• Communications of the Korean Mathematical Society
• /
• v.17 no.1
• /
• pp.121-142
• /
• 2002

We are interested in the curve fitting problems in such a way that the sum of the squares of the orthogonal distances to the given data points is minimized. Especially, the fitting an ellipse to the given data points is a problem that arises in many application areas, e.g. computer graphics, coordinate metrology, etc. In [1] the problem of fitting ellipses was considered and numerically solved with general purpose methods. In this paper we present another new ellipse fitting algorithm. Our algorithm if mainly based on the steepest descent procedure with the view of ensuring the convergence of the corresponding quadratic function Q(u) to a local minimum. Numerical examples are given.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.5
• /
• pp.1153-1165
• /
• 2017

Structural equation modeling (SEM) is a basic tool for testing theories in a variety of disciplines. A maximum likelihood (ML) method for parameter estimation is by far the most widely used in SEM. Alternatively, two-stage least squares (2SLS) estimator has been proposed as a more robust procedure to address model misspecification. A regularized extension of 2SLS, two-stage ridge least squares (2SRLS) has recently been introduced as an alternative to ML to effectively handle the small-sample-size issue. However, it is unclear whether and when misspecification and small sample sizes may pose problems in theory testing with 2SLS, 2SRLS, and ML. The purpose of this article is to evaluate the three estimation methods in terms of inferences errors as well as parameter recovery under two experimental conditions. We find that: 1) when the model is misspecified, 2SRLS tends to recover parameters better than the other two estimation methods; 2) Regardless of specification errors, 2SRLS produces small or relatively acceptable Type II error rates for the small sample sizes.

• Journal of Institute of Control, Robotics and Systems
• /
• v.20 no.9
• /
• pp.908-913
• /
• 2014

In this paper, we propose a method by which QP (Quadratic Programming) problems can be solved in realtime so that we can implement the predictive control algorithm on a microcontroller system. Firstly, we derive a solution to QP problems by converting the original QP problems to its equivalent least squares with inequalities. Secondly, we propose a predictive control algorithm that can give good realtime computation performance by utilizing the fact that some parameters needed for solving QP problems can be computed offline. Finally, we illustrate that the proposed method can give good realtime features by running the C-code application constructed using the proposed method on a microncontroller system.

• Interaction and multiscale mechanics
• /
• v.2 no.4
• /
• pp.333-352
• /
• 2009

We introduce a radial basis collocation method to solve axially moving beam problems which involve $2^{nd}$ order differentiation in time and $4^{th}$ order differentiation in space. The discrete equation is constructed based on the strong form of the governing equation. The employment of multiquadrics radial basis function allows approximation of higher order derivatives in the strong form. Unlike the other approximation functions used in the meshfree methods, such as the moving least-squares approximation, $4^{th}$ order derivative of multiquadrics radial basis function is straightforward. We also show that the standard weighted boundary collocation approach for imposition of boundary conditions in static problems yields significant errors in the transient problems. This inaccuracy in dynamic problems can be corrected by a statically condensed semi-discrete equation resulting from an exact imposition of boundary conditions. The effectiveness of this approach is examined in the numerical examples.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.3
• /
• pp.109-115
• /
• 2020

In this paper, we propose an artificial intelligence(AI) technology analysis using partial least square(PLS) regression model. AI technology is now affecting most areas of our society. So, it is necessary to understand this technology. To analyze the AI technology, we collect the patent documents related to AI from the patent databases in the world. We extract AI technology keywords from the patent documents by text mining techniques. In addition, we analyze the AI keyword data by PLS regression model. This regression model is based on the technique of partial least squares used in the advanced analyses such as bioinformatics, social science, and engineering. To show the performance of our proposed method, we make experiments using AI patent documents, and we illustrate how our research can be applied to real problems. This paper is applicable not only to AI technology but also to other technological fields. This also contributes to understanding other various technologies by PLS regression analysis.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.25 no.2
• /
• pp.139-148
• /
• 2012

The MLS(Moving Least Squares) Difference Method is a numerical scheme that combines the MLS method of Meshfree method and Taylor expansion involving not numerical quadrature or mesh structure but only nodes. This paper presents an dynamic algorithm of MLS difference method for solving transient solid mechanics problems. The developed algorithm performs time integration by using Newmark method and directly discretizes strong forms. It is very convenient to increase the order of Taylor polynomial because derivative approximations are obtained by the Taylor series expanded by MLS method without real differentiation. The accuracy and efficiency of the dynamic algorithm are verified through numerical experiments. Numerical results converge very well to the closed-form solutions and show less oscillation and periodic error than FEM(Finite Element Method).

• Communications for Statistical Applications and Methods
• /
• v.21 no.5
• /
• pp.395-409
• /
• 2014

Data limited, partial, or incomplete are known as an ill-posed problem. If the data with ill-posed problems are analyzed by traditional statistical methods, the results obviously are not reliable and lead to erroneous interpretations. To overcome these problems, we propose a dual generalized maximum entropy (dual GME) estimator for panel data regression models based on an unconstrained dual Lagrange multiplier method. Monte Carlo simulations for panel data regression models with exogeneity, endogeneity, or/and collinearity show that the dual GME estimator outperforms several other estimators such as using least squares and instruments even in small samples. We believe that our dual GME procedure developed for the panel data regression framework will be useful to analyze ill-posed and endogenous data sets.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.930-932
• /
• 2003

For this study we used a CCD video camera to capture the pavement image information via the computer. During investigation processing, the CCD video camera captured 10${\sim}$30 images per second. If the vehicle velocity is too fast, the collected images will be duplicated and if the velocity is too slow there will be a gapped between images. Therefore, in order to control the efficiency of the image grabber we should add accessory tools such as the Differential Global Positioning System (DGPS) and odometer. Furthermore, Kalman Filtering can also solve these problems. After the CCD video camera captured the pavement images, we used the Least-Squares method to eliminate images of gradation which have non-uniform surfaces due to the illumination at night. The Fuzzy Entropy method calculates images of threshold segments and creates binary images. Finally, the Object Labeling algorithm finds objects that are cracks or noises from the binary image based on volume pixels of the object. We used these algorithms and tested them, also providing some discussion and suggestions.

• Journal of Institute of Control, Robotics and Systems
• /
• v.19 no.7
• /
• pp.633-639
• /
• 2013

Several estimation methods used in the range measurement based wireless localization area have individual problems. These problems may not occur according to certain application areas. However, these problems may give rise to serious problems in particular applications. In this paper, three methods, ILS (Iterative Least Squares), DS (Direct Solution), and DSRM (Difference of Squared Range Measurements) methods are considered. Problems that can occur in these methods are defined and a simple hybrid solution is presented to solve them. The ILS method is the most frequently used method in wireless localization and has local minimum problems and a large computational burden compared with closed-form solutions. The DS method requires less processing time than the ILS method. However, a solution for this method may include a complex number caused by the relations between the location of reference nodes and range measurement errors. In the near-field region of the complex solution, large estimation errors occur. In the DSRM method, large measurement errors occur when the mobile node is far from the reference nodes due to the combination of range measurement error and range data. This creates the problem of large localization errors. In this paper, these problems are defined and a hybrid localization method is presented to avoid them by integrating the DS and DSRM methods. The defined problems are confirmed and the performance of the presented method is verified by a Monte-Carlo simulation.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.6A
• /
• pp.835-842
• /
• 2008

The response surface method (RSM) is widely adopted for the structural reliability analysis because of its numerical efficiency. However, the RSM is still time consuming for large-scale applications and sometimes shows large errors in the calculation of sensitivity of reliability index with respect to random variables. Therefore, this study proposes a new RSM in which moving least squares (MLS) approximation is applied. Least squares approximation generally used in the common RSM gives equal weight to the coefficients of the response surface function (RSF). On the other hand, The MLS approximation gives higher weight to the experimental points closer to the design point, which yields the RSF more similar to the limit state at the design point. In the procedure of the proposed method, a linear RSF is constructed initially and then a quadratic RSF is formed using the axial experimental points selected from the reduced region where the design point is likely to exist. The RSF is updated successively by adding one more experimental point to the previously sampled experimental points. In order to demonstrate the effectiveness of the proposed method, mathematical problems and ten-bar truss are considered as numerical examples. As a result, the proposed method shows better accuracy and computational efficiency than the common RSM.