• Journal of the Korean Data and Information Science Society
• /
• v.14 no.1
• /
• pp.103-111
• /
• 2003

Kernel machines are used widely in real-world regression tasks. Kernel ridge regressions(KRR) and support vector machines(SVM) are typical kernel machines. Here, we focus on two types of KRR. One is inductive KRR. The other is transductive KRR. In this paper, we study how differently they work in the interpolation and extrapolation areas. Furthermore, we study prediction interval estimation method for KRR. This turns out to be a reliable and practical measure of prediction interval and is essential in real-world tasks.

• Journal of the Korean Mathematical Society
• /
• v.50 no.3
• /
• pp.557-578
• /
• 2013

For an ideal $\mathcal{I}$ of a preadditive category $\mathcal{A}$, we study when the canonical functor $\mathcal{C}:\mathcal{A}{\rightarrow}\mathcal{A}/\mathcal{I}$ is local. We prove that there exists a largest full subcategory $\mathcal{C}$ of $\mathcal{A}$, for which the canonical functor $\mathcal{C}:\mathcal{C}{\rightarrow}\mathcal{C}/\mathcal{I}$ is local. Under this condition, the functor $\mathcal{C}$, turns out to be a weak equivalence between $\mathcal{C}$, and $\mathcal{C}/\mathcal{I}$. If $\mathcal{A}$ is additive (with splitting idempotents), then $\mathcal{C}$ is additive (with splitting idempotents). The category $\mathcal{C}$ is ample in several cases, such as the case when $\mathcal{A}$=Mod-R and $\mathcal{I}$ is the ideal ${\Delta}$ of all morphisms with essential kernel. In this case, the category $\mathcal{C}$ contains, for instance, the full subcategory $\mathcal{F}$ of Mod-R whose objects are all the continuous modules. The advantage in passing from the category $\mathcal{F}$ to the category $\mathcal{F}/\mathcal{I}$ lies in the fact that, although the two categories $\mathcal{F}$ and $\mathcal{F}/\mathcal{I}$ are weakly equivalent, every endomorphism has a kernel and a cokernel in $\mathcal{F}/{\Delta}$, which is not true in $\mathcal{F}$. In the final section, we extend our theory from the case of one ideal$\mathcal{I}$ to the case of $n$ ideals $\mathcal{I}_$, ${\ldots}$, $\mathca{l}_n$.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.12 no.5
• /
• pp.67-72
• /
• 2003

For the analysis of Kirchhoff plate bending problems, a new meshless method is implemented. For the satisfaction of the $C^1$ continuity condition in which the first derivative is treated an another primary variable, Hermite interpolation is enforced on standard reproducing kernel particle method. In order to impose essential boundary conditions on solving $C^1$ continuity problems, shape function modifications are adopted. Through numerical tests, the characteristics and accuracy of the HRKPM are investigated and compared with the finite element analysis. By this implementatioa it is shown that high accuracy is achieved by using HRKPM for solving Kirchhoff plate bending problems.

• Journal of Magnetics
• /
• v.22 no.2
• /
• pp.227-233
• /
• 2017

The purpose of this study was to evaluate the effects of X-ray photon energy, tissue density, and the kernel essential for image reconstruction on the image quality by measuring HU and noise. Images were obtained by scanning the RMI density phantom within the CT device, and HU and noise were measured as follows: images were obtained by varying the tube voltages, the tube currents and eight different kernels. The greater the voltage-dependent change in the HU value but the noise was decreased. At all densities, changes in the tube current did not exert any significant influence on the HU value, whereas the noise value gradually decreased as the tube current increased. At all densities, changes in the kernel did not exert any significant influence on the HU value. The noise value gradually increased in the lower kernel range, but rapidly increased in the higher kernel range. HU is influenced by voltage and density, and noise is influenced by voltage, current, kernel, and density. This affects contrast resolution and spatial resolution.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2002.04a
• /
• pp.12-18
• /
• 2002

For the analysis of Kirchhoff plate bending problems, a new meshless method is implemented. For the satisfaction of the C¹ continuity condition in which the first derivative is treated as another primary variable, Hermite interpolation is enforced on standard reproducing kernel particle method. In order to impose essential boundary conditions on solving C¹ continuity problems, shape function modifications are adopted. Through numerical tests, the characteristics and accuracy of the HRKPM are investigated and compared with the finite element analysis. By this implementation, it is shown that high accuracy is achieved by using HRKPM fur solving Kirchhoff plate bending problems.

• Interaction and multiscale mechanics
• /
• v.3 no.3
• /
• pp.277-298
• /
• 2010

The complex variable reproducing kernel particle method (CVRKPM) and the FEM are coupled in this paper to analyze the two-dimensional potential problems. The coupled method not only conveniently imposes the essential boundary conditions, but also exploits the advantages of the individual methods while avoiding their disadvantages, resulting in improved computational efficiency. A hybrid approximation function is applied to combine the CVRKPM with the FEM. Formulations of the coupled method are presented in detail. Three numerical examples of the two-dimensional potential problems are presented to demonstrate the effectiveness of the new method.

• Nuclear Engineering and Technology
• /
• v.55 no.6
• /
• pp.1966-1973
• /
• 2023

Virtual reality technology has been widely used in the field of nuclear and radiation safety, dose rate computing in virtual environment is essential for optimizing radiation protection and planning the work in radioactive-controlled area. Because the CPU-based gamma dose rate computing takes up a large amount of time and computing power for voxelization of volumetric radioactive source, it is inefficient and limited in its applied scope. This study is to develop an efficient gamma dose rate computing code and apply into fast virtual simulation. To improve the computing efficiency of the point kernel algorithm in the reference (Li et al., 2020), we design a GPU-based computing framework for taking full advantage of computing power of virtual engine, propose a novel voxelization algorithm of volumetric radioactive source. According to the framework, we develop the GPPK(GPU-based point kernel gamma dose rate computing) code using GPU programming, to realize the fast dose rate computing in virtual world. The test results show that the GPPK code is play and plug for different scenarios of virtual simulation, has a better performance than CPU-based gamma dose rate computing code, especially on the voxelization of three-dimensional (3D) model. The accuracy of dose rates from the proposed method is in the acceptable range.

• Journal of Institute of Control, Robotics and Systems
• /
• v.3 no.2
• /
• pp.187-196
• /
• 1997

This paper presents a semi-automatic methodology to synthesize executable digital controller saftware in a multi-loop control system. A digital controller is described by a task graph and end-to-end timing requirements. A task graph denotes the software structure of the controller, and the end-to-end requirements establish timing relationships between external inputs and outputs. Our approach translates the end-to-end requirements into a set of task attributes such as task periods and deadlines using nonlinear optimization techniques. Such attributes are essential for control engineers to implement control programs and schedule them in a control system with limited resources. In current engineering practice, human programmers manually derive those attributes in an ad hoc manner: they often resort to radical over-sampling to safely guarantee the given timing requirements, and thus render the resultant system poorly utilized. After task-specific attributes are derived, the tasks are scheduled on a single CPU and the compiled kernel is synthesized. We illustrate this process with a non-trivial servo motor control system.

• 한국경영정보학회:학술대회논문집
• /
• 2007.06a
• /
• pp.608-613
• /
• 2007

Prediction of corporate bankruptcies has long been an important topic and has been studied extensively in the finance and management literature because it is an essential basis for the risk management of financial institutions. Recently, support vector machines (SVMs) are becoming popular as a tool for bankruptcy prediction because they use a risk function consisting of the empirical error and a regularized term which is derived from the structural risk minimization principle. In addition, they don't require huge training samples and have little possibility of overfitting. However. in order to Use SVM, a user should determine several factors such as the parameters ofa kernel function, appropriate feature subset, and proper instance subset by heuristics, which hinders accurate prediction results when using SVM In this study, we propose a novel hybrid SVM classifier with simultaneous optimization of feature subsets, instance subsets, and kernel parameters. This study introduces genetic algorithms (GAs) to optimize the feature selection, instance selection, and kernel parameters simultaneously. Our study applies the proposed model to the real-world case for bankruptcy prediction. Experimental results show that the prediction accuracy of conventional SVM may be improved significantly by using our model.

• Korean Journal of Computational Design and Engineering
• /
• v.13 no.5
• /
• pp.334-341
• /
• 2008

Computer-aided engineering (CAE) analysis is considered essential for product development because it decreases the simulation time, reduces the prototyping costs, and enhances the reusability of product parts. The reuse of quality-assured CAE data has been continually increasing due to the extension of product lifecycle management; PLM, which is widely used, shortens the product development cycle and improves the product quality. However, less attention has been paid to systematic research on the interoperability of CAE data because of the diversity of CAE data and because the structure of CAE data is more complex than that of CAD data. In this paper, we suggest a CAE data exchange method for the effective sharing of geometric and analysis data. The method relies on heterogeneous CAE systems, a virtual reality system, and our developed CAE middleware for CAE data exchange. We also designed a generic CAE kernel, which is a critical part of the CAE middleware. The kernel offers a way of storing analysis data from various CAE systems, and, with the aid of a script command, enabling the data to be translated for a different system. The reuse of CAE data is enhanced by the fact that the CAE middle-ware can be linked with a virtual reality system or a product data management system.