• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.315-315
• /
• 2000

In this paper, we perform the time series prediction using the SVM(Support Vector Machine). We make use of two different loss functions and two different kernel functions; i) Quadratic and $\varepsilon$-insensitive loss function are used; ii) GRBF(Gaussian Radial Basis Function) and ERBF(Exponential Radial Basis Function) are used. Mackey-Glass time series are used for prediction. For both cases, we compare the results by the SVM to those by ANN(Artificial Neural Network) and show the better performance by SVM than that by ANN.

• Communications for Statistical Applications and Methods
• /
• v.19 no.5
• /
• pp.655-662
• /
• 2012

Classification is an important research field in pattern recognition with high-dimensional predictors. The support vector machine(SVM) is a penalized feature selector and classifier. It is based on the hinge loss function, the non-convex penalty function, and the smoothly clipped absolute deviation(SCAD) suggested by Fan and Li (2001). We developed the algorithm for the multiclass SVM with the SCAD penalty function using the local quadratic approximation. For multiclass problems we compared the performance of the SVM with the $L_1$, $L_2$ penalty functions and the developed method.

• Journal of the Korean Statistical Society
• /
• v.22 no.1
• /
• pp.55-69
• /
• 1993

Consider the problem of estimating an arbitrary continuous vector function under a weighted quadratic loss in the multiparameter exponential family with the density of the natural form. We first provide, using Blyth's (1951) method, a set of sufficient conditions for the admisibility of (possibly generalized Bayes) estimators and then treat some examples for normal, Poisson, and gamma distributions as applications of the main result.

• Journal of Integrative Natural Science
• /
• v.14 no.3
• /
• pp.73-77
• /
• 2021

Robust design is an approach to reduce the performance variation of mutiple responses in products and processes. In fact, in many experimental designs require the simultaneous optimization of multiple responses. In this paper, we propose how to simultaneously optimize multiple responses for robust design when data are collected from a combined array. The proposed method is based on the quadratic loss function. An example is illustrated to show the proposed method.

• Journal of Korean Society for Quality Management
• /
• v.31 no.1
• /
• pp.109-122
• /
• 2003

This paper presents a robust process capability index(PCI) based on the expected loss derived from the empirical distribution function(EDF). We propose the EDF expected loss in order to develop a PCI that does not depends on the underlying process distribution. The EDF expected loss depends only on the sample data, so the PCI based on it is robust and it does nor require complex calculations. The inverted normal loss function(INLF) is employed in order to overcome the drawback of the quadratic loss which may Increase unboundedly outside the specification limits. A comprehensive simulation study was performed under various process distributions, in order to compare the accuracy and the precision of the proposed PCI with those of the PCI based on the expected loss derived from the normal distribution. The proposed PCI turned out to be more accurate than the normal PCI in most cases, especially when the process distribution has high kurtosis or skewness. It is expected that the proposed PCI can be utilized In real processes where the true distribution family may not be known.

• Proceedings of the KOSOMBE Conference
• /
• v.1995 no.11
• /
• pp.167-170
• /
• 1995

An frequency modulated (FM) DANTE pulse sequence generates a quadratic phase toward the transverse of image by an FM RF pulse. In the image of a serious susceptibility effect, the phase due to the difference of the susceptibility in the pixel occurs susceptibility error which arise signal loss. But the signal loss due to the susceptibility effect in the pixel is reduced when the quadratic phase adds in the pixel. In this paper, we have generated a quadratic function toward the transverse (X-Y) using FM DANTE sequence and the susceptibility effect is reduced in the gradient echo (GE) imaging. Computer simulation and experimental results is obtained by using a whole-body KAIS 2.0T NMR system.

• Journal of Korean Society for Quality Management
• /
• v.28 no.4
• /
• pp.194-203
• /
• 2000

This paper is concerned with an economic selection of both the process mean and the lower limit for a continuous production process with the quadratic loss function. It is assumed that the quality characteristic is normally distributed with a known variability. A profit model is developed which involves selling price, production cost, reprocessing cost and the cost which is incurred by imperfect quality. Methods of finding optimum values of the process mean and the lower limit are presented, and a numerical example is given.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.18 no.36
• /
• pp.275-285
• /
• 1995

In recent years there has been increasing interest in the issue of process centering in manufacturing process, The traditional process capability indices Cp, Cpk and Cpu are used to provide measure of process performance, but these indices do not represent the issue of process centering. A new measure of the process capability index Cpm is proposed that takes into account the proximity to the target value as well as the process variation when assessing process performance. However, Cpm only considers acceptance cost for deviation from target value within specification limits, do not includes economic consideration for rejected items. This paper proposes flexible process performance measures that considered quadratic loss function caused by quality deviation within specification limits, rejection cost associated with the disposition of rejected items, and inspection cost. In this model disposition of rejected items are considered under perfect corrective procedures and the absence of perfect corrective procedures.

• Communications for Statistical Applications and Methods
• /
• v.20 no.6
• /
• pp.481-490
• /
• 2013

Classification is an important research area as data can be easily obtained even if the number of predictors becomes huge. The support vector machine(SVM) is widely used to classify a subject into a predetermined group because it gives sound theoretical background and better performance than other methods in many applications. The SVM can be viewed as a penalized method with the hinge loss function and penalty functions. Instead of $L_2$ penalty function Fan and Li (2001) proposed the smoothly clipped absolute deviation(SCAD) satisfying good statistical properties. Despite the ability of SVMs, they have drawbacks of non-robustness when there are outliers in the data. We develop a robust SVM method using a weight function with the SCAD penalty function based on the local quadratic approximation. We compare the performance of the proposed SVM with the SVM using the $L_1$ and $L_2$ penalty functions.

• Journal of the Korean Data and Information Science Society
• /
• v.16 no.4
• /
• pp.1087-1094
• /
• 2005

e-insensitive support vector regression(e-SVR) is capable of providing more complete description of the linear and nonlinear relationships among random variables. In this paper we propose an iterative reweighted least squares(IRWLS) procedure to solve the quadratic problem of e-SVR with a modified loss function. Furthermore, we introduce the generalized approximate cross validation function to select the hyperparameters which affect the performance of e-SVR. Experimental results are then presented which illustrate the performance of the IRWLS procedure for e-SVR.