• Journal of the Korean Operations Research and Management Science Society
• /
• v.31 no.2
• /
• pp.69-85
• /
• 2006

Transductive Support Vector Machine(TSVM) is one of semi-supervised learning algorithms which exploit the domain structure of the whole data by considering labeled and unlabeled data together. Although it was proposed several years ago, there has been no efficient algorithm which can handle problems with more than hundreds of training examples. In this paper, we propose an efficient branch-and-bound algorithm which can solve large-scale TSVM problems with thousands of training examples. The proposed algorithm uses two bounding techniques: min-cut bound and reduced SVM bound. The min-cut bound is derived from a capacitated graph whose cuts represent a lower bound to the optimal objective function value of the dual problem. The reduced SVM bound is obtained by constructing the SVM problem with only labeled data. Experimental results show that the accuracy rate of TSVM can be significantly improved by learning from the optimal solution of TSVM, rather than an approximated solution.

• Steel and Composite Structures
• /
• v.24 no.6
• /
• pp.659-670
• /
• 2017

In the present work, an analytical solution for the static analysis of laminated composites, functionally graded and sandwich singly and doubly curved panels on the rectangular plan-form, subjected to uniformly distributed transverse loading is presented. Mathematical formulation is based on the higher order shear deformation theory and principle of virtual work is applied to derive the equations of equilibrium subjected to small deformation. A solution methodology based on the fast converging finite double Chebyshev series is used to solve the linear partial differential equations along with the simply supported boundary condition. The effect of span to thickness ratio, radius of curvature to span ratio, stacking sequence, power index are investigated. The accuracy of the solution is checked by the convergence study of non-dimensional central deflection and moments. Present results are compared with those available in the literature.

• Honam Mathematical Journal
• /
• v.45 no.4
• /
• pp.619-628
• /
• 2023

In this paper, the fractional order time-varying linear dynamical systems are investigated by using a residual power series method. A residual power series method (RPSM) is constructed for this problem. The exact solution is obtained by the Laplace transform method and the analytical solution is calculated via the residual power series method (RPSM). As an application, some examples are tested to show the accuracy and efficacy of the proposed methods. The obtained result showed that the proposed methods are effective and accurate for this type of problem.

• Journal of Mechanical Science and Technology
• /
• v.17 no.8
• /
• pp.1148-1155
• /
• 2003

The finite element method is one of the methods widely applied for predicting vibration in mechanical structures. In this paper, the effect of the mesh size of the finite element model on the accuracy of the numerical solutions of the structural vibration problems is investigated with particular focus on obtaining the optimal mesh size with respect to the solution accuracy and computational cost. The vibration response parameters of the natural frequency, modal density, and driving point mobility are discussed. For accurate driving point mobility calculation, the decay method is employed to experimentally determine the internal damping. A uniform plate simply supported at four corners is examined in detail, in which the response parameters are calculated by constructing finite element models with different mesh sizes. The accuracy of the finite element solutions of these parameters is evaluated by comparing with the analytical results as well as estimations based on the statistical energy analysis, or if not available, by testing the numerical convergence. As the mesh size becomes smaller than one quarter of the wavelength of the highest frequency of interest, the solution accuracy improvement is found to be negligible, while the computational cost rapidly increases. For mechanical structures, the finite element analysis with the mesh size of the order of quarter wavelength, combined with the use of the decay method for obtaining internal damping, is found to provide satisfactory predictions for vibration responses.

• Journal of Mechanical Science and Technology
• /
• v.20 no.12
• /
• pp.2218-2229
• /
• 2006

A conservative finite-volume numerical method for unstructured grids with the cell-centered method has been developed for computing flow and heat transfer by combining the attractive features of the existing pressure-based procedures with the advances made in unstructured grid techniques. This method uses an integral form of governing equations for arbitrary convex polyhedra. Care is taken in the discretization and solution procedure to avoid formulations that are cell-shape-specific. A collocated variable arrangement formulation is developed, i.e. all dependent variables such as pressure and velocity are stored at cell centers. For both convective and diffusive fluxes the forms superior to both accuracy and stability are particularly adopted and formulated through a systematic study on the existing approximation ones. Gradients required for the evaluation of diffusion fluxes and for second-order-accurate convective operators are computed by using a linear reconstruction based on the divergence theorem. Momentum interpolation is used to prevent the pressure checkerboarding and a segregated solution strategy is adopted to minimize the storage requirements with the pressure-velocity coupling by the SIMPLE algorithm. An algebraic solver using iterative preconditioned conjugate gradient method is used for the solution of linearized equations. The flow analysis code (PowerCFD) developed by the present method is evaluated for its application to several 2-D structured-mesh benchmark problems using a variety of unstructured quadrilateral and triangular meshes. The present flow analysis code by using unstructured grids with the cell-centered method clearly demonstrate the same accuracy and robustness as that for a typical structured mesh.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.8
• /
• pp.1976-1995
• /
• 2023

The anchor-free object detector CenterNet regards the object as a center point and predicts it based on the Gaussian circle region. For each object's center point, CenterNet directly regresses the width and height of the objects and finally gets the boundary range of the objects. However, the critical range of the object's center point can not be accurately limited by using the Gaussian circle region to constrain the prediction region, resulting in many low-quality centers' predicted values. In addition, because of the large difference between the width and height of different objects, directly regressing the width and height will make the model difficult to converge and lose the intrinsic relationship between them, thereby reducing the stability and consistency of accuracy. For these problems, we proposed a center point prediction method based on the Gaussian elliptic region and a size component regression method based on the small solution space. First, we constructed a Gaussian ellipse region that can accurately predict the object's center point. Second, we recode the width and height of the objects, which significantly reduces the regression solution space and improves the convergence speed of the model. Finally, we jointly decode the predicted components, enhancing the internal relationship between the size components and improving the accuracy consistency. Experiments show that when using CenterNet as the improved baseline and Hourglass-104 as the backbone, on the MS COCO dataset, our improved model achieved 44.7%, which is 2.6% higher than the baseline.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.10
• /
• pp.1-10
• /
• 2009

Precise branch predictor has a profound impact on system performance in modern processor architectures. Recent works show that prediction latency as well as prediction accuracy has a critical impact on overall system performance as well. However, prediction latency tends to be overlooked. In this paper, we propose Branch Pre-Prediction policy to tolerate branch prediction latency. The proposed solution allows that branch predictor can proceed its prediction without any information from the fetch engine, separating the prediction engine from fetch stage. In addition, we propose newly modified BTE structure to support our solution. The simulation result shows that proposed solution can hide most prediction latency with still providing the same level of prediction accuracy. Furthermore, the proposed solution shows even better performance than the ideal case, that is the predictor which always takes a single cycle prediction latency. In our experiments, IPC improvement is up to 11.92% and 5.15% in average, compared to conventional predictor system.

• Journal of Electrical Engineering and Technology
• /
• v.2 no.3
• /
• pp.312-319
• /
• 2007

The solution of the power flow is one of the most important problems in electrical power systems. These traditional methods such as Gauss-Seidel method and Newton-Raphson (NR) method have had drawbacks up to now such as initial values, abnormal operating solutions and divergences in heavy loads. In order to overcome theses problems, the power flow solution incorporating genetic algorithm (GA) is introduced in this paper. General operator of genetic algorithm, arithmetic crossover, and non-uniform mutation operator of GA are suggested to solve the power flow problem. While abnormal solution cannot be obtained by a NR method, multiple power flow solution can be obtained by a GA method. With a heavy load, both normal solution and abnormal solution can be obtained by a proposed method. In this paper, a floating number representation instead of the binary number representation is introduced for accuracy. Simulation results have been compared with traditional methods.

• Structural Engineering and Mechanics
• /
• v.35 no.2
• /
• pp.205-215
• /
• 2010

Recently, earthquake proof technology has been widely applied to both new and existing structures and bridges. The analysis of bridge systems equipped with structural control devices, which possess large degrees of freedom and nonlinear characteristics, is a result in time-consuming task. Therefore, a piecewise exact solution is proposed in this study to simplify the seismic mitigation analysis process for bridge systems equipped with sliding-type isolators. In this study, the simplified system having two degrees of freedom, to reasonably represent the large number of degrees of freedom of a bridge, and is modeled to obtain a piecewise exact solution for system responses during earthquakes. Simultaneously, we used the nonlinear finite element computer program to analyze the bridge responses and verify the accuracy of the proposed piecewise exact solution for bridge systems equipped with sliding-type isolators. The conclusions derived by comparing the results obtained from the piecewise exact solution and nonlinear finite element analysis reveal that the proposed solution not only simplifies the calculation process but also provides highly accurate seismic responses of isolated bridges under earthquakes.

• Computers and Concrete
• /
• v.18 no.1
• /
• pp.17-37
• /
• 2016

This paper compares the arc-length and explicit dynamic solution methods for nonlinear finite element analysis of prestressed concrete members subjected to monotonically increasing loads. The investigations have been conducted using an L-shaped, prestressed concrete spandrel beam, selected as a highly nonlinear problem from the literature to give insight into the advantages and disadvantages of these two solution methods. Convergence problems, computational effort, and quality of the results were investigated using the commercial finite element package ABAQUS. The work in this paper demonstrates that a static analysis procedure, based on the arc-length method, provides more accurate results if it is able to converge on the solution. However, it experiences convergence problems depending upon the choice of mesh configuration and the selection of concrete post-cracking response parameters. The explicit dynamic solution procedure appears to be more robust than the arc-length method in the sense that it provides acceptable solutions in cases when the arc-length approach fails, however solution accuracy may be slightly lower and computational effort may be significantly larger. Furthermore, prestressing forces must be introduced into the finite element model in different ways for the explicit dynamic and arc-length solution procedures.