الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 96 |  |  |
| | | from | 96 | | |
Abstract
In this thesis we present new methodologies for improving the robustness of Simulators which are based on Cellular Neural Networks (CNN). Such a system can process information at very high speeds, only comparable to today’s supercomputers. The regular lattice architecture of CNNs allows massive parallelism, which makes it very suitable for performance-demanding applications in image processing. Its reduced size and power consumption make it easy to embed in portable appliances. In this thesis we describe methodologies for improving existing CNN Simulators. At first we developed a single layer CNN Simulator which used some numerical integration algorithms which have the efficiency of minimizing the error while the simulator making some operations on some images like edge detection and average template. These numerical integration algorithms are RK4(2), RK4(3), RK6(4), RK8(6), RK84(5) and neural networks meet the need in different ways, like Depending on the accuracy required for the simulation, CPU time used and the quality measures of the pictures.
For this simulator, We presented a new global optimization method that is suitable for CNN optimization. The new method of Coupled Simulated Annealing makes use of coupling in order to allow multiple Simulated Annealing (SA) processes to cooperate toward finding the global optimum of multi-modal and multidimensional optimization problems. A number of proof-of-concept applications is presented in order to show the effectiveness of our methodologies.
On the other hand, we also have developed a Time-Multiplexing CNN Simulator which used some of the numerical integration algorithms which are used in Manipulating Single Layer CNN Simulator like RK4(2), RK4(3), RK6(4), RK7(5), RK8(6). Also we added another two new numerical integration algorithms which they are The RK-Embedded Centroidal Mean(RKECM) and the RK-Embedded Harmonic Mean(RKEHM) because of their good results regarding the quality measures of the uses images.
For this simulator, we presented another new global optimization method based on Genetic Algorithm ( GA ) which is suitable for CNN Optimization which are genetic algorithms. Genetic algorithm is a learning algorithm based on the mechanism of natural selection and genetics, which have proved to be effective in a number of applications. It works with a binary coding of the parameter set, searches from a number of points of the parameter space. It uses only the cost function during the optimization.