• Bulletin of the Korean Mathematical Society
• /
• v.35 no.2
• /
• pp.345-362
• /
• 1998

We introduce and anlyze a naturally parallelizable frequency-domain method for parabolic problems with a Neumann boundary condition. After taking the Fourier transformation of given equations in the space-time domain into the space-frequency domain, we solve an indefinite, complex elliptic problem for each frequency. Fourier inversion will then recover the solution of the original problem in the space-time domain. Existence and uniqueness of a solution of the transformed problem corresponding to each frequency is established. Fourier invertibility of the solution in the frequency-domain is also examined. Error estimates for a finite element approximation to solutions fo transformed problems and full error estimates for solving the given problem using a discrete Fourier inverse transform are given.

• Journal of the Korean Mathematical Society
• /
• v.54 no.6
• /
• pp.1683-1698
• /
• 2017

In this paper, we study the $Carath{\acute{e}}odory$ extremal problems on the Hua domain of the first three types. We give the explicit formula for the $Carath{\acute{e}}odory$ extremal problems between the first three types of Hua domain and the unit ball, which improves the works done on Hua domain and Cartan-egg domain and super-Cartan domain.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.23 no.55
• /
• pp.85-96
• /
• 2000

This paper describes IDMS, a graph-based modeling system that supports problem structuring. We employs influence diagram as a problem representation tool, that is, a modeling tool. In particular, IDMS is designed as domain-independent shell. Therefore, a modeler can change the contents of the knowledge base to suit his/her own interested domain. Since the knowledge base of IDMS contains both modeling knowledge and domain knowledge, IDMS provides not only the syntactic support for modeling tool, but also the semantic support for problem domain. To apply the method in the real world context, we tested IDMS on the process selection problem in business reengineering, which is typical semi-structured problem.

• Journal of Information Science Theory and Practice
• /
• v.1 no.1
• /
• pp.10-26
• /
• 2013

Domain transfer is a widely recognized problem for machine learning algorithms because models built upon one data domain generally do not perform well in another data domain. This is especially a challenge for tasks such as opinion classification, which often has to deal with insufficient quantities of labeled data. This study investigates the feasibility of self-training in dealing with the domain transfer problem in opinion classification via leveraging labeled data in non-target data domain(s) and unlabeled data in the target-domain. Specifically, self-training is evaluated for effectiveness in sparse data situations and feasibility for domain adaptation in opinion classification. Three types of Web content are tested: edited news articles, semi-structured movie reviews, and the informal and unstructured content of the blogosphere. Findings of this study suggest that, when there are limited labeled data, self-training is a promising approach for opinion classification, although the contributions vary across data domains. Significant improvement was demonstrated for the most challenging data domain-the blogosphere-when a domain transfer-based self-training strategy was implemented.

• Journal of information and communication convergence engineering
• /
• v.21 no.2
• /
• pp.139-144
• /
• 2023

Engineers prefer deep neural networks (DNNs) for solving computer vision problems. However, DNNs pose two major problems. First, neural networks require large amounts of well-labeled data for training. Second, the covariate shift problem is common in computer vision problems. Domain adaptation has been proposed to mitigate this problem. Recent work on adversarial-learning-based unsupervised domain adaptation (UDA) has explained transferability and enabled the model to learn robust features. Despite this advantage, current methods do not guarantee the distinguishability of the latent space unless they consider class-aware information of the target domain. Furthermore, source and target examples alone cannot efficiently extract domain-invariant features from the encoded spaces. To alleviate the problems of existing UDA methods, we propose the mixup regularization in adversarial discriminative domain adaptation (ADDA) method. We validated the effectiveness and generality of the proposed method by performing experiments under three adaptation scenarios: MNIST to USPS, SVHN to MNIST, and MNIST to MNIST-M.

• Journal of the Korean Mathematical Society
• /
• v.57 no.5
• /
• pp.1267-1286
• /
• 2020

We pay attention to a coupled problem by a penalty term which is induced from non-overlapping domain decomposition methods based on augmented Lagrangian methodology. The coupled problem is composed by two parts mainly; one is a problem associated with local problems in non-overlapping subdomains and the other is a coupled part over all subdomains due to the penalty term. For the speedup of iterative solvers for the coupled problem, we propose two different types of preconditioners: a block-diagonal preconditioner and an additive Schwarz preconditioner as overlapping domain decomposition methods. We analyze the coupled problem and the preconditioned problems in terms of their condition numbers. Finally we present numerical results which show the performance of the proposed methods.

• Structural Engineering and Mechanics
• /
• v.32 no.1
• /
• pp.37-53
• /
• 2009

In this paper the problem of calculating the probability that the responses of a wind-excited structure exceed specified thresholds within a given time interval is considered. The failure domain of the problem can be expressed as a union of elementary failure domains whose boundaries are of quadratic form. The Domain Decomposition Method (DDM) is employed, after being appropriately extended, to solve this problem. The probability estimate of the overall failure domain is given by the sum of the probabilities of the elementary failure domains multiplied by a reduction factor accounting for the overlapping degree of the different elementary failure domains. The DDM is extended with the help of Line Sampling (LS), from its original presentation where the boundary of the elementary failure domains are of linear form, to the current case involving quadratic elementary failure domains. An example involving an along-wind excited steel building shows the accuracy and efficiency of the proposed methodology as compared with that obtained using standard Monte Carlo simulations (MCS).

• Korean Journal of Mathematics
• /
• v.10 no.1
• /
• pp.53-58
• /
• 2002

In this paper we consider the Cauchy problem $$dx(t)/dt=x(t)^n,\;x(0)=x_0$$. The domain of existence is lower semicontinuous for perturbations of the data. We present a simple formula for the time of existence which is exact when there exists no perturbations.

• Journal of Korea Multimedia Society
• /
• v.25 no.6
• /
• pp.819-832
• /
• 2022

Cross-domain recommender system transfers knowledge across different domains to improve the recommendation performance in a target domain that has a relatively sparse model. However, they suffer from the "negative transfer" in which transferred knowledge operates as noise. This paper proposes a novel Multiple Fusion-based Deep Cross-Domain Recommendation named MFDCR. We exploit Doc2Vec, one of the famous word embedding techniques, to fuse data user-wise and transfer knowledge across multi-domains. It alleviates the "negative transfer" problem. Additionally, we introduce a simple multi-layer perception to learn the user-item interactions and predict the possibility of preferring items by users. Extensive experiments with three domain datasets from one of the most famous services Amazon demonstrate that MFDCR outperforms recent single and cross-domain recommendation algorithms. Furthermore, experimental results show that MFDCR can address the problem of "negative transfer" and improve recommendation performance for multiple domains simultaneously. In addition, we show that our approach is efficient in extending toward more domains.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 1992.04a
• /
• pp.58-64
• /
• 1992

It is usual that underground structures are constructed within multi-layered medium. In this paper, an efficient numerical model ling of multi-layered structural systems is studied using coupled analysis of finite elements and boundary elements. The finite elements are applied to the area in which the material nonlinearity is dominated, and the boundary elements are applied to the far field area where the nonlinearity is relatively weak. In the boundary element model 1 ins of the multi-layered medium, fundamental solutions are restricted. Thus, methods which can utilize existing Kelvin and Melan solution are sought for the interior multi-layered domain problem and semi infinite domain problem. Interior domain problem which has piecewise homogeneous layers is analyzed using boundary elements with Kelvin solution; by discretizing each homogeneous subregion and applying compatibility and equilibrium conditions between interfaces. Semi-infinite domain problem is analyzed using boundary elements with Melan solution, by superposing unit stiffness matrices which are obtained for each layer by enemy method. Each methodology is verified by comparing its results which the results from the finite element analysis and it is concluded that coupled analysis using boundary elements and finite elements can be reasonable and efficient if the superposition technique is applied for the multi-layered semi-infinite domain problems.