A new parameter-estimation algorithm, which minimises the cross-validated prediction error for linear-in-the-parameter models, is proposed, based on stacked regression and an evolutionary algorithm. It is initially sh...
详细信息
A new parameter-estimation algorithm, which minimises the cross-validated prediction error for linear-in-the-parameter models, is proposed, based on stacked regression and an evolutionary algorithm. It is initially shown that cross-validation is very important for prediction in linear-in-the-parameter models using a criterion called the mean dispersion error (MDE). Stacked regression, which can be regarded as a sophisticated type of cross-validation, is then introduced based on an evolutionary algorithm, to produce a new parameter-estimation algorithm, which preserves the parsimony of a concise model structure that is determined using the forward orthogonal least-squares (OLS) algorithm. The PRESS prediction errors ale used for cross-validation, and the sunspot and Canadian lynx time series are used to demonstrate the new algorithms.
Nowadays, a promising way to obtain hybrid metaheuristics concerns the combination of several search algorithms with strong specialization in intensification and/or diversification. The flexible architecture of evolut...
详细信息
Nowadays, a promising way to obtain hybrid metaheuristics concerns the combination of several search algorithms with strong specialization in intensification and/or diversification. The flexible architecture of evolutionary algorithms allows specialized models to be obtained with the aim of providing intensification and/or diversification. The outstanding role that is played by evolutionary algorithms at present justifies the choice of their specialist approaches as suitable ingredients to build hybrid metaheuristics. This paper focuses on hybrid metaheuristics with evolutionary algorithms specializing in intensification and diversification. We first give an overview of the existing research on this topic, describing several instances grouped into three categories that were identified after reviewing specialized literature. Then, with the aim of complementing the overview and providing additional results and insights on this line of research, we present an instance that consists of an iterated local search algorithm with an evolutionary perturbation technique. The benefits of the proposal in comparison to other iterated local search algorithms proposed in the literature to deal with binary optimization problems are experimentally shown. The good performance of the reviewed approaches and the suitable results shown by our instance allow an important conclusion to be achieved: the use of evolutionary algorithms specializing in intensification and diversification for building hybrid metaheuristics becomes a prospective line of research for obtaining effective search algorithms. (C) 2009 Elsevier Ltd. All rights reserved.
We analyse the performance of well-known evolutionary algorithms, the (1 + 1) EA and the (1 + similar to) EA, in the prior noise model, where in each fitness evaluation the search point is altered before the evaluatio...
详细信息
We analyse the performance of well-known evolutionary algorithms, the (1 + 1) EA and the (1 + similar to) EA, in the prior noise model, where in each fitness evaluation the search point is altered before the evaluation with probability p. We present refined results for the expected optimisation time of these algorithms on the function -LeadingOnes, where bits have to be optimised in sequence. Previous work showed that the (1 + 1) EA on LeadingOnes runs in polynomial expected time if p = O((log n)/n2) and needs superpolynomial expected time if p = similar to((log n)/n), leaving a huge gap for which no results were known. We close this gap by showing that the expected optimisation time is similar to(n2) . exp(similar to(min{pn2, n})) for all p = 1/2, allowing for the first time to locate the threshold between polynomial and superpolynomial expected times at p = similar to((log n)/n2). Hence the (1 + 1) EA on -LeadingOnes is surprisingly sensitive to noise. We also show that offspring populations of size similar to = 3.42 log n can effectively deal with much higher noise than known before. Finally, we present an example of a rugged landscape where prior noise can help to escape from local optima by blurring the landscape and allowing a hill climber to see the underlying gradient. We prove that in this particular setting noise can have a highly beneficial effect on performance.
The performance of evolutionary algorithms (EAs) may heavily depend severely on a suitable choice of parameters such as mutation and crossover rates. Several methods to adjust those parameters have been developed in o...
详细信息
The performance of evolutionary algorithms (EAs) may heavily depend severely on a suitable choice of parameters such as mutation and crossover rates. Several methods to adjust those parameters have been developed in order to enhance EA performance. For this purpose, it is important to understand the EA dynamics, i.e., to appreciate the behavior of the population. Hence, this paper presents a new model of population dynamics to describe and predict the diversity in any particular generation. The formulation is based on selecting the probability density function of each individual. The population dynamics proposed is modeled for a generational population. The model was tested in several case studies of different population sizes. The results suggest that the prediction error decreases as the population size increases. (C) 2012 Elsevier B. V. All rights reserved.
The initial population of an evolutionary algorithm is an important factor which affects the convergence rate and ultimately its ability to find high quality solutions or satisfactory solutions for that matter. If com...
详细信息
The initial population of an evolutionary algorithm is an important factor which affects the convergence rate and ultimately its ability to find high quality solutions or satisfactory solutions for that matter. If composed of good individuals it may bias the search towards promising regions of the search space right from the beginning. Although, if no knowledge about the problem at hand is available, the initial population is most often generated completely random, thus no such behavior can be expected. This paper proposes a method for initializing the population that attempts to identify i.e., to get close to promising parts of the search space and to generate (relatively) good solutions in their proximity. The method is based on clustering and a simple Cauchy mutation. The results obtained on a broad set of standard benchmark functions suggest that the proposed method succeeds in the aforementioned which is most noticeable as an increase in convergence rate compared to the usual initialization approach and a method from the literature. Also, insight into the usefulness of advanced initialization methods in higher-dimensional search spaces is provided, at least to some degree, by the results obtained on higher-dimensional problem instances-the proposed method is beneficial in such spaces as well. Moreover, results on several very high-dimensional problem instances suggest that the proposed method is able to provide a good starting position for the search. (C) 2016 Elsevier Ltd. All rights reserved.
Two evolutionary algorithms (EAs) are assessed in this paper to design optimal operating rules (ORs) for Water Resource Systems (WRS). The assessment is established through a parameter analysis of both algorithms in a...
详细信息
Two evolutionary algorithms (EAs) are assessed in this paper to design optimal operating rules (ORs) for Water Resource Systems (WRS). The assessment is established through a parameter analysis of both algorithms in a theoretical case, and the methodology described in this paper is applied to a complex, real case. These two applications allow us to analyse an algorithm's properties and performance by defining ORs, how an algorithm's termination/convergence criteria affect the results and the importance of decision-makers participating in the optimisation process. The former analysis reflects the need for correctly defining the important algorithm parameters to ensure an optimal result and how the greater number of termination conditions makes the algorithm an efficient tool for obtaining optimal ORs in less time. Finally, in the complex real case application, we discuss the participation value of decision-makers toward correctly defining the objectives and making decisions in the post-process. (C) 2014 Elsevier Ltd. All rights reserved.
Modern cities are currently facing rapid urban growth and struggle to maintain a sustainable development. In this context, "eco-neighbourhoods" became the perfect place for testing new innovative ideas that ...
详细信息
Modern cities are currently facing rapid urban growth and struggle to maintain a sustainable development. In this context, "eco-neighbourhoods" became the perfect place for testing new innovative ideas that would reduce congestion and optimize traffic flow. The main motivation of this work is a true and stated need of the Department of Transport in Nancy, France, to improve the traffic flow in a central eco-neighbourhood currently under reconfiguration, reduce travel times and test various traffic control scenarios for a better interconnectivity between urban intersections. Therefore, this paper addresses a multi-objective simulation-based signal control problem through the case study of "Nancy Grand Coeur" (NGC) eco-neighbourhood with the purpose of finding the optimal traffic control plan to reduce congestion during peak hours. Firstly, we build the 3D mesoscopic simulation model of the most circulated intersection (C129) based on specifications from the traffic management centre. The simulation outputs from various scenario testing will be then used as inputs for the optimisation and comparative analysis modules. Secondly, we propose a multi-objective optimization method by using evolutionary algorithms and find the optimal traffic control plan to be used in C129 during morning and evening rush hours. Lastly, we take a more global view and extend the 3D simulation model to three other interconnected intersections, in order to analyse the impact of local optimisation on the surrounding traffic conditions in the eco-neighbourhood. The current proposed simulation-optimisation framework aims at supporting the traffic engineering decision-making process and the smart city dynamic by favouring a sustainable mobility.
The characteristics of an unknown source of emissions in the atmosphere are identified using an Adaptive evolutionary Strategy (AES) methodology based on ground concentration measurements and a Gaussian plume model. T...
详细信息
The characteristics of an unknown source of emissions in the atmosphere are identified using an Adaptive evolutionary Strategy (AES) methodology based on ground concentration measurements and a Gaussian plume model. The AES methodology selects an initial set of source characteristics including position, size, mass emission rate, and wind direction, from which a forward dispersion simulation is performed. The error between the simulated concentrations from the tentative source and the observed ground measurements is calculated. Then the AES algorithm prescribes the next tentative set of source characteristics. The iteration proceeds towards minimum error, corresponding to convergence towards the real source. The proposed methodology was used to identify the source characteristics of 12 releases from the Prairie Grass field experiment of dispersion, two for each atmospheric stability class, ranging from very unstable to stable atmosphere. The AES algorithm was found to have advantages over a simple canonical ES and a Monte Carlo (MC) method which were used as benchmarks. (C) 2010 Elsevier Ltd. All rights reserved.
The development of ultra-intense laser-based sources of high energy ions is an important goal, with a variety of potential applications. One of the barriers to achieving this goal is the need to maximize the conversio...
详细信息
The development of ultra-intense laser-based sources of high energy ions is an important goal, with a variety of potential applications. One of the barriers to achieving this goal is the need to maximize the conversion efficiency from laser energy to ion energy. We apply a new approach to this problem, in which we use an evolutionary algorithm to optimize conversion efficiency by exploring variations of the target density profile with thousands of one-dimensional particle-in-cell (PIC) simulations. We then compare this 'optimal' target identified by the one-dimensional PIC simulations to more conventional choices, such as with an exponential scale length pre-plasma, with fully three-dimensional PIC simulations. The optimal target outperforms the conventional targets in terms of maximum ion energy by 20% and show a noticeable enhancement of conversion efficiency to >2 MeV ions. This target geometry enhances laser coupling to the electrons, while still allowing the laser to strongly reflect from an effectively thin target. These results underscore the potential for this statistics-driven approach to guide research into optimizing laser-plasma simulations and experiments.
Convection selection is an approach to multipopulational evolutionary algorithms where solutions are assigned to subpopulations based on their fitness values. Although it is known that convection selection can allow t...
详细信息
Convection selection is an approach to multipopulational evolutionary algorithms where solutions are assigned to subpopulations based on their fitness values. Although it is known that convection selection can allow the algorithm to find better solutions than it would be possible with a standard single population, the convection approach was not yet compared to other, commonly used architectures of multipopulational evolutionary algorithms, such as the island model. In this paper we describe results of experiments which facilitate such a comparison, including extensive multi-parameter analyses. We show that approaches based on convection selection can obtain better results than the island model, especially for difficult optimization problems such as those existing in the area of evolutionary design. We also introduce and test a generalization of the convection selection which allows for adjustable overlapping of fitness ranges of subpopulations;the amount of overlapping influences the exploration vs. exploitation balance. (C) 2018 Elsevier B.V. All rights reserved.
暂无评论