• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.1173-1176
• /
• 1996

Multilayer feedforward networks may be applied to identify the deterministic relationship between input and output data. When the results from the network require a high level of assurance, considering of the stochastic relationship between the data may be very important. The variance is one of the useful parameters to represent the stochastic relationship. This paper presents a new algorithm for a multilayer feedforward network to learn the variance of dispersed data without preliminary calculation of variance. In this paper, the network with this learning algorithm is named as a variance learning neural network(VALEAN). Computer simulation examples are utilized for the demonstration and the evaluation of VALEAN.

• Journal of the Korean Data and Information Science Society
• /
• 제16권2호
• /
• pp.289-300
• /
• 2005

In this article quick variance change point (VCP) detection problem for time series is considered. For this variance VCP detector equipped with tuning parameters is proposed. A major tool for the detector is moving variance ratio (MVR) which monitors variance change of a given time series. Tuning process of detector is investigated via simulation, which shows that tuning parameters are critical in achieving sensitivity and adaptiveness of detector.

• 한국정밀공학회지
• /
• 제14권6호
• /
• pp.121-127
• /
• 1997

Multilayer feedforward networks may be applied to identify the deterministic relationship between input and output data. When the results from the network require a high level of assurance, consideration of the stochastic relationship between the input and output data may be very important. Variance is one of the effective parameters to deal with the stochastic relationship. This paper presents a new algroithm for a multilayer feedforward network to learn the variance of dispersed data without preliminary calculation of variance. In this paper, the network with this learning algorithm is named as a variance learning neural network(VALEAN). Computer simulation examples are utilized for the demonstration and the evaluation of VALEAN.

• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2000년도 추계학술발표회 논문집
• /
• pp.23-26
• /
• 2000

It is noted that many econometric time series have long-memory properties. A long-memory process, or strongly dependent process, is characterized by hyperbolic decaying autocorrelations and unbounded spectral density at the origin. Since the long-memory property can be observed by data obtained from rather a long period, there is some possibility of parameter change in the process. In this paper, we consider the estimation of change-point when there is a change in the variance of a long-memory process. The estimator is based on some reasonable statistic and the consistency is shown using Taqqu's strong reduction theorem

• Journal of the Korean Data and Information Science Society
• /
• 제20권1호
• /
• pp.11-24
• /
• 2009

수정절차에서 공정수정기는 잡음이 존재하지만 제거할 수 없을 때 공정수준을 목표치에 가깝게 수정하는데 종종 유용하게 사용된다. 강건 수정기의 예로는 단일 및 이중 지수가중이동평균 수정기가 있다. 이중 지수가중이동평균 수정기는 단일 지수가중이동평균 수정기가 제거할 수 없는 공정편차의 치우침을 줄일 수 있도록 고안되었다. 이 논문에서는 이 두 가지 수정기가 적용될 때 과도하게 커질 수 있는 수정량분산을 줄일 수 있도록 원래의 수정기에 지수가중이동평균을 적용함으로써 수정되었다. 주어지 수정기에 대한 지수가중이동평균 수정은 편차제곱평균은 조금 증가시키지만, 수정량분산을 줄이는데 성공적임을 보이고 있다.

• 한국품질경영학회:학술대회논문집
• /
• 한국품질경영학회 2007년도 춘계학술대회
• /
• pp.302-309
• /
• 2007

To conduct statistical process control needs the assumption that the process data are independent. However, most of chemical processes, like a semi-conduct processes do not satisfy the assumption because of autocorrelation. It causes abnormal out of control signal in the process control and misleading process capability. In this study, we introduce that Shore's method to solve the problem and to find the optimal subgroup size to estimate variance for AR(l) model. Especially, we focus on finding an actual subgroup size for small samples using simulation. It may be very useful for statistical process control to analyze process capability and to make a Shewhart chart properly.

• 품질경영학회지
• /
• 제26권1호
• /
• pp.61-73
• /
• 1998

The quality control literature contains a substantial number of articles concerned with how to optimally choose control limits in order to minimize production cost. The purpose of the this study is to determine the economic setting for the process mean of an industrial process. In this study it is assumed that the lower control limit is set by government regulations and the u, pp.r limit and process mean are chosen based on economic considerations. Much research has been conducted on this problem under the condition of the fixed rpm(rate per minute). However a variance can be increased in proportion to the level of rpm and the increase of the variance can change the optimal process mean. Therefore, it is desirable to determine both the process mean and the level of rpm simultaneously. In this paper, a mathematical model is presented which considers the u, pp.r limit and the rpm as variables.

• Communications for Statistical Applications and Methods
• /
• 제3권1호
• /
• pp.257-265
• /
• 1996

We consider nonstationary autoregressive autoregressive process with infinite variance of error. In the case of infinite cariance, the limiting distribution of the estimated coefficient is different from that under the finite cariance assumption. In this paper we show that the bootstrap method can be used to approximate the distribution of ordinary least squares estimator of the coefficient in the first order random walk process with infinite variance through some empirical studies and we suggest a test procedure based on bootstrap method for the unit root test.

• Journal of the Korean Statistical Society
• /
• 제24권2호
• /
• pp.507-518
• /
• 1995

The ordinary least squares based estiamte $S^2$ of the disturbance variance is considered in the linear regression model when the disturbances follow the first-order moving-average process. It is shown that $S^2$ is weakly consistent estimate for the disturbance varaince without any restriction on the regressor matrix X. Also, simple exact bounds on the relative bias of $S^2$ are given in finite sample sizes.

• 대한산업공학회지
• /
• 제39권5호
• /
• pp.361-366
• /
• 2013

Prediction of process parameters is very important in parameter design. If predictions are fairly accurate, the quality improvement process will be useful to save time and reduce cost. The concept of dual response approach based on response surface methodology has widely been investigated. Dual response approach may take advantages of optimization modeling for finding optimum setting of input factor by separately modeling mean and variance responses. This study proposes an alternative dual response approach based on machine learning techniques instead of statistical analysis tools. A hybrid neural network-genetic algorithm has been proposed for the purpose of parameter design. A neural network is first constructed to model the relationship between responses and input factors. Mean and variance responses correspond to output nodes while input factors are used for input nodes. Using empirical process data, process parameters can be predicted without performing real experimentations. A genetic algorithm is then applied to find the optimum settings of input factors, where the neural network is used to evaluate the mean and variance response. A drug formulation example from pharmaceutical industry has been studied to demonstrate the procedures and applicability of the proposed approach.