• Journal of the Chungcheong Mathematical Society
• /
• v.27 no.4
• /
• pp.707-720
• /
• 2014

We get a theorem which shows the existence of at least three solutions for some elliptic system with Dirichlet boundary condition. We obtain this result by using the finite dimensional reduction method which reduces the infinite dimensional problem to the finite dimensional one. We also use the critical point theory on the reduced finite dimensioal subspace.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1512-1518
• /
• 2021

Since single cell RNA sequencing provides the expression profiles of individual cells, it provides higher cellular differential resolution than traditional bulk RNA sequencing. Using these single cell RNA sequencing data, clustering analysis is generally conducted to find cell types and understand high level biological processes. In order to effectively process the high-dimensional single cell RNA sequencing data fir the clustering analysis, this paper uses a variational autoencoder to transform a high dimensional data space into a lower dimensional latent space, expecting to produce a latent space that can give more accurate clustering results. By clustering the features in the transformed latent space, we compare the performance of various classical clustering methods for single cell RNA sequencing data. Experimental results demonstrate that the proposed framework outperforms many state-of-the-art methods under various clustering performance metrics.

• Journal of the Korean Solar Energy Society
• /
• v.36 no.3
• /
• pp.9-16
• /
• 2016

This paper discusses a refined model for investigating the micro-mechanical behavior of beam-like structures, which are composed of various elastic moduli and complex geometries varying through the cross-section directions and are also periodically-repeated and heterogeneous along the axial direction. Following the previous work (Lee and Yu, 2011), the original three-dimensional static problem is first formulated in a unified and compact form using the concept of decomposition of the rotation tensor. Taking advantage of the smallness of the cross-sectional dimension-to-length parameter and the micro-to-macro heterogeneity, while also performing homogenization along the dimensional reduction simultaneously, the variational asymptotic method is rigorously used to construct a total energy function, which is asymptotically correct up to the second order. Furthermore, through the transformation procedure based on the pure kinematic relations and the linearized equilibrium equations, a generalized Timoshenko model is systematically established. For the purpose of dealing with realistic and complex geometries and constituent materials at the microscopic level, this present approach is incorporated into a commercial analysis package. A few examples available in literature are used to demonstrate the consistency and efficiency of this proposed model, especially for the structures, in which the effects of transverse shear deformations are significant.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.475-482
• /
• 2022

Signal pre-processing and post-processing are some areas of study around electrical resistance tomography due to the low spatial resolution of pixel-based reconstructed images. In addition, methods that improve integrity and noise reduction are candidates for application in ERT. Lately, formulations of image processing methods provide new implementations and studies to improve the response against noise. For example, compact variational mode decomposition has recently shown good performance in image decomposition and segmentation. The results from this first approach of C-VMD to ERT show an improvement due to image segmentation, providing filtering of noise in the background and location of the target.

• Journal of Environmental Science International
• /
• v.19 no.8
• /
• pp.983-999
• /
• 2010

Wind power energy is one of the favorable and fast growing renewable energies. It is most important for exact analysis of wind to evaluate and forecast the wind power energy. The purpose of this study is to improve the performance of numerical atmospheric model by data assimilation over a complex coastal area. The benefit of the profiler is its high temporal resolution and dense observation data at the lower troposphere. Three wind profiler sites used in this study are inhomogeneously situated near south-western coastal area of Korean Peninsula. The method of the data assimilation for using the profiler to the model simulation is the three-dimensional variational data assimilation (3DVAR). The experiment of two cases, with/without assimilation, were conducted for how to effect on model results with wind profiler data. It was found that the assimilated case shows the more reasonable results than the other case compared with vertical observation and surface Automatic Weather Station(AWS) data. Although the effect of sonde data was better than profiler at a higher altitude, the profiler data improves the model performance at lower atmosphere. Comparison with the results of 4 June and 5 June suggests that the efficiency with hourly assimilated profiler data is strongly influenced by synoptic conditions. The reduction rate of Normalized Mean Error(NME), mean bias normalized by averaged wind speed of observation, on 4 June was 28% which was larger than 13% of 5 June. In order to examine the difference in wind power energy, the wind power density(WPD) was calculated and compared.

• Geomechanics and Engineering
• /
• v.19 no.5
• /
• pp.463-472
• /
• 2019

The mode perturbation method (MPM) is suitable and efficient for solving the eigenvalue problem of a nonuniform soil deposit whose property varies with depth. However, results of the MPM do not always converge to the exact solution, when the variation of soil deposit property is discontinuous. This discontinuity is typical because soil is usually made up of sedimentary layers of different geologic materials. Based on the energy integral of the variational principle, a new mode perturbation method, the energy-based mode perturbation method (EMPM), is proposed to address the convergence of the perturbation solution on the natural frequencies and the corresponding mode shapes and is able to find solution whether the soil properties are continuous or not. First, the variational principle is used to transform the variable coefficient differential equation into an equivalent energy integral equation. Then, the natural mode shapes of the uniform shear beam with same height and boundary conditions are used as Ritz function. The EMPM transforms the energy integral equation into a set of nonlinear algebraic equations which significantly simplifies the eigenvalue solution of the soil layer with variable properties. Finally, the accuracy and convergence of this new method are illustrated with two case study examples. Numerical results show that the EMPM is more accurate and convergent than the MPM. As for the mode shapes of the uniform shear beam included in the EMPM, the additional 8 modes of vibration are sufficient in engineering applications.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2002.05a
• /
• pp.790-795
• /
• 2002

The object of this research is limited to the reduction of compression process noise only among the main sources of compressor noise such as motor noise, compression process noise, and valve port flow noise. Thus the research is focused on the wave motion rather than the particle motion of sound wave travels. A muffler is a commonly used device to reduce the compression process noise, generated by the pressure pulsations caused by the cyclic compression process. In this research, the acoustic characteristics of the muffler are analyzed by using the normal gradient integral equation proposed by Wu and Wan. Moreover, a commercial code SYSNOISE developed by indirect variational boundary integral equation is also used to validate the results. For the noise reduction, the topology optimization technique using a genetic algorithm is used. The number, size and position of the muffler holes are considered as design variables. Compared with original design, the optimized design has very improved acoustic characteristics. Both numerical and experimental analyses are used to evaluate new design.

• Journal of KIISE
• /
• v.43 no.7
• /
• pp.795-800
• /
• 2016

In image classification, recent CNNs compete with human performance. However, there are limitations in more general recognition. Herein we deal with indoor images that contain too much information to be directly processed and require information reduction before recognition. To reduce the amount of data processing, typically variational inference or variational Bayesian methods are suggested for object detection. However, these methods suffer from the difficulty of marginalizing over the given space. In this study, we propose an image-text integrated recognition system using active vision based on Spatial Transformer Networks. The system attempts to efficiently sample a partial region of a given image for a given language information. Our experimental results demonstrate a significant improvement over traditional approaches. We also discuss the results of qualitative analysis of sampled images, model characteristics, and its limitations.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.6
• /
• pp.237-244
• /
• 2022

In this paper, we present VAE-based recommendation using online behavior log and purchase history to overcome data sparsity and cold start. To generate a variable for customers' purchase history, embedding and dimensionality reduction are applied to the customers' purchase history. Also, Variational Autoencoders are applied to online behavior and purchase history. A total number of 12 variables are used, and nDCG is chosen for performance evaluation. Our experimental results showed that the proposed VAE-based recommendation outperforms SVD-based recommendation. Also, the generated purchase history variable improves the recommendation performance.

• Wind and Structures
• /
• v.19 no.1
• /
• pp.35-58
• /
• 2014

A numerical simultaneous solution involving a linear elastic model was applied to study the fluid-structure interaction (FSI) of membrane structures under wind actions, i.e., formulating the fluid-structure system with a single equation system and solving it simultaneously. The linear elastic model was applied to managing the data transfer at the fluid and structure interface. The monolithic equation of the FSI system was formulated by means of variational forms of equations for the fluid, structure and linear elastic model, and was solved by the Newton-Raphson method. Computation procedures of the proposed simultaneous solution are presented. It was applied to computation of flow around an elastic cylinder and a typical FSI problem to verify the validity and accuracy of the method. Then fluid-structure interaction analyses of a saddle membrane structure under wind actions for three typical cases were performed with the method. Wind pressure, wind-induced responses, displacement power spectra, aerodynamic damping and added mass of the membrane structure were computed and analyzed.