• Proceedings of the Korean Information Science Society Conference
• /
• 2001.10b
• /
• pp.307-309
• /
• 2001

분류문제에서 유용한 학습패턴은 클래스들간의 분류경계에 근접한 정상패턴들을 말한다. 본 연구에서는 다양한 구조와 학습 파라미터를 가진 신경망 앙상블을 구성하고 그 출력값의 편기와 분산에 기초한 패턴절수를 정의한다. 전체 학습패턴 중 일정한 임계값 이상의 패턴점수를 가진 패턴들만이 학습패턴으로 선정된다. 제안한 방법은 두 개의 인공문제와 두 개의 실제문제 (UCI Repository)에 적응, 검증되었다. 그 결과 선택된 패턴만으로 학습한 경우, 메모리 공간 절약 및 계산시간 단축의 효과뿐만 아니라 복잡도가 큰 모델이라도 과적합을 하지 않았고 실험적으로 안정된 결과를 산출했으며, 적은 수의 학습패턴만으로도 일반화 성능을 향상시키거나 적어도 저하시키지 않았다는 것을 보였다.

• Journal of Labour Economics
• /
• v.35 no.2
• /
• pp.117-146
• /
• 2012

This study demonstrates the gender difference in the factors that affect job changes and the resulting wage changes in the recent Korean labor market. By using the KEAPS (2003-2007), we found that male workers uniquely tend to stay longer at their current jobs when they have families to support. After controlling self-selection bias, we also found that wage changes resulting from switching jobs differ between male and female workers during this studied period.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.1
• /
• pp.217-227
• /
• 2015

In observational study, handling confounders is a primary issue in measuring treatment effect of interest. Historically, a regression with covariate adjustment (covariate-adjusted regression) has been the typical approach to estimate treatment effect incorporating potential confounders into model. However, ever since the introduction of the propensity score, covariate-adjusted regression has been gradually replaced in medical literatures with various balancing methods based on propensity score. On the other hand, there is only a paucity of researches assessing propensity score methods compared with the covariate-adjusted regression. This paper examined the performance of propensity score methods in estimating risk difference and compare their performance with the covariate-adjusted regression by a Monte Carlo study. The study demonstrated in general the covariate-adjusted regression with variable selection procedure outperformed propensity-score-based methods in terms both of bias and MSE, suggesting that the classical regression method needs to be considered, rather than the propensity score methods, if a performance is a primary concern.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2003.07a
• /
• pp.581-584
• /
• 2003

Most modeling about dynamic behavior of Antiferroelectric Liquid Crystal (AFLC) is limited to the hysteric characteristics of AFLC cells or thresholdless switching of frustrated AFLC cells. In this paper, domain growth of AFLC cells is modeled with extended bilayer model. When driving pulses that consist of a selection voltage, a bias voltage, and a reset voltage are applied to the AFLC cell, its dynamic behavior is simulated.

• Proceedings of the IEEK Conference
• /
• 2000.11b
• /
• pp.55-58
• /
• 2000

This paper describes a single transistor type ferroelectric field effect transistor (1T FeFET) memory celt scheme which can select one unit memory cell and program/read it. To solve the selection problem of 1T FeEET memory cell array, the row direction common well is electrically isolated from different adjacent row direction column. So, we can control voltage of common well line. By applying bias voltage to Gate and Well, respectively, we can implant IT FeEET memory cell scheme which no interface problem and can bit operation. The results of HSPICE simulations showed the successful operations of the proposed cell scheme.

• Publications of The Korean Astronomical Society
• /
• v.25 no.3
• /
• pp.77-82
• /
• 2010

Recent results from large surveys of the local universe show that the galaxy-black hole connection is linked to host morphology at a fundamental level and that there are two fundamentally different modes of black hole growth. The fraction of early-type galaxies with actively growing black holes, and therefore the AGN duty cycle, declines significantly with increasing black hole mass. Late-type galaxies exhibit the opposite trend: the fraction of actively growing black holes increases with black hole mass. Issues of AGN selection bias and prospects for near-future efforts with high redshift data are discussed.

• Proceedings of the IEEK Conference
• /
• 2000.07a
• /
• pp.403-405
• /
• 2000

This paper describes a single transistor type ferroelectric field effect transistor (1Tr FeFET) memory cell scheme, which select one unit memory cell and program/read it. The well voltage can be controlled by isolating the common row well lines. Through applying bias voltage to Gate and Well, respectively, we implement If FeFET memory cell scheme in which interference problem is not generated and the selection of each memory cell is possible. The results of HSPICE simulations showed the successful operations of the proposed cell scheme.

• Survey Research
• /
• v.1 no.1
• /
• pp.125-134
• /
• 2000

It is very important to keep scientific principles in marketing research, Especially in sampling we have to select the scheme to avoid self-selection bias Internet index is ready-made-research service produced by analyzing the log data transferred from PcMeter installed in panelists PC with panelists profile and site data base following the scientific principles Through the index we can figure out the actual behaviour of Korean netizen surfing various web sites and understand their cybergraphics.

• ETRI Journal
• /
• v.31 no.2
• /
• pp.121-128
• /
• 2009

In this paper, we describe a fixed-threshold sequential minimal optimization (FSMO) for structured SVM problems. FSMO is conceptually simple, easy to implement, and faster than the standard support vector machine (SVM) training algorithms for structured SVM problems. Because FSMO uses the fact that the formulation of structured SVM has no bias (that is, the threshold b is fixed at zero), FSMO breaks down the quadratic programming (QP) problems of structured SVM into a series of smallest QP problems, each involving only one variable. By involving only one variable, FSMO is advantageous in that each QP sub-problem does not need subset selection. For the various test sets, FSMO is as accurate as an existing structured SVM implementation (SVM-Struct) but is much faster on large data sets. The training time of FSMO empirically scales between O(n) and O($n^{1.2}$), while SVM-Struct scales between O($n^{1.5}$) and O($n^{1.8}$).

• The Korean Journal of Applied Statistics
• /
• v.11 no.2
• /
• pp.363-376
• /
• 1998

The linear smoothing spline estimator is modified to remove boundary bias effects. The resulting estimator can be calculated efficiently using an O(n) algorithm that is developed for the computation of fitted values and associated smoothing parameter selection criteria. The asymptotic properties of the estimator are studied for the case of a uniform design. In this case the mean squared error properties of boundary corrected linear smoothing splines are seen to be asymptotically competitive with those for standard second order kernel smoothers.