• International Journal of Contents
• /
• v.16 no.2
• /
• pp.8-18
• /
• 2020

Data is critical in deep learning but the scarcity of data often occurs in research, especially in the preparation of the paired training data. In this paper, document image binarization with unpaired data is studied by introducing adversarial learning, excluding the need for supervised or labeled datasets. However, the simple extension of the previous unpaired training to binarization inevitably leads to poor performance compared to paired data training. Thus, a new deep learning approach is proposed by introducing a multi-diversity of higher quality generated images. In this paper, a two-stage model is proposed that comprises the generative adversarial network (GAN) followed by the U-net network. In the first stage, the GAN uses the unpaired image data to create paired image data. With the second stage, the generated paired image data are passed through the U-net network for binarization. Thus, the trained U-net becomes the binarization model during the testing. The proposed model has been evaluated over the publicly available DIBCO dataset and it outperforms other techniques on unpaired training data. The paper shows the potential of using unpaired data for binarization, for the first time in the literature, which can be further improved to replace paired data training for binarization in the future.

• Proceedings of the KSRS Conference
• /
• v.1
• /
• pp.512-515
• /
• 2006

Forest stand height and volume are important indicators for management purpose as well as for the environmental analysis. Shuttle Radar Topography Mission (SRTM) is backscattered over forest canopy and DSM can be acquired from such scattering characteristic, while National Elevation Dataset (NED) provides bare earth elevation data. The difference between SRTM and NED is estimated as tree height, and it is correlated with forest parameters, it is correlated with forest parameters, including average DBH, Trees per acre, net BF per acre, and total Net MBF. Especially, among them, net Board Foot(BF) per acre is the index that well represents forest volume. The Project site was Douglas-fir dominating plantation area in the western Washington an the northern Oregon in the U.S. This study shows a relationship of high correlation between the forest parameters and the product from SRTM, NED, and ETM+. This research performs multi regression analysis and regression tree algorithm, and can get more improved relationship between several parameters.

• Communications for Statistical Applications and Methods
• /
• v.27 no.5
• /
• pp.535-546
• /
• 2020

In genetic association studies, pleiotropy is a phenomenon where a variant or a genetic region affects multiple traits or diseases. There have been many studies identifying cross-phenotype genetic associations. But, most of statistical approaches for detection of pleiotropy are based on individual tests where a single variant association with multiple traits is tested one at a time. These approaches fail to account for relations among correlated variants. Recently, multivariate regularization methods have been proposed to detect pleiotropy in analysis of high-dimensional genomic data. However, they suffer a problem of tuning parameter selection, which often results in either too many false positives or too small true positives. In this article, we applied selection probability to multivariate regularization methods in order to identify pleiotropic variants associated with multiple phenotypes. Selection probability was applied to individual elastic-net, unified elastic-net and multi-response elastic-net regularization methods. In simulation studies, selection performance of three multivariate regularization methods was evaluated when the total number of phenotypes, the number of phenotypes associated with a variant, and correlations among phenotypes are different. We also applied the regularization methods to a wild bean dataset consisting of 169,028 variants and 17 phenotypes.

• Korean Journal of Remote Sensing
• /
• v.22 no.5
• /
• pp.427-431
• /
• 2006

Forest stand height and volume are important indicators for management purpose as well as for the environmental analysis. Shuttle Radar Topography Mission (SRTM) is backscattered over forest canopy and DSM can be acquired from such scattering characteristic, while National Elevation Dataset (NED) provides bare earth elevation data. The difference between SRTM and NED is estimated as tree height, and it is correlated with forest parameters, it is correlated with forest parameters, including average DBH, Trees per acre, net BF per acre, and total Net MBF. Especially, among them, net Board Foot(BF) per acre is the index that well represents forest volume. The Project site was Douglas-fir dominating plantation area in the western Washington an the northern Oregon in the U.S. This study shows a relationship of high correlation between the forest parameters and the product from SRTM, NED, and ETM+. This research performs multi regression analysis and regression tree algorithm, and can get more improved relationship between several parameters.

• Asian Journal of Innovation and Policy
• /
• v.4 no.3
• /
• pp.271-287
• /
• 2015

Universities are now operating in a post-globalized world. They need to be locally engaged while globally visible for continued success. Specific strategies include 1) unbundling education using fractal modules approach to facilitate more flexible and customized learning expectations of net generation of students; 2) open-source software and hardware platforms to facilitate collaborations around a short or multi-year, multidisciplinary brain stretching projects motivated by the real world challenges; 3) new courses on innovation and entrepreneurship to nurture experience seeking and enterprising mindset of students; 4) cross-border university- public- private partnerships for developing solutions specific to the local needs and can be scalable for the world; and 5) encouraging faculty members with glocal mindset.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.06d
• /
• pp.40-43
• /
• 2007

보안운영체제에서는 등급기반 사용자, 프로세스, 파일에 대한 영역분리 및 등급별 보안관리를 하는 다중등급보안(Multi Level Security)을 구현하고 있다. 안전한 운영체제에서는 사용자가 등급 즉, 자신의 보안등급과 보호범주를 설정하기 위해 권한 인증절차를 수행한다. 권한은 보안운영체제에서 강제적 접근 제어(Mandatory Access Control)의 기본이 되므로 그 보안에 중요성이 크다. 따라서, 권한 인증 절차의 보안이 부각되고 있다. 본 논문에서는 PKI 기반 전자서명 인증절차를 이용하여 신원 확인과 권한 인증을 한번에 수행할 수 있는 방법을 제시한다.

• Journal of Electrical Engineering and Technology
• /
• v.9 no.2
• /
• pp.441-451
• /
• 2014

This article addresses an efficient heuristic-based approach to assign static shunt capacitors along radial distribution networks using the artificial bee colony algorithm. The objective function is adapted to enhance the overall system static voltage stability index and to achieve maximum net yearly savings. Load variations have been considered to optimally scope the fixed and switched capacitors required. The numerical results are compared with those obtained using recent heuristic methods and show that the proposed approach is capable of generating high-grade solutions and validated viability.

• Journal of Biomedical Engineering Research
• /
• v.22 no.1
• /
• pp.19-27
• /
• 2001

본 연구에서는 20대 남자 대학생 45명에게 반복 연산 스트레스를 유발시키기 위하여 세 단계의 난이도를 갖는 덧셈 연산을 수행하게 하였으며, 이때 각 피검자들로부터 이에 대한 생체 신호를 측정하였다. 측정된 생체 신호로부터 제시된 연산 스트레스에 대한 감성을 평가하기 위하여 7개의 생체 파라미터를 사용하였고, 비선형 특성을 갖는 연산 스트레스에 대한 감성을 평가하기 위하여 세 단계의 구조를 갖는 감성 평가 시스템을 구성하였다. 또한 감성평가 시스템의 성능을 비교하기 위하여 평가 시스템의 각 단계를 선형 판별 알고리즘인 Least Mean Square Algorithm을 이용한 경우와 비선형 판별 알고리즘인 Radial-Based Functional-link Net을 이용한 경우를 사용하였다. 각 감성 평가 시스템은 Cross Validation을 사용하여 성능을 비교하였으며, 전체 감성 평가 시스템에서의 연산 스트레스에 대한 감성 평가 정확도는 선형 알고리즘을 이용할 경우 63.02%, RBFLN을 이용한 경우는 83.07%를 얻었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.04a
• /
• pp.77-82
• /
• 2000

본 논문에서는 워크플로우 모델링 도구인 ICN(Information Control Net) 모델을 이용하여 워크플로우 역할 의존성 분석 메커니즘을 제안하였다. 즉, ICN 모델로 정의된 워크플로우의 각 액티비티(activity)들 간에 존재하는 역할 의존 관계를 표현하기 위하여 역할 의존 넷(Role Dependency Net)을 정의하였고, ICN 모델로부터 역할 의존 넷을 생성하는 알고리즘을 제안하였다. 본 논문에서 제시된 알고리즘을 이용하여 생성된 역할 기반의 워크플로우 모델은 any-cast 워크플로우와, multi-cast 워크플로우 작업환경을 제공함으로써 현재 대부분 조직에서의 작업 환경인 객체지향 작업 환경 및 분산 작업 환경에서 워크플로우 관리 시스템을 구축할 수 있다.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.41.3-41
• /
• 2001

In this paper, we present a decentralized adaptive neural net(NN) controller for the transient stability and voltage regulation of a multimachine power system. First, an adaptively input-output linearizing controller using NN is designed to eliminate the nonlinearities and interactions between generators. Then, a robust control term which bounds terminal voltage to a neighborhood of the operating point within the desired value is introduced using only local information. In addition, we consider input saturation which exists in the SCR amplifier and prove that the stability of the overall closed-loop system is maintained regardless of the input saturation. The design procedure is tested on a two machine infinite bus power system.