• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2009년도 IWAIT
• /
• pp.282-285
• /
• 2009

This paper proposes a flame verification algorithm using motion and spatial persistency. Most previous vision-based methods using color information and temporal variations of pixels produce frequent false alarms due to the use of many heuristic features. To solve these problems, we used a Bayesian Networks. In addition, since the shape of flame changes upwards irregularly due to the airflow caused by wind or burning material, we distinct real flame from moving objects by checking the motion orientation and temporal persistency of flame regions to remove the misclassification. As a result, the use of two verification steps and a Bayesian inference improved the detection performance and reduced the missing rate.

• Communications for Statistical Applications and Methods
• /
• 제4권3호
• /
• pp.725-734
• /
• 1997

많은 사회과학 조사에서 분할표 형태로 얻어진 범주형 자료에는 오분류(misclassification)로 인한 오차가 내재되는 경우가 종종 있다. 질적속성 추정을 위한 확률화응답은 이러한 오분류 문제의 한 특수한 경우로 여겨지기도 한다. 그래서 확률화응답을 통한 범주형자료는 혼합된 분할표(mixed-up contingency table)로 여길 수 있는 바, 본 논문에서는 이에 대해 대수선형모형(log-linear model)을 설정하고 Chen과 Fienberg(1976)의 Iterative scaling procedure(ISP)에 의하여 얻어진 최우추정량의 극한을 이용하였다. 이 결과 Warner(1965) 형태의 대칭기법에 대해서는 Singh(1976)에 의하여 제안된 최우추정량과 같아지게 됨을 보임으로써 Warner에 의해서 제시된 추정량이 최우추정량으로 적절하지 않음을 확인해 보고, 무관질문기법에 대해서는 Greenberg, et al.(1969)에 의해서 제안된 추정량이 추정의 관점에서 최우추정량으로 적절하지 않음을 알아 보았다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권10호
• /
• pp.4007-4024
• /
• 2020

It is tremendously important to construct decision trees to use as a tool for knowledge representation from a given decision table. However, the usual algorithms may split the decision table based on each value, which is not efficient for numerical attributes. The methodology of this paper is to split the given decision table into binary groups as like the CART algorithm, that uses binary split to work for both categorical and numerical attributes. The difference is that it uses split for each attribute established by the directed acyclic graph in a dynamic programming fashion whereas, the CART uses binary split among all considered attributes in a greedy fashion. The aim of this paper is to study the effect of binary splits in comparison with each value splits when building the decision trees. Such effect can be studied by comparing the number of nodes, local and global misclassification rate among the constructed decision trees based on three proposed algorithms.

• 말소리와 음성과학
• /
• 제10권2호
• /
• pp.7-14
• /
• 2018

The paper investigates the acoustic characteristics of English fricatives in the TIMIT corpus, with a special focus on the role of voicing in rendering fricatives in American English. The TIMIT database includes 630 talkers and 2,342 different sentences, and comprises more than five hours of speech. Acoustic analyses are conducted in the domain of spectral and temporal properties by treating gender, voicing, and place of articulation as independent factors. The results of the acoustic analyses revealed that acoustic signals interact in a complex way to signal the gender, place, and voicing of fricatives. Classification experiments using a multiclass support vector machine (SVM) revealed that 78.7% of fricatives are correctly classified. The majority of errors stem from the misclassification of /θ/ as [f] and /ʒ/ as [z]. The average accuracy of gender classification is 78.7%. Most errors result from the classification of female speakers as male speakers. The paper contributes to the understanding of the effects of voicing and gender on fricatives in a large-scale speech corpus.

• Journal of the Korean Data and Information Science Society
• /
• 제24권1호
• /
• pp.189-200
• /
• 2013

In this paper, it will be assumed that there are two distinct populations which are multivariate normal with equal covariance matrix. We also assume that the two populations are equally likely and the costs of misclassification are equal. The classification rule depends on the situation whether the training samples include missing values or not. We consider the conditional bootstrap confidence intervals for classification error rate when a block of observation is missing.

• 대한산업공학회지
• /
• 제26권3호
• /
• pp.200-205
• /
• 2000

An economic process monitoring procedure is presented using a surrogate variable for the case where performance variable is dichotomous. Every item is inspected with a surrogate variable and determined whether it should be accepted or rejected. When an item is rejected, the previous number of consecutively accepted items is compared with a predetermined number r to decide whether there is a shift in fraction nonconforming or not. The conditional distribution of the surrogate variable given the performance variable is assumed to be normal. A cost model is constructed which includes costs of inspection, misclassification, illegal signal, undetected out-of-control state, and correction. Methods of finding the optimum number r and screening limit are provided. Numerical studies on the effects of cost coefficients are also performed.

• 대한예방의학회:학술대회논문집
• /
• 대한예방의학회 1994년도 교수 연수회(역학)
• /
• pp.129-136
• /
• 1994

A model was developed for a simple clinical trial in which graders had defined probabilities of misclassifying pathologic material to disease present or absent. The authors compared Kappa between graders, and efficiency and bias in the clinical trial in the presence of misclassification. Though related to bias and efficiency, Kappa did not predict these two statistics well. These results pertain generally to evaluation of systems for encoding medical information, and the relevance of Kappa in determining whether such systems are ready for use in comparative studies. The authors conclude that, by itself, Kappa is not informative Enough to evaluate the appropriateness of a grading scheme for comparative studies. Additional, and perhaps difficult, questions must be addressed for such evaluation.

• 한국지능정보시스템학회:학술대회논문집
• /
• 한국지능정보시스템학회 2003년도 춘계학술대회
• /
• pp.206-213
• /
• 2003

Application of existing decision tree algorithms for Internet retail customer classification is apt to construct a bushy tree due to imprecise source data. Even excessive analysis may not guarantee the effectiveness of the business although the results are derived from fully detailed segments. Thus, it is necessary to determine the appropriate number of segments with a certain level of abstraction. In this study, we developed a stopping rule that considers the total amount of information gained while generating a rule tree. In addition to forwarding from root to intermediate nodes with a certain level of abstraction, the decision tree is investigated by the backtracking pruning method with misclassification loss information.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2003년도 학술회의 논문집 정보 및 제어부문 B
• /
• pp.604-607
• /
• 2003

Most recommendation systems recommend the products or other information satisfying preferences of users on the basis of the users' previous profile information and other information related to product searches and purchase of users visiting web sites. This study aims to apply these application categories to e-mail more necessary to users. The E-Mail System has the strong personality so that there will be some problems even if e-mails are automatically classified by category through the learning on the basis of the personal rules. In consideration with this aspect, we need the semi-automatic system enabling both automatic classification and recommendation method to enhance the satisfaction of users. Accordingly, this paper uses two approaches as the solution against the misclassification that the users consider as the accuracy of classification itself using the dynamic threshold in Bayesian Learning Algorithm and the second one is the methodological approach using the recommendation agent enabling the users to make the final decision.

• Communications for Statistical Applications and Methods
• /
• 제27권2호
• /
• pp.177-187
• /
• 2020

This article is concerned with the issue of forecasting and evaluation of threshold-asymmetric volatility models for time series of count data. In particular, threshold integer-valued models with conditional Poisson and conditional negative binomial distributions are highlighted. Based on the parametric bootstrap method, some evaluation measures are discussed in terms of one-step ahead forecasting. A parametric bootstrap procedure is explained from which directional measure, magnitude measure and expected cost of misclassification are discussed to evaluate competing models. The cholera data in Bangladesh from 1988 to 2016 is analyzed as a real application.