An improved Least Squares Support Vector Machine (LS-SVM) approach was proposed to overcome the drawback of "losing sparsity" in original LS-SVM. At the same time, real-coded genetic algorithm (RC-GA) was in...
详细信息
ISBN:
(纸本)9781424421138
An improved Least Squares Support Vector Machine (LS-SVM) approach was proposed to overcome the drawback of "losing sparsity" in original LS-SVM. At the same time, real-coded genetic algorithm (RC-GA) was introduced to solve the difficult problem of parameters selection in LS-SVM. By discarding most data points with too large or too small training errors in the trained LS-SVM model, a more sparse and anti-noise LS-SVM model was obtained. The parameters selection of LS-SVM was regard as an optimization problem. The objective function of the optimization problem was established. And a RC-GA with high global searching capability was used to search the optimal parameters of LS-SVM. Both simulation and experiment results demonstrate the success of the improved LS-SVM and the effectiveness of RC-GA parameters optimization method.
We proposed Evolutionary Particle Swarm Optimization (EPSO) which provides a new paradigm of meta-optimization for model selection in swarm intelligence. In this paper, we extend the technique of online evolutionary c...
详细信息
ISBN:
(纸本)9783540875352
We proposed Evolutionary Particle Swarm Optimization (EPSO) which provides a new paradigm of meta-optimization for model selection in swarm intelligence. In this paper, we extend the technique of online evolutionary computation of EPSO to Canonical Particle Swarm Optimizer (CPSO), and propose Evolutionary Canonical Particle Swarm Optimizer (ECPSO) for optimizing CPSO. In order to effectually evalute the performance of CPSO, a temporally cumulative fitness function of the best particle is adoped in ECPSO as the behavioral representative for entire swarm. Applications of the proposed method to a suite of 5-dimensional benchmark problems well demonstrate the effectiveness. Our experimental results clearly indicate that (1) the proper parameter sets in CPSO for solving various optimization problems are not unique;(2) the values of parameters in them are quite different from that of the original CPSO;(3) the search performance of the optimized CPSO is superior to that of the original CPSO, and to that of RGA/E except for the result of the Rastrigin's benchmark problem.
Parameters of real-valued crossover operators ha,,e been of ten tuned under a constraint for preserving statistics of infinite parental population. For applications in actual scenes, in a previous study, all alternati...
详细信息
ISBN:
(纸本)9783540896937
Parameters of real-valued crossover operators ha,,e been of ten tuned under a constraint for preserving statistics of infinite parental population. For applications in actual scenes, in a previous study, all alternative constraint, called unbiased constraint, considering finiteness of the population has been derived. To clarify the wide applicability of the unbiased constraint, this paper presents two additional Studies: (1) applying it to various crossover operators in higher dimensional search space, and (2) generalization of it, for preserving statistics of overall population. Appropriateness of the parameter setting based oil the unbiased constraint has been supported in discussion on robust search.
The conventional ordinary least squares (OLS) variance-covariance matrix estimator for a linear regression model under heteroscedastic errors is biased and inconsistent. Accordingly, several estimators have so far bee...
详细信息
The conventional ordinary least squares (OLS) variance-covariance matrix estimator for a linear regression model under heteroscedastic errors is biased and inconsistent. Accordingly, several estimators have so far been proposed by various researchers. However, none of these perform well under the finite-sample situation. In this paper, the powerful optimization technique of geneticalgorithm (GA) is used to modify these estimators. Properties of these newly developed estimators are thoroughly studied by Monte Carlo method for various sample sizes. It is shown that GA-versions of the estimators are superior to corresponding non-GA versions as there are significant reductions in the Total relative bias as well as Total root mean square error.
The methodology of developing Fuzzy Cognitive Map (FCM) still exhibited weaknesses. This paper investigated a hybrid framework for learning FCM, which was combined of the real-codedgenetic (RCGA) algorithm and nonlin...
详细信息
ISBN:
(纸本)9781424421077
The methodology of developing Fuzzy Cognitive Map (FCM) still exhibited weaknesses. This paper investigated a hybrid framework for learning FCM, which was combined of the real-codedgenetic (RCGA) algorithm and nonlinear Hebbian learning (NHL) algorithm. This approach combined the synergistic theories of neural networks and fuzzy logic. The hybrid algorithm is introduced, presented and applied successfully in real-world problems of partner selection.
Water quality models are helpful for rationalizing water quality management. Their success depends largely on how well they are calibrated. The automatic calibration of models usually outperforms traditional trial-and...
详细信息
Water quality models are helpful for rationalizing water quality management. Their success depends largely on how well they are calibrated. The automatic calibration of models usually outperforms traditional trial-and-error process. In this study, a realcodedgeneticalgorithm (GA) was combined with the Nelder-Mead simplex (NMS) algorithm to form a hybrid approach, GA-NMS. It was employed to calibrate simulated vertical profiles of temperature and concentration of chlorophyll a by CE-QUAL-W2 to measured values in Lake Maumelle, USA. A set of parameter values that had the lowest objective function value was obtained in hundreds of objective function evaluations. It produces reasonable agreement between measurements and simulations. This application demonstrated that the approach can be used in the automatic calibration of water quality models.
An improved Least Squares Support Vector Machine (LS-SVM) approach was proposed to overcome the drawback of "losing sparsity" in original LS-SVM. At the same time, real-coded genetic algorithm (RC-GA) was in...
详细信息
An improved Least Squares Support Vector Machine (LS-SVM) approach was proposed to overcome the drawback of "losing sparsity" in original LS-SVM. At the same time, real-coded genetic algorithm (RC-GA) was introduced to solve the difficult problem of parameters selection in LS-SVM. By discarding most data points with too large or too small training errors in the trained LS-SVM model, a more sparse and anti-noise LS-SVM model was obtained. The parameters selection of LS-SVM was regard as an optimization problem. The objective function of the optimization problem was established. And a RC-GA with high global searching capability was used to search the optimal parameters of LS-SVM. Both simulation and experiment results demonstrate the success of the improved LS-SVM and the effectiveness of RC-GA parameters optimization method.
Many real-world design problems involve simultaneous optimization of multiple objectives. The considered grillage design is a multi-objective optimization problem, since there are two objective functions, namely, the ...
详细信息
Many real-world design problems involve simultaneous optimization of multiple objectives. The considered grillage design is a multi-objective optimization problem, since there are two objective functions, namely, the volume design and the cost design. Thus, the solution of this optimization problem requires specialized method suitable for multi-objective problems. In this article, a multi-objective design optimization using real-coded genetic algorithm which is proposed for the Pareto-optimality of grillage system. The objective functions in the optimization problem measure the design sensitivities in grillage system. The nonlinear constrained multi-objective optimization is a very important from the point of view of practical problem solving. Therefore, the real-coded genetic algorithm with multiple genetic operators is proposed to find the optimum grillage system without handling any of penalty functions. Direct strength calculation defined from the class rules of DNV was applied for structural design of grillage system. The hybrid method (real-coded genetic algorithm including the non-dominated sorting and sharing approaches) performs a marvelous explorability in finding a diverse set of solutions and in converging near the true Pareto-optimal set. The results obtained are very encouraging, since they show that we can produce an important portion of the Pareto-front at a very low computational time frame.
In this paper, we will propose a modified crossover formula in geneticalgorithms (GAs), and use this method to determine PID controller gains for multivariable processes. It is well known that GA is globally optimal ...
详细信息
In this paper, we will propose a modified crossover formula in geneticalgorithms (GAs), and use this method to determine PID controller gains for multivariable processes. It is well known that GA is globally optimal search method borrowing the concepts from biological evolutionary theory. In the traditional crossover fashion, only two parent chromosomes are usually used to be crossed by each other. The proposed algorithm, however, is designed to provide a more accurate adjusting direction for evolving offspring because of the use of multi-crossover genetic operations. Then we apply the innovative GA into the design of multivariable PID control systems for deriving optimal or near optimal control gains such that the defined performance criterion of integrated absolute error (IAE) is minimized as much as possible. Finally, a 2 x 2 multivariable controlled plant with strong interactions between input and output pairs will be illustrated to demonstrate the effectiveness of the proposed method. Some comparison results with the traditional GA and BLT method are also demonstrated in the simulation. (c) 2006 Elsevier Ltd. All rights reserved.
Parameter identification problem will be presented, and solved through our new real-coded genetic algorithm. The algorithm is a modified version from normal GA but it includes biased initialization, dynamic parameters...
详细信息
Parameter identification problem will be presented, and solved through our new real-coded genetic algorithm. The algorithm is a modified version from normal GA but it includes biased initialization, dynamic parameters, and elitism. The algorithm will be tested on three cases. (c) 2006 Elsevier Inc. All rights reserved.
暂无评论