• Journal of the Korean Statistical Society
• /
• v.29 no.4
• /
• pp.407-422
• /
• 2000

This article is concerned with the selecting predictor variables to be included in building a class of binary response t-link regression models where both probit and logistic regression models can e approximately taken as members of the class. It is based on a modification of the stochastic search variable selection method(SSVS), intended to propose and develop a Bayesian procedure that used probabilistic considerations for selecting promising subsets of predictor variables. The procedure reformulates the binary response t-link regression setup in a hierarchical truncated normal mixture model by introducing a set of hyperparameters that will be used to identify subset choices. In this setup, the most promising subset of predictors can be identified as that with highest posterior probability in the marginal posterior distribution of the hyperparameters. To highlight the merit of the procedure, an illustrative numerical example is given.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.03a
• /
• pp.81-84
• /
• 1998

Genetic Algorithms (GAs) are stochastic algorithms whose search methods model some natural phenomena. The procedure of GAs may be divided into two sub-procedures : Operation and Selection. Chromosomes can produce new offspring by means of operation, and the fitter chromosomes can produce more offspring than the less fit ones by means of selection. However, operation which is executed randomly and has some limits to its execution can not guarantee to produce fitter chromosomes. Thus, we propose a method which gives a directional information to the genetic operator by reinforcement learning. It can be achived by using neural networks to apply reinforcement learning to the genetic operator. We use the amount of fitness change which can be considered as reinforcement signal to calcualte the error terms for the output units. Then the weights are updated using backpropagtion algorithm. The performance improvement of GAs using reinforcement learning can be measured by applying the pr posed method to GA-hard problem.

• Food Science and Biotechnology
• /
• v.17 no.6
• /
• pp.1352-1356
• /
• 2008

Bootstrap method, a computer-intensive statistical technique to estimate the distribution of a statistic was applied to deal with uncertainty and variability of the experimental data in stochastic prediction modeling of microbial growth on a chill-stored food. Three different bootstrapping methods for the curve-fitting to the microbial count data were compared in determining the parameters of Baranyi and Roberts growth model: nonlinear regression to static version function with resampling residuals onto all the experimental microbial count data; static version regression onto mean counts at sampling times; dynamic version fitting of differential equations onto the bootstrapped mean counts. All the methods outputted almost same mean values of the parameters with difference in their distribution. Parameter search according to the dynamic form of differential equations resulted in the largest distribution of the model parameters but produced the confidence interval of the predicted microbial count close to those of nonlinear regression of static equation.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.34 no.3
• /
• pp.821-831
• /
• 2014

Stochastic rainfall generators or stochastic simulation have been widely employed to generate synthetic rainfall sequences which can be used in hydrologic models as inputs. The calibration of Poisson cluster stochastic rainfall generator (e.g. Modified Bartlett-Lewis Rectangular Pulse, MBLRP) is seriously affected by local minima that is usually estimated from the local optimization algorithm. In this regard, global optimization techniques such as particle swarm optimization and shuffled complex evolution algorithm have been proposed to better estimate the parameters. Although the global search algorithm is designed to avoid the local minima, reliable parameter estimation of MBLRP model is not always feasible especially in a limited parameter space. In addition, uncertainty associated with parameters in the MBLRP rainfall generator has not been properly addressed yet. In this sense, this study aims to develop and test a Bayesian model based parameter estimation method for the MBLRP rainfall generator that allow us to derive the posterior distribution of the model parameters. It was found that the HBM based MBLRP model showed better performance in terms of reproducing rainfall statistic and underlying distribution of hourly rainfall series.

• Journal of the Korean Society for Precision Engineering
• /
• v.14 no.3
• /
• pp.89-97
• /
• 1997

A genetic algorithm (GA) is a stochastic direct search strategy that mimics the process of genetic evolution. The GA applied herein works on a population of structural designs at any one time, and uses a structured information exchange based on the principles of natural selection and wurvival of the fittest to recombine the most desirable features of the designs over a sequence of generations until the process converges to a "maximum fitness" design. Principles of genetics are adapted into a search procedure for structural optimization. The methods consist of three genetics operations mainly named selection, cross- over and mutation. In this study, a method of finding the optimum topology of truss/beam structure is pro- posed by using the GA. In order to use GA in the optimum topology problem, chromosomes to FEM elements are assigned, and a penalty function is used to include constraints into fitness function. The results show that the GA has the potential to be an effective tool for the optimal design of structures accounting for sizing, geometrical and topological variables.variables.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.15 no.4
• /
• pp.591-598
• /
• 2002

In the ship structural design, the material cost of hull weight and the overall cost of construction processes should be minimized considering safety and reliability. In the past, minimum weight design has been mainly focused on reducing material cost and increasing dead weight reflect the interests of a ship's owner. But, in the past experience, the minimum weight design has been inevitably lead to increasing the construction cost. Therefore, it is necessary that the designer of ship structure should consider both structural weight and construction cost. In this point of view, multi-objective optimization technique is proposed to design the ship structure in this study. According to the proposed algorithm, the results of optimization were compared to the structural design of actual VLCC(Very Large Crude Oil Carrier). Objective functions were weight cost and construction cost of VLCC, and ES(Evolution Strategies), one of the stochastic search methods, was used as an optimization solver. For the scantlings of members and the estimations of objectives, classification rule was adopted for the longitudinal members, and the direct calculation method, GSDM(Generalized Slope Deflection Method), lot the transverse members. To choose the most economical design point among the results of Pareto optimal set, RFR(Required Freight Rate) was evaluated for each Pareto point, and compared to actual ship.

• Management & Information Systems Review
• /
• v.2
• /
• pp.75-93
• /
• 1998

In this paper, we propose a new algorithm(N_BP) to be capable of overcoming limitations of the traditional backpropagation(O_BP). The N_BP is based on the method of conjugate gradients and calculates learning parameters through the line search which may be characterized by order statistics and golden section. Experimental results showed that the N_BP was definitely superior to the O_BP with and without a stochastic term in terms of accuracy and rate of convergence and might surmount the problem of local minima. Furthermore, they confirmed us that the stagnant phenomenon of learning in the O_BP resulted from the limitations of its algorithm in itself and that unessential approaches would never cured it of this phenomenon.

• Proceedings of the Korea Committee for Ocean Resources and Engineering Conference
• /
• 2000.04a
• /
• pp.96-103
• /
• 2000

This research is a new design method, which will be presented as a basic concept for a more efficient minimum weight design of grillages, as an attempt to describe true collapse mechanism in as overall search as possible. It serves as introduction to the numerical technique of Linear Programming(LP) and Automatic Modified Direct Plastic Frame Analysis(AMDPFA). Attention is directed to both analysis and design, and emphasis is placed on the physical significance of Systematic Searching Techniques(SST) involved. In weight minimum grillages design, the parameterisation study in optimum beam configuration which was carried out over the range of beam sections for a given plastic section modulus likely to occur in structures by suing an adaptive stochastic optimisation technique, Genetic Algorithms.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.12
• /
• pp.1680-1686
• /
• 1995

GAs, effective stochastic search algorithms based on the model of natural evolution and genetics, have been successfully applied to various optimization problems. When population size is not large, GAs often suffer from the phenomenon of premature convergence in which all chromosomes in the population lose the diversity of genes before they find the optimal solution. In this paper, we propose that a new heuristic that maintains the diversity of genes by adding some chromosomes with random mutation and selective mutation into population during evolution. And population size changes dynamically with supplement of new chromosomes. Experimental results for several test functions show that when population size is rather small and the length of chromosome is not long, this method is effective.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 1997.10a
• /
• pp.163-170
• /
• 1997

The present paper describes an investigation into the application of the genetic algorithm(GA) in the optimization of structural design. Stochastic processes generate an initial population of designs and then apply principles of natural selection/survival of the fittest to improve the designs. The five test functions are used to verify the robustness and reliability of GA, and as a numerical example, minimum weight of a cantilever composite laminated beam with a mix of continuous, integer and discrete design variables is obtained by using GA with exterior penalty function method. The design problem has constraints on strength, displacements, and natural frequencies, and is formulated to a multidimensional nonlinear form. From the results, it is found that the GA search technique is very effective at finding the good optimum solution as well as has higher robustness.