Artificial Neural Networks (ANNs) have been applied to a variety of classification and learning tasks. The use of evolutionary Algorithms (EA) as one of the fastest, robust and efficient global search techniques has a...
详细信息
Artificial Neural Networks (ANNs) have been applied to a variety of classification and learning tasks. The use of evolutionary Algorithms (EA) as one of the fastest, robust and efficient global search techniques has allowed different properties of artificial neural networks to be evolved. This paper proposes the possibility' of using differential evolution for Determining an ANN Architecture (DNNA). We explain how to use differential evolution's application for determining an ANN architecture. The approach we describe is innovative and has only been successfully applied and implemented for the first time, although the idea of Differential Evolution has been applied in various fields since the last decade. In this work, we proposed an algorithm based on Differential Evolution that uses a minimum number of user specified parameters in determining an ANN architecture. By using back-propagation algorithm to train the ANN architecture partially during the evolution process, DNNA is evaluated on five benchmark classification problems, namely, Cancer, Diabetes, Heart Disease, Thyroid, and the Australian Credit Card problem. Through performance analysis and simulation studies, we show that DNNA can produce ANN architecture with good generalization abilities, but with less number of training cycles when compared with an evolutionary programming approach and standard back-propagation.
A cluster analysis on a set of Retro-Transcribing viral proteomic sequences is described in this paper. A Lysine-Arginine concentration vector is calculated from the sequences and analyzed to identify correlations amo...
详细信息
ISBN:
(纸本)9781424452583
A cluster analysis on a set of Retro-Transcribing viral proteomic sequences is described in this paper. A Lysine-Arginine concentration vector is calculated from the sequences and analyzed to identify correlations among species. The computational strategy is based on the K-Means algorithm to partition the data into disjoint sets of points. A search method based on evolutionary programming is incorporated, in order to optimize the cluster structures. Experimental results show a number of interesting and unexpected similarities. These similarities could suggest bioelectronics relationships, in the context of the electronic mobility theory.
The nominal optimal tracker for the chaotic, nonlinear, interval system is first proposed in this paper. Initially we use an optimal linearization methodology to obtain the exact linear models of a class of discrete-t...
详细信息
The nominal optimal tracker for the chaotic, nonlinear, interval system is first proposed in this paper. Initially we use an optimal linearization methodology to obtain the exact linear models of a class of discrete-time, nonlinear, time-invariant systems at operating states of interest, so that the conventional tracker will work for the nonlinear systems. A prediction-based digital tracker using the state-matching digital redesign method from a predesigned, state-feedback, continuous-time tracker for a hybrid chaotic system is presented. Then, we discuss the case in which the system has unknown-but-bounded interval parameters. The proposed evolutionary programming (EP) technique yields the strongest species to survive, reproduce themselves, and create more outstanding offspring. The worst-case realization of the sampled-data, nonlinear, uncertain system represented by the interval form with respect to the implemented 'best' tracker is also found in this paper for demonstrating the effectiveness of the proposed tracker.
A Bayesian network is a graphical model for probabilistic relationships among a set of variables. Over the last two decades, the Bayesian network has become a popular representation for encoding uncer
A Bayesian network is a graphical model for probabilistic relationships among a set of variables. Over the last two decades, the Bayesian network has become a popular representation for encoding uncer
Recent developments in the global fuel markets imposed the need of increased fuel economy and cost effectiveness of sea-going vessels. Optimization of the ship's total energy system, as a whole, is now a demand of...
详细信息
Recent developments in the global fuel markets imposed the need of increased fuel economy and cost effectiveness of sea-going vessels. Optimization of the ship's total energy system, as a whole, is now a demand of the marine industry sector in order to address the significant increase of installation and operational costs. This study is focused on the synthesis, design and operation optimization of a marine energy system. A realistic example of a cruise liner energy system has been selected. Basic technology options have been identified and a generic energy system model has been constructed. Various configuration options.. types of technologies and existence of components have been incorporated in the generic system. In addition, time varying operational requirements for this cruise liner ship have been considered, resulting in a time dependent operation optimization problem. The complete optimization problem has been solved using a novel algorithm, inspired by evolutionary and social behavior metaphors. A parametric analysis with respect to the fuel price demonstrated changes in the optimum synthesis of the system. (C) 2007 Elsevier Ltd. All rights reserved.
The application of evolutionary computation methods in search and optimization has been growing over the past few decades. As a promising approach in metaheuristic optimization algorithms, differential evolution (DE) ...
详细信息
The application of evolutionary computation methods in search and optimization has been growing over the past few decades. As a promising approach in metaheuristic optimization algorithms, differential evolution (DE) has been attracting increasing attention for wide engineering applications including power engineering. Different from conventional evolutionary algorithms using predefined probability distribution function for mutation process, differential evolution exploits the differences of randomly sampled pairs of objective vectors for its mutation process. Consequently the variation between vectors will outfit the objective functions topographical information toward the optimization process, and therefore provides efficient global optimization capability. However, although DE is shown to be precise, fast as well as robust, its search efficiency will be impaired during solution process with fast descending diversity of population. In this paper, detailed numerical studies are carried out to propose the characterization of the performance of several DE mutation methods with and without fitness sharing scheme. All the approaches using the proposed modified DE are presented on an example in power system planning.
This article presents hybrid evolutionary programming (EP), particle swarm optimization (PSO), and sequential quadratic programming (SQP) methods to solve the dynamic economic dispatch problem (DEDP) of generating uni...
详细信息
This article presents hybrid evolutionary programming (EP), particle swarm optimization (PSO), and sequential quadratic programming (SQP) methods to solve the dynamic economic dispatch problem (DEDP) of generating units considering non-convex features. The non-convex feature considered is the valve-point effects, which is modeled in two different representations in the DEDP formulation. The proposed method is a two-phase optimizer In the first phase, the candidates are treated by both, the EP and PSO techniques to explore the solution space freely. In the second phase, the SQP method will be invoked when there is an improvement of solution (a feasible solution) in the first phase of the run. This hybrid optimization mechanism leads a better performance of the solution algorithm to effectively search the complex solution space. To validate the effectiveness of the proposed method, several non-convex DEDP test systems are studied and shown in general.
In this study, a new classification technique based on rough set theory and MEPAR-miner algorithm for association rule mining is introduced. Proposed method is called as 'Reduced MEPAR-miner Algorithm'. In the...
详细信息
In this study, a new classification technique based on rough set theory and MEPAR-miner algorithm for association rule mining is introduced. Proposed method is called as 'Reduced MEPAR-miner Algorithm'. In the method being improved rough sets are used in the preprocessing stage in order to reduce the dimensionality of the feature space and improved MEPAR-miner algorithms are then used to extract the classification rules. Besides, a new and an effective default class structure is also defined in this proposed method. Integrating rough set theory and improved MEPAR-miner algorithm, an effective rule mining structure is acquired. The effectiveness of our approach is tested on eight publicly available binary and n-ary classification data sets. Comprehensive experiments are performed to demonstrate that Reduced MEPAR-miner Algorithm can discover effective classification rules which are as good as (or better) the other classification algorithms. These promising results show that the rough set approach is a useful tool for preprocessing of data for improved MEPAR-miner algorithm. (c) 2007 Elsevier Inc. All rights reserved.
Using evolutionary programming (EP), monopulse Cassegrain antennas with four feeds can be designed for the desired sum gain, side lobe level and minimum possible antenna size. A method is proposed to achieve the optim...
详细信息
Using evolutionary programming (EP), monopulse Cassegrain antennas with four feeds can be designed for the desired sum gain, side lobe level and minimum possible antenna size. A method is proposed to achieve the optimum monopulse difference pattern. cos(q) (theta) type feeds are considered as the feed system and the final designs have been checked using real feeds. Proper cost functions are proposed to achieve the desired gain, side lobe level and optimum slope in the difference pattern, taking into account the feasibility of the feed system. The effects of the parameters involved in the optimization are investigated. (c) 2007 Elsevier GmbH. All rights reserved.
We present a new model for detection of noun phrases in unrestricted text, whose most outstanding feature is its flexibility: the system is able to recognize noun phrases similar enough to the ones given by the inferr...
详细信息
We present a new model for detection of noun phrases in unrestricted text, whose most outstanding feature is its flexibility: the system is able to recognize noun phrases similar enough to the ones given by the inferred noun phrase grammar. The system provides a probabilistic finite-state automaton able to recognize the part-of-speech tag sequences which define a noun phrase. The recognition flexibility is possible by using a very accurate set of rankings for the FSA transitions. These accurate rankings are obtained by means of an evolutionary algorithm, which works with both, positive and negative examples of the language, thus improving the system coverage while maintaining its precision. We have tested the system on different corpora and evaluated different aspects of the system performance. We have also investigated other ways of improving the performance such as the application of certain filters in the training sets. The comparison of our results with other systems has revealed a considerable performance improvement. (c) 2007 Elsevier B.V. All rights reserved.
暂无评论