• Transactions of the Korean Society of Mechanical Engineers
• /
• v.12 no.1
• /
• pp.137-151
• /
• 1988

Thurbulent thermal convection between two plates, bottom plate is at higher temperature $T_{h}$ and the upper plate is at lower temperature $T_{i}$ is numerically investigated. Model equations are abridged Reynolds stress equations; full Reynolds stress equations are simplified to yield algebraic relations in case of mean square velocity fluctuations in vertical and horizontal directions. Boundary conditions for turbulent kinetic energy k and mean square temperature variance .thera.$^{2}$oner bar at the plate surfaces are set to be zero and those of dissipation rate of turbulent kinetic energy .epsilon. and dissipation rate of mean square temperature variance .epsilon.$_{\theta}$ are assumed at first grid point nearest to the boundary surfaces, whose values are approximated by inviscid estimates. Results show that temperature profiles are in good agreement with experimental data except transition region, in which temperature is over-predicted. Such discrepancy becomes larger as the Rayleigh number becomes smaller. Nusselt numbers, which are calculated from the temperature gradients at the boundary surfaces, are also in good agreement with experimental data.a.a.

• Asian-Australasian Journal of Animal Sciences
• /
• v.21 no.4
• /
• pp.478-483
• /
• 2008

Records of a Northeastern Thai indigenous cattle line population were used to estimate genetic parameters and annual trends for calf weights. The data set comprised records of 1,922 and 1,489 animals for birth and weaning weight, respectively born from 1993 to 2004. A bivariate analysis was carried out for variance and covariance components estimations using average information restricted maximum likelihood procedure. Average estimated breeding value and maternal breeding value of the animals born in 1993 were set to zero as a base group. Genetic trends of each trait were calculated by regressing average estimated breeding values and maternal breeding values on birth year of calves. Phenotypic trends for each trait were calculated by regressing the yearly adjusted weight on birth year of calves. The results revealed that the estimate of direct heritability, maternal heritability and maternal permanent environmental variance as a proportion of phenotypic variance for birth and weaning weight was 0.40, 0.14 and 0.04; 0.27, 0.05 and 0.23, respectively. Direct heritability was moderately heritable and genetic improvement through selection can be achieved. The estimate of phenotypic, direct genetic, maternal genetic and maternal permanent environmental correlation between birth and weaning weight was 0.48, 0.65, 0.98 and 0.73, respectively. The phenotypic trend, genetic trends of estimated breeding value and maternal breeding value for birth weight was 0.18, 0.04 and 0.01 kg/year, respectively. The phenotypic trend, genetic trends of estimated breeding value and maternal breeding value for weaning weight was -1.36, 0.32 and 0.03 kg/year, respectively. As maternal genetic effect was considerably less important than direct genetic effect, selection for improved weaning weight of this Northeastern Thai indigenous cattle line can place more emphasis on the direct genetic effect.

• Animal Bioscience
• /
• v.34 no.1
• /
• pp.12-19
• /
• 2021

Objective: The objectives of the current study were to study the main environmental factors affecting birth weight (BW) and weaning weight (WW), estimate variance components, genetic parameters and genetic trend and to evaluate the variability and relationships among breeding value of BW and WW using principal components analysis (PCA). Methods: A total of 16,370 records were collected from 8,271 buffalo calves. Genetic parameters and breeding values were estimated using a bivariate animal model which includes direct, maternal and permanent maternal effects. These estimates were standardized and used in PCA. Results: The direct heritability estimates were 0.06 and 0.41 for BW and WW, respectively whereas direct maternal heritability values were 0.03 and 0.14, respectively. Proportions of variance due to permanent environmental effects of dam were 0.455 and 0.280 for BW and WW respectively. The genetic correlation between BW and WWs was weak approaching zero, but the maternal correlation was 0.26. The first two principal components (PC1 and PC2) were estimated utilizing the standardized breeding values according to Kaiser method. The total variance explained by the first two PCs was 71.17% in which 45.91% and 25.25% were explained by PC1 and PC2, respectively. The direct breeding values of BW were related to PC2 but those of WW and maternal breeding values of BW and WWs were associated with PC1. Conclusion: The results of genetic parameters and PCA indicate that BW and WWs were not genetically correlated and improving growth traits of Egyptian buffaloes could be achieved using WW without any adverse effect by BW.

• The Korean Journal of Applied Statistics
• /
• v.32 no.5
• /
• pp.693-702
• /
• 2019

Deep learning has gained popularity for the classification and prediction task. Neural network layers become deeper as more data becomes available. Saturation is the phenomenon that the gradient of an activation function gets closer to 0 and can happen when the value of weight is too big. Increased importance has been placed on the issue of saturation which limits the ability of weight to learn. To resolve this problem, Glorot and Bengio (Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, 249-256, 2010) claimed that efficient neural network training is possible when data flows variously between layers. They argued that variance over the output of each layer and variance over input of each layer are equal. They proposed a method of initialization that the variance of the output of each layer and the variance of the input should be the same. In this paper, we propose a new method of establishing initialization by adopting truncated normal distribution and truncated cauchy distribution. We decide where to truncate the distribution while adapting the initialization method by Glorot and Bengio (2010). Variances are made over output and input equal that are then accomplished by setting variances equal to the variance of truncated distribution. It manipulates the distribution so that the initial values of weights would not grow so large and with values that simultaneously get close to zero. To compare the performance of our proposed method with existing methods, we conducted experiments on MNIST and CIFAR-10 data using DNN and CNN. Our proposed method outperformed existing methods in terms of accuracy.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.747-750
• /
• 2001

The paper presents an algorithm for estimating the variance of additive zero mean Gaussian noise in an image. The algorithm uses the wavelet transform which is a good tool for energy compaction. The algorithm consists of three steps. At first, high frequency components, wavelet coefficients in HH band, are generated from a noisy image by the wavelet transform. In a second step, high frequency components which are out of the noise range ate eliminated. Finally, if the image has many components eliminated in the previous step, then its noise estimated value is reduced. Experimental results show that the wavelet filter has better performance than the other high pass filters such as a Laplacian filter, residual from a median filter, residual from a mean filter, and a difference operator. In various images, the algorithm reduces 50% of estimated error on an average.

• Bulletin of the Korean Mathematical Society
• /
• v.48 no.6
• /
• pp.1137-1146
• /
• 2011

Let {$X_i$, $i{\geq}1$} be a sequence of i.i.d. nondegenerate random variables which is in the domain of attraction of the normal law with mean zero and possibly infinite variance. Denote $S_n={\sum}_{i=1}^n\;X_i$, $M_n=max_{1{\leq}i{\leq}n}\;{\mid}S_i{\mid}$ and $V_n^2={\sum}_{i=1}^n\;X_i^2$. Then for d > -1, we showed that under some regularity conditions, $$\lim_{{\varepsilon}{\searrow}0}{\varepsilon}^2^{d+1}\sum_{n=1}^{\infty}\frac{(loglogn)^d}{nlogn}I\{M_n/V_n{\geq}\sqrt{2loglogn}({\varepsilon}+{\alpha}_n)\}=\frac{2}{\sqrt{\pi}(1+d)}{\Gamma}(d+3/2)\sum_{k=0}^{\infty}\frac{(-1)^k}{(2k+1)^{2d+2}}\;a.s.$$ holds in this paper, where If g denotes the indicator function.

• Journal of the Korean Mathematical Society
• /
• v.45 no.6
• /
• pp.1601-1611
• /
• 2008

Let X, $X_1$, $X_2$, ... be i.i.d. random variables with zero means, variance one, and set $S_n\;=\;{\sum}^n_{i=1}\;X_i$, $n\;{\geq}\;1$. Gut and $Sp{\check{a}}taru$ [3] established the precise asymptotics in the law of the iterated logarithm and Li, Nguyen and Rosalsky [7] generalized their result under minimal conditions. If P($|S_n|\;{\geq}\;{\varepsilon}{\sqrt{2n\;{\log}\;{\log}\;n}}$) is replaced by E{$|S_n|/{\sqrt{n}}-{\varepsilon}{\sqrt{2\;{\log}\;{\log}\;n}$}+ in their results, the new one is called the moment version of precise asymptotics in the law of the iterated logarithm. We establish such a result for self-normalized sums, when X belongs to the domain of attraction of the normal law.

• International Journal of Reliability and Applications
• /
• v.17 no.1
• /
• pp.85-105
• /
• 2016

In this paper accelerated life testing is incorporated in quality control technique of acceptance sampling plan to induce early failures in high reliability products.Stress under accelerated condition can be applied in constant-stress, step-stress and progressive-stress or combination of such loadings. A ramp-stress results when stress is increased linearly (from zero) with time. In this paper optimum failure-censored ramp-stress accelerated life test sampling plan for log-logistic distribution has been formulated with cost considerations. The log-logistic distribution has been found appropriate for insulating materials. The optimal plans consist in finding optimum sample size, sample proportion allocated to each stress, and stress rate factor such that producer's and consumer's interests are safeguarded. Variance optimality criterion is used when expected cost per lot is not taken into consideration, and bilevel programming approach is used in cost optimization problems. The methods developed have been illustrated using some numerical examples, and sensitivity analyses carried out in the context of ramp-stress ALTSP based on variable SSP for proportion nonconforming.

• Journal of Korean Society for Quality Management
• /
• v.22 no.2
• /
• pp.89-97
• /
• 1994

We consider the linear calibration model: $y_1={\alpha}+{\beta}x_i+{\sigma}{\varepsilon}_i$, i = 1, ${\cdots}$, n, $y={\alpha}+{\beta}x+{\sigma}{\varepsilon}$ where ($y_1$, ${\cdots}$, $y_n$, y) stands for an observation vector, {$x_i$} fixed design vector, (${\alpha}$, ${\beta}$) vector of regression parameters, x unknown true value of interest and {${\varepsilon}_i$}, ${\varepsilon}$ are mutually uncorrelated measurement errors with zero mean and unit variance but otherwise unknown distributions. On the basis of simple small-sample low-noise approximation, we introduce a new method of comparing the mean squared errors of the various competing estimators of the true value x for finite sample size n. Then we show that a class of estimators including the classical and the inverse estimators are consistent and first-order efficient within the class of all regular consistent estimators irrespective of type of measurement errors.

• The Korean Journal of Applied Statistics
• /
• v.17 no.3
• /
• pp.547-556
• /
• 2004

This paper consider the testing problem of variance component for the unbalanced two-way error component model with nuisance parameter. We derive the one-sided LM test statistic for testing zero individual(time) effects assuming that the other time-specific(individual) effects are present. Using the Monte Carlo experiments, the computational more demanding LR test slightly underestimates the nominal size and has the low powers relative to LM test statistic.