In the last few years, evolutionary computing (EC) approaches have been successfully used for many real world optimization applications in scientific and engineering areas. One of these areas is computational nanoscie...
详细信息
ISBN:
(纸本)9781424478354
In the last few years, evolutionary computing (EC) approaches have been successfully used for many real world optimization applications in scientific and engineering areas. One of these areas is computational nanoscience. Semi-empirical models with physics-based symmetries and properties can be developed by using EC to reproduce theoretically the experimental data. One of these semi-empirical models is the Valence Force Field (VFF) method for lattice properties. An accurate understanding of lattice properties provides a stepping stone for the investigation of thermal phenomena and has large impact in thermoelectricity and nano-scale electronic device design. The VFF method allows for the calculation of static properties like the elastic constants as well as dynamic properties like the sound velocity and the phonon dispersion. In this paper a parallel genetic algorithm (PGA) is employed to develop the optimal VFF model parameters for gallium arsenide (GaAs). This methodology can also be used for other semiconductors. The achieved results agree qualitatively and quantitatively with the experimental data.
Proteins are one of the most vital macromolecules on the cellular level. In order to understand the function of a protein, its structure needs to be determined. For this purpose, different computational approaches hav...
详细信息
ISBN:
(纸本)9783642246685
Proteins are one of the most vital macromolecules on the cellular level. In order to understand the function of a protein, its structure needs to be determined. For this purpose, different computational approaches have been introduced. geneticalgorithms can be used to search the vast space of all possible conformations of a protein in order to find its native structure. A framework for design of such algorithms that is both generic, easy to use and performs fast on distributed systems may help further development of geneticalgorithm based approaches. We propose such a framework based on a parallel master-slave model which is implemented in C++ and Message Passing interface. We evaluated its performance on distributed systems with a different number of processors and achieved a linear acceleration in proportion to the number of processing units.
There are significant challenges related to estimating the source term of the atmospheric release. Urged on by robots in performing emergency responding tasks, a fast and accurate algorithm for this inversion problem ...
详细信息
There are significant challenges related to estimating the source term of the atmospheric release. Urged on by robots in performing emergency responding tasks, a fast and accurate algorithm for this inversion problem is indispensable. Sometimes the NM simplex algorithm is efficient in the optimization problem, but sometimes the quality of convergence is unacceptable as a numerical breakdown, even for smooth and well-behaved functions. In contrast, full convergence might be seen in parallel genetic algorithms with a comparative slower convergence. In this paper we combine the PGA and the NM simplex algorithm by initializing simplex from the final individual of PGA results and obtaining the best vertex through simplex algorithm thereafter. A numerical simulation of the proposed algorithm shows noteworthy improvement of efficiency and robustness, compared with the PGA or the NM algorithm only.
In this paper a coarse-grain execution model for evolutionary algorithms is proposed and used for solving numerical and combinatorial optimization problems. This model does not use migration as the solution dispersion...
详细信息
In this paper a coarse-grain execution model for evolutionary algorithms is proposed and used for solving numerical and combinatorial optimization problems. This model does not use migration as the solution dispersion mechanism, in its place a more efficient population-merging mechanism is used that dynamically reduces the population size as well as the total number of parallel evolving populations. Even more relevant is the fact that the proposed model incorporates an entropy measure to determine how to merge the populations such that no valuable information is lost during the evolutionary process. Extensive experimentation, using geneticalgorithms over a well-known set of classical problems, shows the proposed model to be faster and more accurate than the traditional one.
This study presents new reliability models for k-out-of-n systems using a structured continuous-time Markov chain. The approach makes it comfortable to identify the characteristics of the lifetime of the system, such ...
详细信息
This study presents new reliability models for k-out-of-n systems using a structured continuous-time Markov chain. The approach makes it comfortable to identify the characteristics of the lifetime of the system, such as reliability and expected lifetime. To demonstrate the advantage, the analysis results of them for the multi-spectral camera system, which is the only payload of Korea multi-purpose satellite-2 as an example of a real system. Furthermore, existing studies on a k-out-of-n system with standby redundancy has provided the approximated reliability for it. In this paper, it is confirmed that the approximation error could have an effect on the reliability design for a system. Moreover, new versions of reliability optimization problems, redundancy allocation problem (RAP) and reliability-redundancy allocation problem(RRAP), are proposed. In order to maximize system reliability, they further determine the redundancy strategy, either active or standby redundancy for each k-out-of-n subsystem from the traditional problems. A parallel genetic algorithm is proposed for an RRAP modeled by nonlinear mixed integer programming. (C) 2017 Elsevier Ltd. All rights reserved.
This paper focuses on the development of a methodology to identify the latent factors leading to changes in the planned itineraries of travellers that result in their actual activity patterns. Specifically, we propose...
详细信息
This paper focuses on the development of a methodology to identify the latent factors leading to changes in the planned itineraries of travellers that result in their actual activity patterns. Specifically, we propose a way to utilise patterns of activities established by individuals across multiple days to generate possible alternative actions by these individuals when faced with conditions that produce a discrepancy between performed and planned patterns on a particular day. The choice alternatives, which are unobserved, are inferred by rules applied to comprehensive multiday data collected in Belgium, consisting of information regarding planned activity itineraries, performed activity/travel diaries, and demographics of travellers. These data are utilised to analyse and explore the underlying reasons preventing individuals from performing their planned activities on a given day, and to identify the influential parameters that lead individuals to trade their planned patterns with those actually performed. Using multiday data, we generate all possible combinations of categories of activities - mandatory, maintenance, discretionary, and pickup/drop off activities - that can form patterns for individuals. Under the assumption that the performed patterns have the closest utility to the planned patterns, we estimate the latent factors that influence travellers' time use behaviour using a multinomial probit choice structure in which the covariance structure of the choice alternatives is specified in terms of the overlap in activities. We further identify the 'costs' associated with making changes in planned agenda (replacing, inserting, or deleting an activity). These penalty values are estimated using 'parallel genetic algorithm', where the fitness function is the likelihood function estimated under the multinomial choice model structure. The results show that individuals' mobility decisions related to mandatory activities are more robust than those associated with the
In this study, a comprehensive methodology for overcoming the design problem of the Fuzzy ARTMAP neural network is proposed. The issues addressed are the sequence of training data for supervised learning and optimum p...
详细信息
In this study, a comprehensive methodology for overcoming the design problem of the Fuzzy ARTMAP neural network is proposed. The issues addressed are the sequence of training data for supervised learning and optimum parameter tuning for parameters such as baseline vigilance. A geneticalgorithm search heuristic was chosen to solve this multi-objective optimization problem. To further augment the ARTMAP's pattern classification ability, multiple ARTMAPs were optimized via geneticalgorithm and assembled into a classifier ensemble. An optimal ensemble was realized by the inter-classifier diversity of its constituents. This was achieved by mitigating convergence in the geneticalgorithms by employing a hierarchical parallel architecture. The best-performing classifiers were then combined in an ensemble, using probabilistic voting for decision combination. This study also integrated the disparate methods to operate within a single framework, which is the proposed novel method for creating an optimum classifier ensemble configuration with minimum user intervention. The methodology was benchmarked using popular data sets from UCI machine learning repository.
The paper is devoted to the problem of machine-made synthesis of control for robotic teams. The goal of synthesis is to find a multidimensional control function that depends on the current states of all robots. The sy...
详细信息
The paper is devoted to the problem of machine-made synthesis of control for robotic teams. The goal of synthesis is to find a multidimensional control function that depends on the current states of all robots. The synthesised control function provides any time the optimal control values to allow each robot achieving the objectives with the best value of functional quality. The approach is based on multilayer network operator method that belongs to a symbolic regression class. Formations of multi-robot systems require individual robots to satisfy their kinematic equations while constantly maintaining inter-robot dynamic constraints. Verification of these dynamic constraints on each iteration of the evolutionary algorithm greatly increases the computational costs of the numerical synthesis. In the paper we propose to accelerate existing designs through taking advantage of newest programming tools of MPI framework for automatic parallelization. Experiments show that our approach reduces greatly computational time. (C) 2017 The Authors. Published by Elsevier B.V.
In this paper, aiming at solving the problem arising from the application of the two-dimensional entropy method in double, threshold segmentation, which is time-consuming and highly complex, the author introduces the ...
详细信息
In this paper, aiming at solving the problem arising from the application of the two-dimensional entropy method in double, threshold segmentation, which is time-consuming and highly complex, the author introduces the parallelgenetic simulated annealing algorithm to optimize it. Besides, the author structures parallelgenetic simulated annealing algorithm to search for a two-dimensional maximum entropy value of the optimal threshold. This optimized algorithm shortens the calculation time, accelerates the speed of obtaining the optimal threshold and improves the efficiency of image segmentation. (C) 2015 Elsevier GmbH. All rights reserved.
Feature selection is a key step in data analysis. However, most of the existing feature selection techniques are serial and inefficient to be applied to massive data sets. We propose a feature selection method based o...
详细信息
ISBN:
(纸本)9781538608371
Feature selection is a key step in data analysis. However, most of the existing feature selection techniques are serial and inefficient to be applied to massive data sets. We propose a feature selection method based on a multi-population weighted intelligent geneticalgorithm to enhance the reliability of diagnoses in e-Health applications. The proposed approach, called PIGAS, utilizes a weighted intelligent geneticalgorithm to select a proper subset of features that leads to a high classification accuracy. In addition, PIGAS takes advantage of multi-population implementation to further enhance accuracy. To evaluate the subsets of the selected features, the KNN classifier is utilized and assessed on UCI Arrhythmia dataset. To guarantee valid results, leave-one-out validation technique is employed The experimental results show that the proposed approach outperforms other methods in terms of accuracy and efficiency. The results of the 16-class classification problem indicate an increase in the overall accuracy when using the optimal feature subset. Accuracy achieved being 99.70% indicating the potential of the algorithm to be utilized in a practical auto-diagnosis system. This accuracy was obtained using only half of features, as against an accuracy of 66.76% using all the features.
暂无评论