Results 1  10
of
12
An Exact and Direct Analytical Method for the Design of Optimally Robust CNN Templates
 IEEE TRANS. CIRCUITS & SYST.I
, 1999
"... In this paper, we present an analytical design approach for the class of bipolar cellular neural networks (CNN's) which yields optimally robust template parameters. We give a rigorous definition of absolute and relative robustness and show that all welldefined CNN tasks are characterized by a finit ..."
Abstract

Cited by 5 (2 self)
 Add to MetaCart
In this paper, we present an analytical design approach for the class of bipolar cellular neural networks (CNN's) which yields optimally robust template parameters. We give a rigorous definition of absolute and relative robustness and show that all welldefined CNN tasks are characterized by a finite set of linear and homogeneous inequalities. This system of inequalities can be analytically solved for the most robust template by simple matrix algebra. For the relative robustness of a task, a theoretical upper bound exists and is easily derived, whereas the absolute robustness can be arbitrarily increased by template scaling. A series of examples demonstrates the simplicity and broad applicability of the proposed method.
Adaptive Simulated Annealing in CNN Template Learning
, 1999
"... Introduction Opportunities for the application of template optimization (or "learning") for a Cellular Neural Network (CNN) [1] are prevalent insG hareas as pattern recognition andtexture clasturedG50G Itis highly des788Ed to employ an algorithm which not only can produce optimal simald0GT but whic ..."
Abstract

Cited by 4 (0 self)
 Add to MetaCart
Introduction Opportunities for the application of template optimization (or "learning") for a Cellular Neural Network (CNN) [1] are prevalent insG hareas as pattern recognition andtexture clasturedG50G Itis highly des788Ed to employ an algorithm which not only can produce optimal simald0GT but which can als findthesdthed5 e#ciently, in as sGGG a timeas posM507d Various templatelearning methods have been propos6E5 date [2]. In [3] and[4], a hybridDirectSearch methodandSimulatedAnnealing (SA) were inves8F gatedforDisford00G6d CNN template optimization. Another algorithm, whichhas been widelyuse for CNN template learningtasni is the Genetic Algorithm (GA) [5], [6]. A variant of GA, from aclas calledEvolutionaryStrategies was appliedin [7] to obtain featureextraction templates Inthis letter, we compare the performance of a recentlydeveloped optimization algorithm called Adaptive Simulated Annealing (ASA)agains GA. In a publisG87T8dsds sbl y, ASAsAd5TM5 outperformedGA on asG of
Automatic Design of Cellular Neural Networks by means of Genetic Algorithms: Finding a Feature Detector
, 1994
"... This paper aims to examine the use of genetic algorithms to optimize subsystems of cellular neural network architectures. The application at hand is character recognition: the aim is to evolve an optimal feature detector in order to aid a conventional classifier network to generalize across differen ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
This paper aims to examine the use of genetic algorithms to optimize subsystems of cellular neural network architectures. The application at hand is character recognition: the aim is to evolve an optimal feature detector in order to aid a conventional classifier network to generalize across different fonts. To this end, a performance function and a genetic encoding for a feature detector are presented. An experiment is described where an optimal feature detector is indeed found by the genetic algorithm. 1. Introduction We are interested in the application of cellular neural networks in computer vision. Genetic algorithms (GA's) [13] can serve to optimize the design of cellular neural networks. Although the design of the global architecture of the system could still be done by human insight, we propose that specific submodules of the system are best optimized using one or other optimization method. GA's are a good candidate to fulfill this optimization role, as they are well suited ...
Analogic CNN Computing: Architectural, Implementation, and Algorithmic Advances  a Review
"... : In this paper, first, an overview is given about the whole scenario of analogic CNN computing. Next, two areas on CNN Computing Technology are considered briefly: (i) the architectural advances, especially the variable resolution and adaptation in space, time, and value and (ii) the computational ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
: In this paper, first, an overview is given about the whole scenario of analogic CNN computing. Next, two areas on CNN Computing Technology are considered briefly: (i) the architectural advances, especially the variable resolution and adaptation in space, time, and value and (ii) the computational infrastructure from high level language and compiler to physical implementations. Three basic physical implementations are supposed : analogic CMOS, emulated digital CMOS and optical. The computational infrastructure is the same for all implementations, except the physical interfaces. 1. Introduction A few months ago, Intel shipped the first Tera FLOPS supercomputer consisting almost ten thousand 200 MHz Pentium microprocessors. In many image processing applications we really need this trillion operations per second, except the operations are special and do not require the 32 bit floating point accuracy. The alternative is the analogic CNN array computer performing about Tera equivalent op...
Texture segmentation by the 64x64 CNN chip
 in Proceedings of IEEE Workshop on CNN and Their Applications
, 2002
"... CNN’s fast image processing technology helps us to run highspeed filtering tasks for image enhancement, recognition or segmentation. Texture analysis is a specific task, since the whole image is processed massively parallel while we have a limited number of texturespecific filtering and evaluation ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
CNN’s fast image processing technology helps us to run highspeed filtering tasks for image enhancement, recognition or segmentation. Texture analysis is a specific task, since the whole image is processed massively parallel while we have a limited number of texturespecific filtering and evaluation steps. Former results of simulations and recognition results of simple CNN chips show that the CNN is an appropriate tool for this imageprocessing task. Now we see what the grayscale image processor CNN chip at its limited memory capability and datahandling/processing accuracy can complete for multitexture images. We demonstrate and compare some of our earlier CNNrelated texture analysis methods. Some methods to improve CNN configuration are proposed. 1 Introduction: CNN for texture recognition Texture recognition and segmentation ask for a massively parallel computation, based on a series of simplelike (mostly earlyvision effects) image processing steps. These steps are usually robust filtering and statistical evaluation effects. It is just a good role for the CNN applications. In [7,12] we see how the kernel matrices of a feedback & feedforward
Article A Novel Cloning Template Designing Method by Using an Artificial Bee Colony Algorithm for Edge Detection of CNN Based Imaging Sensors
, 2011
"... sensors ..."
A 0.8 m CMOS TwoDimensional Programmable MixedSignal FocalPlane Array Processor with OnChip Binary Imaging and Instructions Storage
"... Abstract—This paper presents a CMOS chip for the parallel acquisition and concurrent analog processing of twodimensional (2D) binary images. Its processing function is determined by a reduced set of 19 analog coefficients whose values are programmable with 7b accuracy. The internal programming si ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
Abstract—This paper presents a CMOS chip for the parallel acquisition and concurrent analog processing of twodimensional (2D) binary images. Its processing function is determined by a reduced set of 19 analog coefficients whose values are programmable with 7b accuracy. The internal programming signals are analog, but the external control interface is fully digital. Onchip nonlinear digitaltoanalog converters (DAC’s) map digitally coded weight values into analog control signals, using feedback to predistort their transfer characteristics in accordance to the response of the analog programming circuitry. This strategy cancels out the nonlinear dependence of the analog circuitry with the programming signal and reduces the influence of interchip technological parameters random fluctuations. The chip includes a small digital RAM memory to store eight sets of processing parameters in the periphery of the cell array and four 2D binary images spatially distributed over the processing array. It also includes the necessary control circuitry to realize the stored instructions in any order and also to realize programmable logic operations among images. The chip architecture is based on the cellular neural/nonlinear network universal machine (CNNUM). It has been fabricated in a 0.8 m singlepoly doublemetal technology and features 2 s operation speed (time required to process an image) and around 7b accuracy in the analog processing operations. Index Terms — Analog array processors, cellular neural networks, focal plane processors, vision chips. I.
Optimization of Incompletely Specified MVL Functions Using Genetic Algorithm
, 1996
"... . The genetic algorithm for optimization of logical polynomial forms of ..."
Extraction and Optimization of BSpline PBD Templates for Recognition of Connected Handwritten Digit Strings
, 2002
"... Recognition of connected handwritten digit strings is a challenging task due mainly to two problems: poor character segmentation and unreliable isolated character recognition. In this paper, we first present a rational Bspline representation of digit templates based on PixeltoBoundary Distance (P ..."
Abstract
 Add to MetaCart
Recognition of connected handwritten digit strings is a challenging task due mainly to two problems: poor character segmentation and unreliable isolated character recognition. In this paper, we first present a rational Bspline representation of digit templates based on PixeltoBoundary Distance (PBD) maps. We then present a neural network approach to extract Bspline PBD templates and an evolutionary algorithm to optimize these templates. In total, 1,000 templates (100 templates for each of 10 classes) were extracted from and optimized on 10,426 training samples from the NIST Special Database 3. By using these templates, a nearest neighbor classifier can successfully reject 90.7 percent of nondigit patterns while achieving a 96.4 percent correct classification of isolated test digits. When our classifier is applied to the recognition of 4,958 connected handwritten digit strings (4,555 2digit, 355 3digit, and 48 4digit strings) from the NIST Special Database 3 with a dynamic programming approach, it has a correct classification rate of 82.4 percent with a rejection rate of as low as 0.85 percent. Our classifier compares favorably in terms of correct classification rate and robustness with other classifiers that are tested.
Automatic ChipSpecific CNN Template Optimization Using Adaptive Simulated Annealing
 IN PROCEEDINGS OF EUROPEAN CONFERENCE ON CIRCUIT THEORY AND DESIGN (ECCTD’03), KRAKOW
, 2003
"... This paper describes a solution proposal for automatically tuning cellular neural network CNN templates for given CNN Universal Machine  CNNUM chips in order to make them respond in the same fashion as a simulator, i.e. to minimize or even eliminate the erroneous behavior of actual CNNUM chi ..."
Abstract
 Add to MetaCart
This paper describes a solution proposal for automatically tuning cellular neural network CNN templates for given CNN Universal Machine  CNNUM chips in order to make them respond in the same fashion as a simulator, i.e. to minimize or even eliminate the erroneous behavior of actual CNNUM chips. The approach uses measurements of actual CNNUM chips as part of the cost function for the adaptive simulated annealingASA algorithm to find an optimal template given an initial approximation, e.g. a template used for a simulator. The tuned templates are therefore customized versions that are expected to be much less sensitive to imperfections on the manufacturing process and other reasons of erroneous behavior of CNNUM chips. Results are presented for the binary and gray scale input cases. The automatic tuning was able to find better templates for all considered tasks. It is expected that the maturity of this technique will give to CNNUM chips enough reliability to compete with digital systems in terms of robustness in addition to advantages in speed.