In this paper, the implementation of a Parallel Genetic Algorithm (PGA) for the training stage, and the optimi zation of a monolithic and modular neural network, for pattern recognition are presented. The optimization con sists in obtaining the best architecture in layers, and neu rons per layer achieving the less training error in a shor ter time. The implementation was performed in a multicore architecture, using parallel programming techniques to exploit its resources. We present the results obtained in terms of performance by comparing results of the training stage for sequential and parallel implementations.
In this paper, we propose a new optimization algorithm for soft computing problems, which is inspired on a nature paradigm: the reaction methods existing on chemistry, and the way the elements combine with each other to form compounds, in other words, quantum chemistry. This paper is the first approach for the proposed method, and it presents the background, main ideas, desired goals and preliminary results in optimization.
This paper describes a modular neural network (MNN) with fuzzy integration for the problem of signature recognition. Currently, biometric identification has gained a great deal of research interest within the pattern recognition community. For instance, many attempts have been made in order to automate the process of identifying a person’s handwritten signature; however this problem has proven to be a very difficult task. In this work, we propose a MNN that has three separate modules, each using different image features as input, these are: edges, wavelet coefficients, and the Hough transform matrix. Then, the outputs from each of these modules are combined using a Sugeno fuzzy integral and a fuzzy inference system. The experimental results obtained using a database of 30 individual’s shows that the modular architecture can achieve a very high 99.33% recognition accuracy with a test set of 150 images. Therefore, we conclude that the proposed architecture provides a suitable platform to build a signature recognition system. Furthermore we consider the verification of signatures as false acceptance, false rejection and error recognition of the MNN.
This paper proposes a novel method for genetic optimi zation of triangular and trapezoidal membership functions of fuzzy systems, for hardware applications such as the FPGA (Field Programmable Gate Array). This method con sists in taking only certain points of the membership func tions, with the purpose of giving more efficiency to the algorithm. The genetic algorithm was tested in a fuzzy con troller to regulate engine speed of a direct current (DC) motor, using the Xilinx System Generator (XSG) toolbox of Matlab, which simulate VHDL (Very High Description Lang uage) code.
We describe in this paper a comparative study between Fuzzy Inference Systems as methods of integration in modular neural networks for multimodal biometry. These methods of integration are based on techniques of type-1 fuzzy logic and type-2 fuzzy logic. Also, the fuzzy systems are optimized with simple genetic algorithms. First, we considered the use of type-1 fuzzy logic and later the approach with type-2 fuzzy logic. The fuzzy systems were developed using genetic algorithms to handle fuzzy inference systems with different membership functions, like the triangular, trapezoidal and Gaussian; since these algorithms generate the fuzzy systems automatically. Then the response integration of the modular neural network was tested with the optimized fuzzy integration systems. The comparative study of type-1 and type-2 fuzzy inference systems was made to observe the behavior of the two different integration methods for modular neural networks for multimodal biometry.
Combining the outputs of multiple neural networks has been used in Ensemble architectures to improve the decision accuracy in many applications fields, including pattern recognition, in particular for the case of fingerprints. In this paper, we describe a set of experiments performed in order to find the optimal individual networks in terms of the architecture and training rule. In the second step, we used the fuzzy Sugeno Integral to integrate results of the ensemble neural networks. This method combines objective evidence in the form of the network's outputs, with subjective measures of their performance. In the third step, we used a Fuzzy Inference System for the decision process of finding the output of the ensemble neural networks, and finally a comparison of experimental results between Fuzzy Sugeno Integral and the Fuzzy Inference System are presented.
The Particle Swarm Optimization (PSO) and the Genetic Algorithms (GA) have been used successfully in solving problems of optimization with continuous and combinatorial search spaces. In this paper the results of the application of PSO and GAs for the optimization of mathematical functions are presented. These two methodologies have been implemented with the goal of making a comparison of their performance in solving complex optimization problems. This paper describes a comparison between a GA and PSO for the optimization of complex mathematical functions.
We describe in this paper the combination of soft computing techniques and fractal theory for achieving intelligent manufacturing. Soft computing techniques can be used to develop hybrid intelligent systems. Fractal theory can be used to analyze the geometrical complexity of natural and artificial objects. The careful combination of soft computing and fractal theory can provide us with a good mix of intelligent techniques and fractal mathematical tools, which can help in achieving automation of manufacturing processes. We consider in this paper several manufacturing and automation problems that are efficiently solved with the proposed approach.
We describe in this paper a general overview oj the analysis and design of hybrid intelligent systems for pattern recognition applications. Hybrid intelligent systems can be developed by a careful combination of several soft-computing techniques. The combination of soft computing techniques has to take advantage of the capabilities of each technique in solving port of the pattern recognition problem. We review the problems of face, fingerprint and mice recognition and their soiution using hybrid intelligent systems. Recognition rates achieved with the hybrid approaches are comparable with the best approaches known for solving these recognition problems.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.