• The Journal of Asian Finance, Economics and Business
• /
• 제9권2호
• /
• pp.37-47
• /
• 2022

This study uses monthly data from January 2009 to December 2020 to examine the effectiveness of foreign currency intervention and its influence on monetary policy in Vietnam using a Hierarchical Bayesian VAR model. The findings suggest that foreign exchange intervention has little influence on the exchange rate level or exports, but it can significantly minimize exchange rate volatility. As a result, we can demonstrate that the claim that Vietnam is a currency manipulator is false. As well, the forecast error variance decomposition results reveal that interest rate differentials mainly determine the exchange rate level instead of foreign exchange intervention. Moreover, the findings suggest that foreign exchange intervention is not effectively sterilized in Vietnam. Inflation is caused by an increase in international reserves, which leads to an expansion of the money supply and a decrease in interest rates. Although the impact of foreign exchange intervention grows in tandem with the growth of international reserves, if the sterilizing capacity does not improve, rising foreign exchange intervention will instead result in inflation. Finally, we use a rolling window approach to examine the time-varying effect of foreign exchange intervention.

• Fisheries and Aquatic Sciences
• /
• 제24권4호
• /
• pp.139-152
• /
• 2021

The main purpose of this study is to fit catch-per-unit-effort (CPUE) data about Korea chub mackerel (Scomber japonicus) stock with a state-space production (SSP) model, and to provide stock assessment results. We chose a surplus production model for the chub mackerel data, namely annual yield and CPUE. Then we employed a state-space layer for a production model to consider two sources of variability arising from unmodelled factors (process error) and noise in the data (observation error). We implemented the model via script software ADMB-RE because it reduces the computational cost of high-dimensional integration and provides Markov Chain Monte Carlo sampling, which is required for Bayesian approaches. To stabilize the numerical optimization, we considered prior distributions for model parameters. Applying the SSP model to data collected from commercial fisheries from 1999 to 2017, we estimated model parameters and management references, as well as uncertainties for the estimates. We also applied various production models and showed parameter estimates and goodness of fit statistics to compare the model performance. This study presents two significant findings. First, we concluded that the stock has been overexploited in terms of harvest rate from 1999 to 2017. Second, we suggest a SSP model for the smallest goodness of fit statistics among several production models, especially for fitting CPUE data with fluctuations.

• Genomics & Informatics
• /
• 제18권4호
• /
• pp.45.1-45.5
• /
• 2020

With the ongoing rise of coronavirus disease 2019 (COVID-19) pandemic across the globe, interests in COVID-19 antibody testing, also known as a serology test has grown, as a way to measure how far the infection has spread in the population and to identify individuals who may be immune. Recently, many countries reported their population based antibody titer study results. South Korea recently reported their third antibody formation rate, where it divided the study between the general population and the young male youths in their early twenties. As previously stated, these simple point estimates may be misinterpreted without proper estimation of standard error and confidence intervals. In this article, we provide an updated 95% confidence intervals for COVID-19 antibody formation rate for the Korean population using asymptotic, exact and Bayesian statistical estimation methods. As before, we found that the Wald method gives the narrowest interval among all asymptotic methods whereas mid p-value gives the narrowest among all exact methods and Jeffrey's method gives the narrowest from Bayesian method. The most conservative 95% confidence interval estimation shows that as of 00:00 November 23, 2020, at least 69,524 people were infected but not confirmed. It also shows that more positive cases were found among the young male in their twenties (0.22%), three times that of the general public (0.051%). This thereby calls for the quarantine authorities' need to strengthen quarantine managements for the early twenties in order to find the hidden infected people in the population.

• Genomics & Informatics
• /
• 제18권3호
• /
• pp.31.1-31.8
• /
• 2020

The coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has become a global pandemic. No specific therapeutic agents or vaccines for COVID-19 are available, though several antiviral drugs, are under investigation as treatment agents for COVID-19. The use of convalescent plasma transfusion that contain neutralizing antibodies for COVID-19 has become the major focus. This requires mass screening of populations for these antibodies. While several countries started reporting population based antibody rate, its simple point estimate may be misinterpreted without proper estimation of standard error and confidence intervals. In this paper, we review the importance of antibody studies and present the 95% confidence intervals COVID-19 antibody rate for the Korean population using two recently performed antibody tests in Korea. Due to the sparsity of data, the estimation of confidence interval is a big challenge. Thus, we consider several confidence intervals using Asymptotic, Exact and Bayesian estimation methods. In this article, we found that the Wald method gives the narrowest interval among all Asymptotic methods whereas mid p-value gives the narrowest among all Exact methods and Jeffrey's method gives the narrowest from Bayesian method. The most conservative 95% confidence interval estimation shows that as of 00:00 on September 15, 2020, at least 32,602 people were infected but not confirmed in Korea.

• 경영과학
• /
• 제11권2호
• /
• pp.1-27
• /
• 1994

A reliability data processing MPRDP (Multi-Purpose Reliability Data Processor) has been developed in FORTRAN language since Jan. 1992 at KAERI (Korean Atomic Energy Research Institute). The purpose of the research is to construct a reliability database(plant-specific as well as generic) by processing various kinds of reliability data in most objective and systematic fashion. To account for generic estimates in various compendia as well as generic plants' operating experience, we developed a 'three-stage' Bayesian procedure[1] by logically combining the 'two-stage' procedure[2] and the idea for processing generic estimates[3]. The first stage manipulates generic plant data to determine a set of estimates for generic parameters,e.g. the mean and the error factor, which accordingly defines a generic failure rate distribution. Then the second stage combines these estimates with the other ones proposed by various generic compendia (we call these generic book type data). This stage adopts another Bayesian procedure to determine the final generic failure rate distribution which is to be used as a priori distribution in the third stage. Then the third stage updates the generic distribution by plant-specific data resulting in a posterior failure rate distribution. Both running failure and demand failure data can be handled in this code. In accordance with the growing needs for a consistent and well-structured reliability database, we constructed a generic reliability database by the MPRDP code[4]. About 30 generic data sources were reviewed and available data were collected and screened from them. We processed reliability data for about 100 safety related components frequently modeled in PSA. The underlying distribution for the failure rate was assumed to be lognormal or gamma, according to the PSA convention. The dependencies among the generic sources were not considered at this time. This problem will be approached in further study.

• 응용통계연구
• /
• 제32권2호
• /
• pp.213-227
• /
• 2019

본 논문에서는 베이즈 신경망을 결합한 종단 간 딥러닝 모형을 한국어 음성인식에 적용하였다. 논문에서는 종단 간 학습 모형으로 연결성 시계열 분류기(connectionist temporal classification), 주의 기제, 그리고 주의 기제에 연결성 시계열 분류기를 결합한 모형을 사용하였으며. 각 모형은 순환신경망(recurrent neural network) 혹은 합성곱신경망(convolutional neural network)을 기반으로 하였다. 추가적으로 디코딩 과정에서 빔 탐색과 유한 상태 오토마타를 활용하여 자모음 순서를 조정한 최적의 문자열을 도출하였다. 또한 베이즈 신경망을 각 종단 간 모형에 적용하여 일반적인 점 추정치와 몬테카를로 추정치를 구하였으며 이를 기존 종단 간 모형의 결괏값과 비교하였다. 최종적으로 본 논문에 제안된 모형 중에 가장 성능이 우수한 모형을 선택하여 현재 상용되고 있는 Application Programming Interface (API)들과 성능을 비교하였다. 우리말샘 온라인 사전 훈련 데이터에 한하여 비교한 결과, 제안된 모형의 word error rate (WER)와 label error rate (LER)는 각각 26.4%와 4.58%로서 76%의 WER와 29.88%의 LER 값을 보인 Google API보다 월등히 개선된 성능을 보였다.

• Journal of Information Processing Systems
• /
• 제14권4호
• /
• pp.980-988
• /
• 2018

Motion estimation is a key Natural User Interface/Natural User Experience (NUI/NUX) technology to utilize motions as commands. HTC VIVE is an excellent device for estimating motions but only considers the positions of hands, not the orientations of arms. Even if the positions of the hands are the same, the meaning of motions can differ according to the orientations of the arms. Therefore, when the positions of arms are measured and utilized, their orientations should be estimated as well. This paper proposes a method for estimating the arm orientations based on the Bayesian probability of the hand positions measured in advance. In experiments, the proposed method was used to measure the hand positions with HTC VIVE. The results showed that the proposed method estimated orientations with an error rate of about 19%, but the possibility of estimating the orientation of any body part without additional devices was demonstrated.

• Communications for Statistical Applications and Methods
• /
• 제13권1호
• /
• pp.89-99
• /
• 2006

This article deals with the problem of estimating parameters of Burr Type XII distribution, on the basis of a general progressive Type II censored sample using Bayesian viewpoints. The maximum likelihood estimator does not admit closed form but explicit sharp lower and upper bounds are provided. Assuming squared error loss and linex loss functions, Bayes estimators of the parameter k, the reliability function, and the failure rate function are obtained in closed form. Finally, a simulation study is also included.

• 한국측량학회지
• /
• 제24권1호
• /
• pp.65-71
• /
• 2006

공공건물, 대학교, 공항 등에 무선 네트워크의 설치가 증가하면서 장소와 시간에 관계없이 모바일 환경에 접근 할 수 있게 되었으며, 모바일 사용자의 급격한 증가로 위치기반서비스의 중요성과 활용에 대한 관심이 증가하고 있다. 본 연구는 무선 랜의 신호세기를 이용하여 모바일 사용자의 위치를 추적하는 실내측위 시스템을 개발하는 것이다. 사용자의 위치를 결정하기 위해 유클리디안 거리 모델과 베이시안 추론 모델을 사용하였다. 실험 결과 유클리디안 거리 모델보다 베이시안 추론 모델이 더 높은 정확도로 위치를 결정하는 것으로 나타났다. 정지상태에서 베이시안 추론 모델은 약 2m 이내의 측위 정확도를 제공하며, 누적좌표수가 증가할수록 그 정확도는 더 향상되었다. 그러나 모바일 사용자의 이동에 따른 누적좌표의 거리오차 및 모바일 기기의 연산량을 감소시키기 위하여, 누적좌표가 5개 일 때의 베이시안 추론 모델이 실내측위에 가장 최적화된 방법이라 생각된다.

• 정보처리학회논문지B
• /
• 제13B권7호
• /
• pp.653-662
• /
• 2006

사용자의 상황에 따라 적절한 서비스를 제공하는 컴퓨팅 환경을 구현하려는 유비쿼터스 컴퓨팅에서 사람과 기계간의 효과적인 상호작용과 사용자의 상황 인식을 위해 사용자의 얼굴 표정 기반의 감정 인식이 HCI의 중요한 수단으로 이용되고 있다. 본 연구는 새로운 베이지안 분류기를 이용하여 상황에 민감한 얼굴 표정에서 기본 감정을 강건하게 인식하는 문제를 다룬다. 표정에 기반한 감정 인식은 두 단계로 나뉘는데 본 연구에서는 얼굴 특징 추출 단계는 색상 히스토그램 방법을 기반으로 하고 표정을 이용한 감정 분류 단계에서는 학습과 테스트를 효과적으로 실행하는 새로운 베이지안 학습 알고리즘인 EADF(Extended Assumed-Density Filtering)을 이용한다. 상황에 민감한 베이지안 학습 알고리즘은 사용자 상황이 달라지면 복잡도가 다른 분류기를 적용할 수 있어 더 정확한 감정 인식이 가능하도록 제안되었다. 실험 결과는 표정 분류 정확도가 91% 이상이며 상황이 드러나지 않게 얼굴 표정 데이터를 모델링한 결과 10.8%의 실험 오류율을 보였다.