron transistor and metal oxide semiconductor, named SETMOS. Then single scroll and double scrolls are obtained. SETMOS transconductance amplifier and SETMOS voltage comparator are designed, and the double-scroll-like chaotic circuit built of SETMOS is proposed. The double-scroll-like chaos attractor is verified by simulation. All the simulation results show that the designed circuits have the characteristics of simple structure and low power dissipation, and they can further improve the density of integrated circuits. It also provides a new method for the chaos to be used in engineering.

NEW LEARNING METHOD FOR CELLULAR NEURAL NETWORKS TEMPLATE BASED ON COMBINATION BETWEEN ROUGH SETS AND GENETIC PROGRAMMING

ABSTRACT A new learning algorithm for space invariant Cellular Neural Network (CNN) is introduced. Learning is formulated as an optimization problem by combining rough sets and genetic programming. Rough Sets approach has been selected for creating priori knowledge about the actual effective cells, determining their significance in classifying the output, and discovering the optimal CNN structure. According to the lattice of CNN architecture and depending on the priori knowledge gained by rough sets, genetic programming will be used in deriving the cloning template. Exploration of any stable domain is possible by the current approach. Details of the algorithm are discussed and several application results are shown.

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.

Toward CNN Chip-Specific Robustness

The promising potential of cellular neural networks (CNN) has resulted in the development of several template design methods. The CNN universal machine (CNN-UM), a programmable CNN, has made it possible to create image-processing algorithms that run on this platform. However, very large-scale integration implementations of CNN-UMs presented parameter deviations that do not occur on ideal CNN structures. Consequently, new design methods were developed aiming at more robust templates. Although these new templates were indeed more robust, erroneous behavior can still be observed. An alternative for chip-independent robustness is chip-specific optimization, where the template is targeted to an individual chip. This paper describes a solution proposal in this sense to automatically tune templates in order to make the chip react as an ideal CNN structure. The approach uses measurements of actual CNN-UM chips as part of the cost function for a global optimization method to find an optimal template given an initial approximation. Further improvements are achieved by generating chip-specific robust templates by doing a search for the best template among the optimal ones. The tuned templates are therefore customized versions that are expected to be much less sensitive to imperfections on the operation of CNN-UM chips. Results are presented for the binary and grayscale cases, including the case of grayscale output. It is expected that as this technique matures, it will give CNN-UM chips enough reliability to compete with digital systems in terms of robustness in addition to advantages of speed.

BIFURCATIONS AND CHAOS IN CELLULAR NEURAL NETWORKS

The dynamic behavior of first-order autonomous space invariant cellular neural networks (CNNs) is investigated. It is shown that complex dynamics may occur in very simple CNN structures, described by two-dimensional templates that present only vertical and horizontal couplings. The bifurcation processes are analyzed through the computation of the limit cycle Floquet's multipliers, the evaluation of the Lyapunov exponents and of the signal spectra. As a main result a detailed and accurate two-dimensional bifurcation diagram is reported. The diagram allows one to distinguish several regions in the parameter space of a single CNN. They correspond to stable, periodic, quasi-periodic, and chaotic behavior, respectively. In particular it is shown that chaotic regions can be reached through two different routes: period doubling and torus breakdown. We remark that most practical CNN implementations exploit first order cells and space-invariant templates: so far only a few examples of complex dynamics and no complete bifurcation analysis have been presented for such networks.

Analog cellular locomotion control of hexapod robots

This article discusses analog neural processing structures for artificial locomotion in mechatronic devices. The main inspiration comes from the biological paradigm of the central pattern generator, used to model the neural populations responsible for locomotion planning and control in animals. We start by considering locomotion by legs as a complex spatiotemporal, nonlinear dynamic system, modeled referring to particular types of reaction-diffusion nonlinear partial differential equations. Spatiotemporal phenomena are then obtained by implementing the whole mathematical model on a new reaction-diffusion cellular neural network (CNN) architecture. Wavelike solutions as well as patterns are obtained and are able to induce and control locomotion in some prototypes of biologically inspired walking machines. The CNN structure is subsequently designed using analog circuits; this makes it possible to generate locomotion in real time and also to control transition between several types of locomotion. The methodology presented is applied to the experimental prototype of a hexapod robot.

A learnable cellular neural network structure with ratio memory for image processing

In this paper, a learnable cellular neural network (CNN) with space-variant templates and ratio memory (RM) called the RMCNN, is proposed and analyzed. By incorporating both a modified Hebbian learning rule and RM into the CNN architecture, the RMCNN as an associative memory can generate the absolute weights and then transform them into the ratioed A-template weights as the ratio memories for recognition of noisy input patterns. It is found from simulation results that, due to the feature enhancement effect of RM, the RMCNN under constant leakage on template coefficients can store and recognize more patterns than CNN associative memories without RM, but with the same learning rule and the same constant leakage on space-variant template coefficients. For 9/spl times/9 (18/spl times/18) RMCNNs, three (five) patterns can be learned, stored and recognized. Based upon the RMCNN architecture, an experimental CMOS 9/spl times/9 RMCNN chip is designed and fabricated by using 0.35 /spl mu/m CMOS technology. The measurement results have successfully verified the correct functions of RMCNN.

Multi‐template approach to realize central pattern generators for artificial locomotion control

AbstractBiologically inspired control of artificial locomotion often makes use of the concept of central pattern generator (CPG), a network of neurons establishing the locomotion pattern within a lattice of neural activity. In this paper a new approach, based on cellular neural networks (CNNs), for the design of CPGs is presented. From a biological point of view this new approach includes an approximated chemical synapse realized and implemented in a CNN structure. This allows to extend the results, previously obtained with a reaction‐diffusion‐CNN (RD‐CNN) for the locomotion control of a hexapod robot, to a more general class of artificial CPGs in which the desired locomotion pattern and the switching among patterns are realized by means of a spatio‐temporal algorithm implemented in the same CNN structure. Copyright © 2002 John Wiley & Sons, Ltd.

A SIMPLE NEURAL NETWORK APPROACH TO INVARIANT IMAGE RECOGNITION

In this paper we propose a simple neural network architecture for invariant image recognition. The proposed neural network architecture contains three specialized modules. The neurons from the first module are connected in a cellular neural network structure, which is responsible for image processing: edge detection and segmentation. The second module is a feed forward neural network for invariant feature extraction from the sensorial layer: computation of the pair distribution function and bond angle distribution function. The third module is responsible for image classification. An application to the face recognition problem is also presented.

IMPLEMENTING BACK-PROPAGATION- THROUGH-TIME LEARNING ALGORITHM USING CELLULAR NEURAL NETWORKS

In a programmable (multistage) cellular neural network (CNN) structure, the CPU is a CNN universal chip which supports massively parallel computations on patterns and images, including videos. In this paper, we decompose the structure of a class of simultaneous recurrent networks (SRN) into a CNN program and run it on a von Neumann-like stored program CNN structure. To train the SRN, we map the back-propagation-through-time (BTT) learning algorithm into a sequence of CNN subroutines to achieve real-time performance via a CNN universal chip. By computing in parallel, the CNN universal chip can be programmed to implement in real time the BTT learning algorithm, which has a very high time complexity. An estimate of the time complexity of the BTT learning algorithm based on the CNN universal chip is presented. For small-scale problems, our simulation results show that a CNN implementation of the BTT learning algorithm for a two-dimensional SRN is at least 10,000 times faster than that based on state-of-the-art sequential workstations. For the few large-scale problems which we have so far simulated, the CNN implemented BTT learning algorithm maintained virtually the same time complexity with a learning time of a few seconds, while those implemented on state-of-the-art sequential workstations dramatically increased their time complexity, often requiring several days of running time. Several examples are presented to demonstrate how efficiently a CNN universal chip can speed up the learning algorithm for both off-line and on-line applications.

Reaction-diffusion CNN algorithms to generate and control artificial locomotion

In this paper a physiological-behavioral approach to neural processing is used to realize artificial locomotion in mechatronic devices. The task has been realized by using a particular model of reaction-diffusion cellular neural networks (RD-CNN's) generating autowave fronts as well as Turing patterns. Moreover a programmable hardware cellular neural network structure is presented in order to model, generate, and control in real time some biorobots. The programmable hardware implementation gives the possibility of generating locomotion in real time and also to control the transition among several types of locomotion, with particular attention to hexapodes. The approach proposed allows not only the design of walking robots, but also the ability to build structures able to efficiently solve typical problems in industrial automation, such as online routing of objects moved on conveyor belts.