scholarly journals Global Optimization of Redescending Robust Estimators

2021 ◽  
Vol 2021 ◽  
pp. 1-13
Author(s):  
Sergio Baselga ◽  
Ivandro Klein ◽  
Stefano Sampaio Suraci ◽  
Leonardo Castro de Oliveira ◽  
Marcelo Tomio Matsuoka ◽  
...  

Robust estimation has proved to be a valuable alternative to the least squares estimator for the cases where the dataset is contaminated with outliers. Many robust estimators have been designed to be minimally affected by the outlying observations and produce a good fit for the majority of the data. Among them, the redescending estimators have demonstrated the best estimation capabilities. It is little known, however, that the success of a robust estimation method depends not only on the robust estimator used but also on the way the estimator is computed. In the present paper, we show that for complicated cases, the predominant method of computing the robust estimator by means of an iteratively reweighted least squares scheme may result in a local optimum of significantly lower quality than the global optimum attainable by means of a global optimization method. Further, the sequential use of the proposed global robust estimation proves to successfully solve the problem of M-split estimation, that is, the determination of parameters of different functional models implicit in the data.

2019 ◽  
Vol 2019 ◽  
pp. 1-11
Author(s):  
Jun Qin ◽  
ChuTing Wang ◽  
GuiHe Qin

Multilevel thresholding is to find the thresholds to segment the image with grey levels. Usually, the thresholds are so determined that some indicator functions of the segmented image are optimized. To improve the computational efficiency, we presented an optimization method for multilevel thresholding. First, the solution space is divided into subspaces. Second, the subspaces are searched to obtain their current local optimal value. Third, the subspaces that are of worse current optimal value are eliminated. Then, the next round of elimination is exerted in the remainder. The elimination is repeated until only one subspace is left and its optimal value is taken as the global optimum. In principle, any random search algorithm can be used to find the local optimum in a subspace block because it is a strategy to enhance the searching efficiency through eliminating hopeless regions as early as possible, rather than to improve the searching algorithm itself. To verify its performance, taking PSO (Particle swarm optimization) as the basic searching algorithm of subspaces, the presented method is applied to Otsu’s and Kapur’s multilevel thresholding of four different kinds of digital images. The presented method is compared with PSO, and it behaves better in efficiency.


Author(s):  
J. Gu ◽  
G. Y. Li ◽  
Z. Dong

Metamodeling techniques are increasingly used in solving computation intensive design optimization problems today. In this work, the issue of automatic identification of appropriate metamodeling techniques in global optimization is addressed. A generic, new hybrid metamodel based global optimization method, particularly suitable for design problems involving computation intensive, black-box analyses and simulations, is introduced. The method employs three representative metamodels concurrently in the search process and selects sample data points adaptively according to the values calculated using the three metamodels to improve the accuracy of modeling. The global optimum is identified when the metamodels become reasonably accurate. The new method is tested using various benchmark global optimization problems and applied to a real industrial design optimization problem involving vehicle crash simulation, to demonstrate the superior performance of the new algorithm over existing search methods. Present limitations of the proposed method are also discussed.


Author(s):  
Liqun Wang ◽  
Songqing Shan ◽  
G. Gary Wang

The presence of black-box functions in engineering design, which are usually computation-intensive, demands efficient global optimization methods. This work proposes a new global optimization method for black-box functions. The global optimization method is based on a novel mode-pursuing sampling (MPS) method which systematically generates more sample points in the neighborhood of the function mode while statistically covers the entire search space. Quadratic regression is performed to detect the region containing the global optimum. The sampling and detection process iterates until the global optimum is obtained. Through intensive testing, this method is found to be effective, efficient, robust, and applicable to both continuous and discontinuous functions. It supports simultaneous computation and applies to both unconstrained and constrained optimization problems. Because it does not call any existing global optimization tool, it can be used as a standalone global optimization method for inexpensive problems as well. Limitation of the method is also identified and discussed.


Author(s):  
Alireza Saremi ◽  
Nasr Al-Hinai ◽  
G. Gary Wang ◽  
Tarek ElMekkawy

The current work discusses a novel global optimization method called the Multi-Agent Normal Sampling Technique (MANST). MANST is based on systematic sampling of points around agents; each agent in MANST represents a candidate solution of the problem. All agents compete with each other for a larger share of available resources. The performance of all agents is periodically evaluated and a specific number of agents who show no promising achievements are deleted; new agents are generated in the proximity of those promising agents. This process continues until the agents converge to the global optimum. MANST is a standalone global optimization technique. It is benchmarked with six well-known test cases and the results are then compared with those obtained from Matlab™ 7.1 GA Toolbox. The test results showed that MANST outperformed Matlab™ 7.1 GA Toolbox for the benchmark problems in terms of accuracy, number of function evaluations, and CPU time.


Author(s):  
Alireza Saremi ◽  
Amir H. Birjandi ◽  
G. Gary Wang ◽  
Tarek ElMekkawy ◽  
Eric Bibeau

This paper describes an enhanced version of a new global optimization method, Multi-Agent Normal Sampling Technique (MANST) described in reference [1]. Each agent in MANST includes a number of points that sample around the mean point with a certain standard deviation. In each step the point with the minimum value in the agent is chosen as the center point for the next step normal sampling. Then the chosen points of all agents are compared to each other and agents receive a certain share of the resources for the next step according to their lowest mean function value at the current step. The performance of all agents is periodically evaluated and a specific number of agents who show no promising achievements are deleted; new agents are generated in the proximity of those promising agents. This process continues until the agents converge to the global optimum. MANST is a standalone global optimization technique and does not require equations or knowledge about the objective function. The unique feature of this method in comparison with other global optimization methods is its dynamic normal distribution search. This work presents our recent research in enhancing MANST to handle variable boundaries and constraints. Moreover, a lean group sampling approach is implemented to prevent sampling in the same region for different agents. The overall capability and efficiency of the MANST has been improved as a result in the newer version. The enhanced MANST is highly competitive with other stochastic methods such as Genetic Algorithm (GA). In most of the test cases, the performance of the MANST is significantly higher than the Matlab™ GA Toolbox.


Author(s):  
WEN-LIANG HUNG ◽  
YUAN-CHEN LIU

The purpose of this paper is to find a robust estimation method for a two-parameter Weibull distribution when outliers are present. This is a relevant problem because of the usefulness of the Weibull distribution in life testing and reliability theory. For that purpose, a cluster-wise fuzzy least-squares algorithm with a noise cluster is used. This is because a noise cluster can be used for compensating the effects of outliers. Numerical comparisons between this fuzzy least-squares algorithm and the existing methods are implemented. According to these comparisons, it is suggested that the proposed fuzzy least-squares algorithm is preferable when the sample size is large.


1993 ◽  
Vol 115 (4) ◽  
pp. 770-775 ◽  
Author(s):  
P. Jain ◽  
A. M. Agogino

Multistart is a stochastic global optimization method for finding the global optimum of highly nonlinear mechanical problems. In this paper we introduce and develop a variant of the multistart method in which a fraction of the sample points in the feasible region with smallest function value are clustered using the Vector Quantization technique. The theories of lattices and sphere packing are used to define optimal lattices. These lattices are optimal with respect to quantization error and are used as code points for vector quantization. The implementation of these ideas has resulted in the VQ-multistart algorithm for finding the global optimum with substantial reductions in both the incore memory requirements and the computation time. We solve several mathematical test problems and a mechanical optimal design problem using the VQ-multistart algorithm.


Author(s):  
С.И. Носков

Описываются свойства методов оценивания параметров регрессионных моделей - наименьших квадратов, модулей, антиробастного, а также их применения для решения конкретных практических проблем. При этом метод наименьших модулей не реагирует на аномальные наблюдения выборки, метод антиробастного оценивания сильно отклоняет линию регрессии в их направлении, метод наименьших квадратов занимает промежуточное положение. Показано, что если целью построения модели является проведение на ее основе многовариантных прогнозных расчетов значений зависимой переменной, то выбор метода численной идентификации параметров модели следует производить на основе анализа характера выбросов. Если есть основания полагать, что подобные им ситуации могут иметь место в будущем, следует выбрать метод антиробастного оценивания, в противном же случае - метод наименьших модулей. Построена регрессионная модель грузооборота Красноярской железной дороги на основе применения всех трех методов оценивания параметров. Проведен анализ причин, имеющих место в 2010 году в ситуации резкого падения величины грузооборота, которая вполне может характеризоваться как аномальное наблюдение в данных. Сделаны рекомендации по выбору метода оценивания параметров в этом случае The article describes the properties of methods for estimating the parameters of regression models - least squares, moduli, anti-robust - as well as their application for solving specific practical problems. At the same time, the method of least modules does not respond to anomalous observations of the sample, the method of anti-robust estimation strongly deviates the regression line in their direction, the method of least squares occupies an intermediate position. I show that if the purpose of constructing a model is to carry out multivariate predictive calculations of the values of the dependent variable on its basis, then the choice of a method for the numerical identification of model parameters should be based on an analysis of the nature of emissions. If there is a reason to believe that similar situations may occur in the future, the anti-robust estimation method should be chosen, otherwise - the least modulus method. I built a regression model of the freight turnover of the Krasnoyarsk railway on the basis of the application of all three methods of parameter estimation. I carried out the analysis of the reasons for the situation of a sharp drop in the value of cargo turnover in 2010, which may well be characterized as anomalous observation in the data. I give recommendations on the choice of the parameter estimation method in this case


2015 ◽  
Vol 2015 ◽  
pp. 1-14 ◽  
Author(s):  
Jian-Guo Zheng ◽  
Chao-Qun Zhang ◽  
Yong-Quan Zhou

Artificial bee colony (ABC) algorithm is a popular swarm intelligence technique inspired by the intelligent foraging behavior of honey bees. However, ABC is good at exploration but poor at exploitation and its convergence speed is also an issue in some cases. To improve the performance of ABC, a novel ABC combined with grenade explosion method (GEM) and Cauchy operator, namely, ABCGC, is proposed. GEM is embedded in the onlooker bees’ phase to enhance the exploitation ability and accelerate convergence of ABCGC; meanwhile, Cauchy operator is introduced into the scout bees’ phase to help ABCGC escape from local optimum and further enhance its exploration ability. Two sets of well-known benchmark functions are used to validate the better performance of ABCGC. The experiments confirm that ABCGC is significantly superior to ABC and other competitors; particularly it converges to the global optimum faster in most cases. These results suggest that ABCGC usually achieves a good balance between exploitation and exploration and can effectively serve as an alternative for global optimization.


Author(s):  
P. Jain ◽  
A. M. Agogino

Abstract Multistart is a novel stochastic global optimization method for finding the global optimum of highly nonlinear mechanical problems. In this paper we introduce and develop a variant of the multistart method in which a fraction of the sample points in the feasible region with smallest function value are clustered using the Vector Quantization technique. The theories of lattices and sphere packing are used to define optimal lattices. These lattices are optimal with respect to quantization error and are used as code points for vector quantization. The implementation of these ideas has resulted in the VQ-multistart algorithm for finding the global optimum with substantial reductions in both the incore memory requirements and the computation time. We solve several mathematical test problems and a mechanical optimal design problem using the VQ-multistart algorithm.


Sign in / Sign up

Export Citation Format

Share Document