• International Journal of Quality Innovation
• /
• 제3권2호
• /
• pp.13-28
• /
• 2002

Higher efficiency and effectiveness of Research & Development phases can be attained using advanced statistical methodologies. In this work statistical methodologies are combined with a deterministic approach to engineering design. In order to show the potentiality of such integration, a simple but effective example is presented. It concerns the problem of optimising the performances of a paper helicopter. The design of this simple device is not new in quality engineering literature and has been mainly used for educational purposes. Taking full advantage of fundamental engineering knowledge, an aerodynamic model is originally formulated in order to describe the flight of the helicopter. Screening experiments were necessary to get first estimates of model parameters. Subsequently, deterministic evaluations based on this model were necessary to set up further experimental phases needed to search (or a better design. Thanks to this integration of statistical and deterministic phases, a significant performance improvement is obtained. Moreover, the engineering knowledge かms out to be developed since an explanation of the “why” of better performances, although approximate, is achieved. The final design solution is robust in a broader sense, being both validated by experimental evidence and closely examined by engineering knowledge.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2001년도 제14회 신호처리 합동 학술대회 논문집
• /
• pp.231-234
• /
• 2001

In this paper, we present a robust watermarking technique based on a DCT-domain watermarking approach and an order statistic(OS) filter. The proposed technique inserts one watermark into each of four coefficients within a 2 ${\times}$ 2 block which is scanned on DCT coefficients in the zig-zag ordering from the medium frequency range. The detection algorithm uses an adaptive trimmed mean operator as a local estimator of the embedded watermark to obtain the desired robustness in the presence of additive Gaussian noise and JPEG compression attacks. The performance is analyzed through statistical analysis and numerical experiments. It is shown that the robustness properties against additive noise and JPEG compression attacks are more enhanced than the previous techniques.

• Journal of the Korean Statistical Society
• /
• 제35권4호
• /
• pp.419-439
• /
• 2006

The article develops a new class of distributions by introducing a nonnegative perturbing function to $t_\nu$ distribution having location and scale parameters. The class is obtained by using transformations and conditioning. The class strictly includes $t_\nu$ and $skew-t_\nu$ distributions. It provides yet other models useful for selection modeling and robustness analysis. Analytic forms of the densities are obtained and distributional properties are studied. These developments are followed by an easy method for estimating the distribution by using Markov chain Monte Carlo. It is shown that the method is straightforward to specify distribution ally and to implement computationally, with output readily adopted for constructing required criterion. The method is illustrated by using a simulation study.

• 응용통계연구
• /
• 제8권2호
• /
• pp.77-86
• /
• 1995

본 논문에서는 두 회귀직선의 평행성에 대한 로버스트 검정법을 제안하고, 모의실험과 예를 통하여 기존의 방법들과 유의수준의 안정성 및 검정력의 측면에서 비교하였다. 제안된 검정법은 Song et al. (1994b)에 의하여 제안된 최소절사제곱 추정량을 초기치로 하는 일단계 GM-추정량에 기초를 두고 있다. 이 추정량은 최대붕괴점과 유계영향함수를 갖는 것으로 알려져 있다.

• 응용통계연구
• /
• 제8권1호
• /
• pp.119-131
• /
• 1995

윈저화 자료에 기초한 분산분석법 개발의 일차시도로 고정효과 일원 분산분석 모형에 대한 윈저화 분산분석을 제시한다. 몬테 칼로 모의실험 기법을 사용하여 각 요인 수준마다 g-g 대칭 윈저화를 적용시켰을 때 윈저화 자료에 기초한 제곱합들의 비의 경험적 분포가 통상의 F 분포로 근사됨을 보인다. 이 근사 F 분포의 자유도는 윈저화 카이제곱 통계량의 경험적 분포가 자유도 (n-3g-1)의 통상적인 카이제곱 분포에 만족할만하게 근사되어진다는 성내경(1994)의 연구 성과를 토대로 결정된다. 여기서 n은 표본 크기, g는 한쪽 꼬리 부분에서 윈저화가 적용되는 양이다. 산출된 분산비의 경험적 분위수의 일부를 수록하였다. 이 연구는 non-adaptive 로버스트 분산분석법을 제안하는 것으로 이상점이 존재하는 분산분석 자료에 적용하면 자료 해석이 단순화되는 실용성을 위주로 한다.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1996년도 춘계공동학술대회논문집; 공군사관학교, 청주; 26-27 Apr. 1996
• /
• pp.15-18
• /
• 1996

Shewhart control chart is a basic technique to monitor the state of a process. We observe observations of a group of size four or five in a rational way and plot some statistics (e.g., means and ranges) on the chart. When setting up the control chart, the control limits are calculated based on preliminary 20-40 samples, which were supposedly obtained from stable operating conditions. But it may be hard to believe, especially at the beginning of constructing the chart for the first time, whether the process is stable and hence all samples were generated under the homogeneous operating conditions. In this report we suggest a mechanism to obtain robust control limits under self-criticism. When outliers are present in the sample, we obtain tighter control limits and hence increase the sensitivity of the chart. Examples will be given via simulation study.

• Communications for Statistical Applications and Methods
• /
• 제12권1호
• /
• pp.39-46
• /
• 2005

We propose an equivariant and robust estimator in multivariate regression model based on the least quartile difference (LQD) estimator in univariate regression. We call this estimator as the multivariate least quartile difference (MLQD) estimator. The MLQD estimator considers correlations among response variables and it can be shown that the proposed estimator has the appropriate equivariance properties defined in multivariate regressions. The MLQD estimator has high breakdown point as does the univariate LQD estimator. We develop an algorithm for MLQD estimate. Simulations are performed to compare the efficiencies of MLQD estimate with coordinatewise LQD estimate and the multivariate least trimmed squares estimate.

• Communications for Statistical Applications and Methods
• /
• 제10권3호
• /
• pp.1037-1046
• /
• 2003

We propose an equivariant and robust estimator in multivariate regression model based on the least trimmed squares (LTS) estimator in univariate regression. We call this estimator as multivariate least trimmed squares (MLTS) estimator. The MLTS estimator considers correlations among response variables and it can be shown that the proposed estimator has the appropriate equivariance properties defined in multivariate regression. The MLTS estimator has high breakdown point as does LTS estimator in univariate case. We develop an algorithm for MLTS estimate. Simulation are performed to compare the efficiencies of MLTS estimate with coordinatewise LTS estimate and a numerical example is given to illustrate the effectiveness of MLTS estimate in multivariate regression.

• 응용통계연구
• /
• 제6권1호
• /
• pp.29-40
• /
• 1993

모델을 이용한 표본론에서는 오차에 대한 함수식이 불확실한 경우가 종종 발생되는데 이러 한 오차에 대한 지식이 결여 되었을 때 발생되는 잘못된 효과를 줄일 수 있는 방법이 연구 되었다. 제시된 표본방법론은 모든 가능한 오차함수식에 대한 비효율성에 대한 평균을 최소 화하는데 그 목적이 있다. 컴퓨터를 이용한 알고리즘이 제시되었고 회계감사에 관련된 특수 한 경우의 예를 들어 이러한 방법의 효율성을 알아 보았다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1993년도 정기총회 및 추계학술대회 논문집 학회본부
• /
• pp.287-289
• /
• 1993

In this paper, we improve feedforward neural network performance by eliminating the effect of gross error using classifier structure. At first, we prove the output of classifier converges to the posteriori probability of each pattern given input x, $f_0({\theta}_1|x)$. And we apply filtering approach based on the robust statistics before reconstructing continuous output. The data distorted with noise can be rejected by this process. Finally, we suggest neurofilter structure. Simulation result shows that our structure yields consistent estimates even in the presence of noise.