1 |
An Adaptive Multiobjective Evolutionary Approach To Optimize Artmap Neural NetworksKaylani, Assem 01 January 2008 (has links)
This dissertation deals with the evolutionary optimization of ART neural network architectures. ART (adaptive resonance theory) was introduced by a Grossberg in 1976. In the last 20 years (1987-2007) a number of ART neural network architectures were introduced into the literature (Fuzzy ARTMAP (1992), Gaussian ARTMAP (1996 and 1997) and Ellipsoidal ARTMAP (2001)). In this dissertation, we focus on the evolutionary optimization of ART neural network architectures with the intent of optimizing the size and the generalization performance of the ART neural network. A number of researchers have focused on the evolutionary optimization of neural networks, but no research has been performed on the evolutionary optimization of ART neural networks, prior to 2006, when Daraiseh has used evolutionary techniques for the optimization of ART structures. This dissertation extends in many ways and expands in different directions the evolution of ART architectures, such as: (a) uses a multi-objective optimization of ART structures, thus providing to the user multiple solutions (ART networks) with varying degrees of merit, instead of a single solution (b) uses GA parameters that are adaptively determined throughout the ART evolution, (c) identifies a proper size of the validation set used to calculate the fitness function needed for ART's evolution, thus speeding up the evolutionary process, (d) produces experimental results that demonstrate the evolved ART's effectiveness (good accuracy and small size) and efficiency (speed) compared with other competitive ART structures, as well as other classifiers (CART (Classification and Regression Trees) and SVM (Support Vector Machines)). The overall methodology to evolve ART using a multi-objective approach, the chromosome representation of an ART neural network, the genetic operators used in ART's evolution, and the automatic adaptation of some of the GA parameters in ART's evolution could also be applied in the evolution of other exemplar based neural network classifiers such as the probabilistic neural network and the radial basis function neural network.
|
2 |
Estudo comparativo de diferentes representações cromossômicas nos algoritmos genéticos em problemas de sequenciamento da produção em job shopMódolo Junior, Valdemar 10 June 2015 (has links)
Submitted by Nadir Basilio (nadirsb@uninove.br) on 2016-06-01T14:43:08Z
No. of bitstreams: 1
Valdemar Modolo Junior.pdf: 2802590 bytes, checksum: f3956818acd10efc3244abc007294827 (MD5) / Made available in DSpace on 2016-06-01T14:43:08Z (GMT). No. of bitstreams: 1
Valdemar Modolo Junior.pdf: 2802590 bytes, checksum: f3956818acd10efc3244abc007294827 (MD5)
Previous issue date: 2015-06-10 / Among the optimization methods, the Genetic Algorithm (GA) has been producing good results in problems with high order of complexity, such as, for example, the production scheduling problem in job shop environment. The production sequencing problems must be translated into a mathematical representation, so that the AG can act. In this process we came up a problematic, the choice between different ways to represent the solution as some representations have limitations, how to present not feasible and / or redundant solutions. Therefore the aim of this study is to conduct a comparative study between different representations of the solution in the AG in production sequencing problems in job shop environments. Two representations of the solution were analyzed, the priority lists based and based on order of operations and compared with a binary representation, in the context of sequencing problem set defined by Lawrence (1984). The results were evaluated according to the total processing time (makespan), the computational cost and the proportion of generated feasible solutions. It was noticed that the representation of the solution based on order of operations, which produced 100% of feasible solutions, was the one that showed the best results although no convergence to the best known solution to every problem. / Dentre os métodos de otimização, o Algoritmo Genético (AG) vem produzindo bons resultados em problemas com ordem de complexidade elevada, como é o caso, por exemplo, do problema de sequenciamento da produção em ambiente job shop. Os problemas de sequenciamento da produção devem ser traduzidos para uma representação matemática, para que o AG possa atuar. Neste processo surgi uma problemática, a escolha entre as diferentes formas de se representar a solução visto que algumas representações apresentam limitações, como apresentar soluções não factíveis e/ou redundantes. Portanto o objetivo deste trabalho é realizar um estudo comparativo entre diferentes representações da solução no AG em problemas de sequenciamento da produção em ambientes job shop. Duas representações da solução foram analisadas, a baseada em listas de prioridades e a baseada em ordem de operações e comparada com uma representação binária, no contexto do conjunto de problemas de sequenciamento definidos por Lawrence (1984). Os resultados foram avaliados em função do tempo total de processamento (makespan), do custo computacional e da proporção de soluções factíveis geradas. Percebeu-se que, a representação da solução baseada em ordem de operações, a qual produziu 100% de soluções factíveis, foi a que mostrou os melhores resultados apesar de não apresentar convergência para a melhor solução conhecida em todos os problemas.
|
3 |
Analýza genetických algoritmů / Analysis of Genetic AlgorithmSnášelová, Petra January 2013 (has links)
This thesis deals with analysis of genetic algorithms. It is focused on various approaches to creation of new populations. A comparison between basic principles of operation of genetic algorithms and processes occurring in living organisms is drawn here. Some methods of application of particular steps of genetic algorithms are introduced and a suitability of the methods to certain types of problems is considered. The main goal in the thesis is to apply genetic algorithms in solving three types of optimization problems, namely the solution of functions with a single major extreme, functions with flat (slight) extreme and also functions with many local extremes.
|
Page generated in 0.1102 seconds