• 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.

• East Asian mathematical journal
• /
• v.17 no.1
• /
• pp.1-17
• /
• 2001

In this paper, fuzzy D-continuous function is defined. Some basic properties of this continuity are summarized; and sufficient conditions on domain and/or ranges implying fuzzy D-continuity of fuzzy D-continuous functions are given. Also fuzzy D-regular space is defined and by using fuzzy D-continuity, the condition which is equivalent to fuzzy D-regular space, is given.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1282-1285
• /
• 2022

기존의 video domain adaptation은 closed set 환경에서 주로 연구되었다. 하지만 이는 source와 target의 label이 같다는 비현실적인 전제를 요구한다. 따라서 본 논문에서는 target의 label space가 source보다 넓은 open set video domain adaptation 문제를 다룬다. 우린 open set image domain adaptation에서 사용되는 방법들을 video로 확장 시켜 모델을 설계하고 UCF to HMDB, HMDB to UCF 와 같은 video dataset에서 실험하였다. 그 결과 source only 대비 UCF to HMDB에서 12%, HMDB to UCF 7.1% 향상된 결과를 얻었다.

• The Pure and Applied Mathematics
• /
• v.15 no.1
• /
• pp.47-55
• /
• 2008

Local convergence results for three Newton-like methods in Banach space are provided. A comparison is given between the three convergence radii. Then we show that using the largest convergence radius we can pick an initial guess from with we start the corresponding iteration. It turns out that after a finite number of steps we can always use the iterate found as the starting guess for a faster method, since this iterate will be inside the convergence domain of the new method.

• Journal of Information Technology Applications and Management
• /
• v.23 no.2
• /
• pp.143-152
• /
• 2016

We can organize, manage, search, and process the documents efficiently by a document clustering. In general, the documents are clustered in a high dimensional feature space because the documents consist of many terms. In this paper, we propose a new method to cluster the documents efficiently in a low dimensional feature space by finding the core concepts from a domain ontology corresponding to the particular area documents. The experiment shows that our clustering method has a good performance.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• v.2
• /
• pp.341-345
• /
• 2006

The GPS is widely used in various parts, therefore it is required higher integrity and continuity. These integrity and continuity are threatened by outer jamming signals which are intended or not. And various anti-jamming ways have been studied to remove these jamming signals. In this paper, we are going to test the efficiency of the anti-jamming algorithm in space and time-space domain, and analyze

• Journal of the Korean Mathematical Society
• /
• v.41 no.1
• /
• pp.175-192
• /
• 2004

This talk discusses conditions on the numerical range of a holomorphic function defined on a bounded convex domain in a complex Banach space that imply that the function has a unique fixed point. In particular, extensions of the Earle-Hamilton Theorem are given for such domains. The theorems are applied to obtain a quantitative version of the inverse function theorem for holomorphic functions and a distortion form of Cartan's unique-ness theorem.

• Bulletin of the Korean Chemical Society
• /
• v.7 no.3
• /
• pp.204-210
• /
• 1986

It is shown that the linear stability coincides with the thermodynamic stability in the case of stress tensor evolution for simple dense fluids even if the constitutive (evolution) equation for the stress tensor is nolinear. The domain of coincidence can be defined in the space of parameters appearing in the constitutive equation and we find the domain is confined in an elliptical cone in a three-dimensional parameter space. The corresponding state theory in rheology of simple dense fluids is also further examined. The validity of the idea is strengthened by the examination.

• Journal of Information Technology and Architecture
• /
• v.9 no.3
• /
• pp.243-251
• /
• 2012

This paper presents the Conceptual Models of the Mission Space-Korea (CMMS-K), which is an ontology-based conceptual modeling framework of the mission space. Through modeling and simulating military trainings, we can reduce the cost of actual military trainings in terms of time, space, and supplies. CMMS-K is being developed to improve the interoperability and reusability of defense models and simulations. CMMS-K reflects the needs and characteristics of Korean military while referring to existing military conceptual modeling frameworks. The main components of CMMS-K contain domain ontologies, a mission space model description language, a mission space modeling tool, and a CMMS-K management system. CMMS-K domain ontologies consist of entity and task ontologies. In this paper, the CMMS-K domain ontologies are described in detail and the feasibility of the proposed method is discussed with a case study.

• International Journal of Aeronautical and Space Sciences
• /
• v.16 no.2
• /
• pp.177-189
• /
• 2015

This paper presents an all-direct domain decomposition approach for large-scale structural analysis. The proposed approach achieves computational robustness and efficiency by enforcing the compatibility of the displacement field across the sub-domain boundaries via local Lagrange multipliers and augmented Lagrangian formulation (ALF). The proposed domain decomposition approach was compared to the existing FETI approach in terms of the computational time and memory usage. The parallel implementation of the proposed algorithm was described in detail. Finally, a preliminary validation was attempted for the proposed approach, and the numerical results of two- and three-dimensional problems were compared to those obtained through a dual-primal FETI approach. The results indicate an improvement in the performance as a result of the implementing the proposed approach.