• The KIPS Transactions:PartB
• /
• v.12B no.3 s.99
• /
• pp.343-348
• /
• 2005

The goal of this paper is to study the joint effect of factors of neural network teaming procedure. There are many factors, which may affect the generalization ability and teaming speed of neural networks, such as the initial values of weights, the learning rates, and the regularization coefficients. We will apply a constructive training algerian for neural network, then patterns are trained incrementally by considering them one by one. First, we will investigate the effect of these factors on generalization performance and learning speed. Based on these factors' effect, we will propose a joint method that simultaneously considers these three factors, and dynamically hue the learning rate and regularization coefficient. Then we will present the results of some experimental comparison among these kinds of methods in several simulated nonlinear data. Finally, we will draw conclusions and make plan for future work.

• Spatial Information Research
• /
• v.19 no.2
• /
• pp.29-36
• /
• 2011

This study performed quantitative assessment of the performance of adjustment models by a-priori analysis of the statistics of the surface parameter estimates against modeling factors. Lidar, airborne imagery, and SAR imagery have been used to acquire the earth surface elevation, where the shape properties of the surface need to be determined through neighboring observations around target location. In this study, parameters which are selected to be estimated are elevation, slope, second order coefficient. In this study, several factors which are needed to be specified to compose adjustment models are classified into three types: mathematical functions, kernel sizes, and weighting types. Accordingly, a-priori standard deviations of the parameters are computed for varying adjustment models. Then their corresponding confidence regions for both the standard deviation of the estimate and the estimate itself are calculated in association with probability distributions. Thereafter, the resulting confidence regions are compared to each other against the factors constituting the adjustment models and the quantitative performance of adjustment models are ascertained.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.11a
• /
• pp.493-496
• /
• 2004

본 연구에서는 신경망 학습의 중요한 평가 척도로써 고려될 수 있는 일반화 성능과 학습속도를 개선시키기 위한 방안으로써 초기 가중값과 학습률과 같은 주요 인자들을 이용한 신경망 학습 영향을 살펴본다. 특히 초기 가중값과 학습률을 고정시킨 후 새롭게 조정된 계수들을 점차적으로 변화시키는 새로운 인자 결합방법을 이용하여 신경망 학습량과 학습속도를 비교해 보고 계수조정을 통한 개선된 학습 영향을 살펴본다. 그리고 단순한 예제를 이용한 실증분석을 통하여 신경망 모형의 일반화 성능과 학습 속도 개선을 위한 각 인자들의 개별 효과와 결합 효과를 살펴보고 그 개선 방안을 제시한다.

• Journal of the Korean Geotechnical Society
• /
• v.38 no.2
• /
• pp.15-27
• /
• 2022

Despite the increasing applicability of rock grouting as a method for strengthening or disaster prevention by improving the stability of ground, criteria for planning parameters which can be used as minimum guideline are required since the current practice is mainly dependent on experience. In this study, the fundamental criteria for important parameters of rock grouting in terms of injection conditions such as water-cement ratio, injecting pressure, cement take and resulting effects such as deformation modulus and permeability are proposed. Those criteria are the results of analyses of a series of hydraulic fracturing tests and Lugeon tests, in-situ grouting tests at 17 sites in Korea and other countries, combined with the literature analyses of standards and previous research. In addition, the method for modifying proposed criteria according to water-cement ratio is also addressed since that in Korean practice is too high and therefore, should be adjusted to satisfy the conditions of balanced stable grouting. The results of this study can be used as a fundamental reference for more refined research in the future although they are still somewhat experience-dependent.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10b
• /
• pp.355-357
• /
• 2004

소프트웨어의 측정은 소프트웨어의 생명주기 초기에 분석되어 규모와 비용의 예측에 도움을 주어야 한다. 초기에는 정확한 LOC 기반 규모예측이 어려워 기능 점수 기반의 측정에 의하여 예측할 수 있다. 그러나 현재의 기능 점수 기반 측정은 모든 시스템에 획일화되어 있어 시스템의 특성을 반영할 수 없으며, IFPUG에서 제시한 일반 시스템 특성은 웹 기반 어플리케이션에서 사용자의 인지도가 매우 낮다. 본 논문에서는 웹 기반 어플리케이션 중에 동적 웹사이트에 특화된 시스템 특성 집합 및 영향도 측정 방법을 제시한다. 이 요소를 근거로 동적 웹사이트의 시스템 특성을 분석하여 기능 정수의 값 조정 인자에 반영하고 실제 시스템을 대상으로 조정된 기능 점수를 산정한다.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.12a
• /
• pp.213-216
• /
• 2000

퍼지제어기를 실제 플랜트에 적용할 경우, 모델링 오차와 플랜트에 대한 관련지식의 부족으로 만족할 만한 제어 결과를 기대하지 못한다. 이러한 경우 제어기 성능을 향상시키기 위해 제어인자를 다시 조정해야 하고, 이 조정과정은 시행착오방법으로 수행되기 때문에 많은 시간과 비용을 필요로 한다. 본 논문에서는 퍼지 논리를 사용하여 스케일링 계수를 자동조정하는 퍼지 레귤레이터를 제안하였다. 스케일 계수의 출력 범위와 오차, 오차 변화량 따라 퍼지 레귤레이터는 적절한 스케일링 계수값을 결정함으로써 퍼지제어기의 성능을 향상시키게 된다. 제안한 방법의 타당성을 확인하기 위해 2차 플랜트에 적용하여 모의실험을 수행하였다.

• Journal of Preventive Medicine and Public Health
• /
• v.29 no.2 s.53
• /
• pp.255-264
• /
• 1996

To study reliability and validity of PWI, this newly developed self-administered questionnaire which measures stress, was given to medical students. All respondent(133) were followed 4 weeks later and 92 were retested. Cronbach's $\alpha$ coefficient of data was 0.93. Test-retest reliability measured by Pearson's correlation coefficient was 0.72(P<0.01). Exploratory factor analysis(EFA) performed by principal axis factor method without iteration and by varimax rotation explored 13 principal components(eigenvalues > 1). After exploring 4 factor structure according to previous study results, factor 1 showed good agreement but other factors did not. Confirmatory factor analysis(CFA) showed poor fit of 4 factor model to data. In the further study, it may be considered to model that has unidimensional factor structure.

• The KIPS Transactions:PartD
• /
• v.13D no.6 s.109
• /
• pp.815-822
• /
• 2006

It is very important that they accurately predict the software development cost in the early stage of a software development. Because cost estimations are required when bidding for a contract or determining whether a project is feasible in terms of a cost-benefit analysis. The criterions of the software cost estimation was set up to calculate software development cost more exactly, which is applied to made up a budget of the software business or to calculate a suitable cost to start the business in our country. However, as the software technology and environment are changing very rapidly, it need to enhance the criterion of the cost estimation continuously. Therefore, we tried to apply technology of software and a variety of factors of environment changes in present. Most of all, we proposed an introduction and readjustment of the adjustment factor applying 14 general system characteristics to improve the accuracy of the cost estimation and the schedule adjustment factor that is required by practicians. For evaluating the accuracy in terms of the real data, we have used MMRE & PRED. In result, we proved that the accuracy was clearly improved by applying the scale factor and readjusted VAF with 14 general system characteristics. Moreover, we evaluated the accuracy of the schedule adjustment factor.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.4 s.316
• /
• pp.17-25
• /
• 2007

Multilevel inkjet printer employs multiple ink droplets with variable dot size and/or different concentrations intended to preserve high fidelity color reproduction and the appearance of continuous tone. A variety of research efforts on multitoning techniques has progressed toward better image quality. However, banding artifacts appear due to the same dot distributions near the printable output levels. This results in discontinuity and visually unpleasing output, especially at the smooth tone transition region. In this paper, to reduce the banding artifacts, a multitoning method to arrange ink distribution by controlling the blending proportion of adjacent output pixels based on an improved threshold scaling function is proposed. Ink distributions across the banding regions are changed according to two factors of the threshold scaling function because these factors handle the blending point of adjacent output pixel. Therefore, 8 observers, subjectively investigated ink distributions around the printable output levels for a set of the improved threshold scaling function. For a threshold scaling function with the specific factor values, we can achieve smoother visual transition. In the experiment, the proposed method showed a reduction of banding artifacts in both u-ay and color image and represented better Performance of color reproduction.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2007.05a
• /
• pp.1479-1483
• /
• 2007

낙동강상류에 위치한 제1지류인 반변천 유역에는 대규모 홍수조절을 위한 임하다목적댐이 위치해있다. 임하댐의 발전 또는 수문방류시 방류수의 원활한 흐름을 위해서 본댐하류에 조정지댐이 존재하게 되는 데 본댐 및 조정지댐에서는 운영자의 원활한 운영을 위해서 수위에 대한 조견표가 작성되어 있다. 대규모 홍수가 발생하는 경우 조정지댐에서는 수문을 완전개도하기 때문에 본댐의 방류량과 조정지댐의 방류량은 실시간으로 유사한 값을 나타내지만, 평수기의 경우는 발전방류가 특정시간에 이루어지는 관계로 이를 조절하면서 발생하는 조정지댐의 방류량은 본댐의 방류량과 다른 형태를 나타내게 된다. 따라서 조정지댐의 시간대별 방류량의 정확한 산정법은 하류 하천의 흐름분석에 중요한 인자가 될 수 있다. 본댐의 경우는 여수로를 통한 흐름이 대부분 자유흐름의 형태로 유하하기 때문에 상류수위의 영향에 따른 문비개도별 조견표가 작성되어 있지만, 조정지댐의 흐름은 특성에 따라서 자유흐름 또는 잠수흐름 그리고 완전개도의 흐름이 존재하게 된다. 따라서 잠수된 흐름이 발생하는 경우 합리적인 조정지댐의 방류량산정을 위해서 하류의 수위를 고려한 산정법을 고려할 필요가 있다. 임하조정지댐에서는 2004년 11월부터 댐하류 방수로의 계측이 시스템에 입력되고 있어서 현재 방수로의 수위를 실시간으로 취득할 수 있는 바, 이를 이용한 방류량의 산정방법을 검토하게 되었다.