• The Korean Journal of Financial Studies
• /
• v.7 no.1
• /
• pp.1-47
• /
• 2001

증권가격의 시계열을 그래프로 표시하면 이 시계열의 운동양태가 파악될 수도 있다. 그래프를 통하여 추세가 존재하고 있는지 아니면 존재하지 않는지를 파악할 수 있을 것이다. 그리고 이 그래프를 통하여 증권가격 시계열이 정상적과정에 의하여 생성되는지의 여부가 인식될 수도 있을 것이며, (조건부) 이분산이 존재하고 있는지 또는 (조건부) 동분산이 존재하고 있는지도 인식될 수 있을 것이다. 간단한 기술통계량을 통하여 증권시계열의 성질을 파악할수도 있다. 이 시계열이 선형과정에 의하여 생성되는지 아니면 비선형과정에 의하여 생성되는지도 인식할 수 있을 것이다. 뿐만아니라 비선형과정중 하나인 카오스 과정에 의하여 증권가격이 생성되는지의 여부도 파악할 수 있을 것이다. 증권가격의 실현된 표본경로와 시뮬레이션을 통하여 얻은 표본경로가 일치하는지 또는 불일치하는지에 대한 판별을 통하여 모형정립에서 특히 많이 사용되고 있는 확률과정들이 생성시키는 증권가격 시계열이 실제로 관찰된 가격 시계열과 일치하여 현실적합성을 가지고 있는지의 여부도 판단할 수 있을 것이다. 주가시계열 그 자체를 출발점으로 하여 이 시계열의 움직임과 행동양식을 파악해가면 수많은 연구를 통하여 축적된 이론들과 주가를 형성시키는 성질들이 현실적으로 성립하고 있는지도 밝힐 수 있고 개발된 이론들의 장점과 단점을 강도높게 밝힐 수 있는 계기도 갖게 될 것이다. 데이터를 있는 그대로 면밀하게 검토하면 이미 공개된 문제점(open question)도 확인할 수 있을 것이고 아직 알려지지 않은 문제점들과 질문들을 찾게 될 수도 있을 것이다. 이것들은 앞으로의 연구를 위한 중요한 발견이 될 수 있을 것이다. 이 논문에서는 문제와 질문의 발견에 초점을 둔다. 이 논문에서는 한국의 주식시장과 미국의 주식시장을 대비하여 다룬다. 우리가 그동안의 연구를 통하여 미국의 문헌과 미국의 시장에 대한 지식을 상당히 축적하고 있는 만큼 이 대비를 통하여 두 시장이 동일하게 가지고 있는 행동양태와 서로 상이하게 가지고 있는 점들을 파악하면 두 시장에 대한 이해의 폭도 넓어질 것이며 동시에 미국의 연구결과를 수용하는 큰 방향을 결정하는데에도 일조가 되리라고 생각된다.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1994.06c
• /
• pp.138-141
• /
• 1994

기존의 HMM을 이용한 음성인식기는 대부분 ML 추정에 기초한 Baum-Welch 알고리듬으로 학습되었다. ML학습은 기본적으로 무한한 양의 학습 데이터가 주어지고, 각 모델들이 서로 독립이라는 가정에 기초한다. 하지만 실제적인 학습의 경우에 각 모델들이 서로 독립이라고 보기 어렵고, 학습 데이터의 양도 상당히 제한되어 있어서 인식기의 변별력을 저하시키는 주된 원인이 되고 있다. 본 논문에서는 전통적인 패턴분류기법인 Bayes 결정이론에 따라 최소오차율분류를 위한 MAP 수식화를 유도하고, 그에 기초한 HMM의 변별력 있는 학습 알고리듬을 제안한다. 최소오차율분류를 근사화한 사후확률로 표현된 비용함수를 정의하고, 그 비용함수에 조건부 경사강하법을 적용한다. 제안된 알고리듬을 분류하기 어려운 한국어 단음절 인식에 적용한 결과, 기존의 ML 알고리듬으로 학습한 경우 발생한 오인식 개수의 약 10% 가량이 개선되었다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04a
• /
• pp.395-397
• /
• 2003

서비스거부공격(DoS)은 공격의 목표가 되는 호스트에 악의적인 패킷을 단지 하나만 보냄으로써 공격이 이루어 질 수 없다. 그렇기 때문에 침입탐지시스템에서는 패킷에 대한 일련의 순차성을 알아냄으로써 연속적인 패킷을 보내는 DoS 공격으로서의 침입을 탐지해 내는 것이 가능하다. 본 연구에서는 네트워크 패킷에 대한 송신지의 주소와 서비스, 목적지의 주소와 서비스를 이용하여 이벤트를 정의 하였으며. 이렇게 정의된 이벤트를 이용하여 정상적인 상황에서의 이벤트에 대한 순차의존성과 DoS 공격 상황에서의 이벤트에 대한 순차의존성을 알아내었다. 이벤트에 대한 순차의존성을 측정하기 위해서 정상 이벤트가 발생하는 확률을 이용하였으며. 더 나아가서는 정보이론 분야에서의 조건부 엔트로피의 사용도 제안하였다.

• Journal of KIISE:Software and Applications
• /
• v.27 no.8
• /
• pp.855-863
• /
• 2000

Although Behavior-Knowledge Space (BKS) method, one of well known decision combination methods, does not need any assumptions in combining the multiple experts, it should theoretically build exponential storage spaces for storing and managing jointly observed K decisions from K experts. That is, combining K experts needs a (K+1)st-order probability distribution. However, it is well known that the distribution becomes unmanageable in storing and estimating, even for a small K. In order to overcome such weakness, it has been studied to decompose a probability distribution into a number of component distributions and to approximate the distribution with a product of the component distributions. One of such previous works is to apply a conditional independence assumption to the distribution. Another work is to approximate the distribution with a product of only first-order tree dependencies or second-order distributions as shown in [1]. In this paper, higher order dependency than the first-order is considered in approximating the distribution and a dependency-based framework is proposed to optimally approximate the (K+1)st-order probability distribution with a product set of dth-order dependencies where ($1{\le}d{\le}K$), and to combine multiple experts based on the product set using the Bayesian formalism. This framework was experimented and evaluated with a standardized CENPARMI data base.

• The Korean Journal of Applied Statistics
• /
• v.11 no.2
• /
• pp.377-387
• /
• 1998

A Markov Chain Monte Carlo method with data augmentation is developed to compute the features of the posterior distribution. For each observed failure epoch, we introduced latent variables that indicates with component of the Superposition model. This data augmentation approach facilitates specification of the transitional measure in the Markov Chain. Metropolis algorithms along with Gibbs steps are proposed to preform the Bayesian inference of such models. for model determination, we explored the Pre-quential conditional predictive Ordinate(PCPO) criterion that selects the best model with the largest posterior likelihood among models using all possible subsets of the component intensity functions. To relax the monotonic intensity function assumptions, we consider in this paper Superposition of Musa-Okumoto and Erlang(2) models. A numerical example with simulated dataset is given.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.16 no.7
• /
• pp.1234-1243
• /
• 1992

A systematic procedure to design fuzzy PID controllers is developed in this paper. The concept of local fuzzy control cell is proposed by introducing both an adequate global control rule and membership functions to simplify a fuzzy logic controller. Fuzzy decision is made by using algebraic product and parallel firing arithematic mean, and a defuzzification strategy is adopted for improving the computational efficiency based on nonfuzzy micro-processor. A direct method, transforming the typical output of quasi-linear fuzzy operator to the digital compensator of PID form, is also proposed. Finally, the proposed algorithm is applied to an DC-servo motor. It is found that this algorithm is systematic and robust through computer simulations and implementation of controller using Intel 8097 micro-processor.

• KDI Journal of Economic Policy
• /
• v.13 no.4
• /
• pp.117-140
• /
• 1991

The purpose of this paper is primarily to introduce a nonparametric statistical tool developed by Baek and Brock to detect a unidirectional causal ordering between two economic variables and apply it to interesting macroeconomic relationships among money, production and prices. It can be applied to any other causal structure, for instance, defense spending and economic performance, stock market index and market interest rates etc. A key building block of the test for nonlinear Granger causality used in this paper is the correlation. The main emphasis is put on nonlinear causal structure rather than a linear one because the conventional F-test provides high power against the linear causal relationship. Based on asymptotic normality of our test statistic, the nonlinear causality test is finally derived. Size of the test is reported for some parameters. When it is applied to a money, production and prices model, some evidences of nonlinear causality are found by the corrected size of the test. For instance, nonlinear causal relationships between production and prices are demonstrated in both directions, however, these results were ignored by the conventional F-test. A similar results between money and prices are obtained at high lag variables.

• Journal of KIISE:Software and Applications
• /
• v.28 no.9
• /
• pp.681-687
• /
• 2001

In order to raise a class discrimination power by combining multiple classifiers under the Bayesian decision theory, the upper bound of a Bayes error rate bounded by the conditional entropy of a class variable and decision variables obtained from training data samples should be minimized. Wang and Wong proposed a tree dependence first-order approximation scheme of a high order probability distribution composed of the class and multiple feature pattern variables for minimizing the upper bound of the Bayes error rate. This paper presents an extended high order product approximation scheme dealing with higher order dependency more than the first-order tree dependence, based on the minimization of the upper bound of the Bayes error rate. Multiple recognizers for unconstrained handwritten numerals from CENPARMI were combined by the proposed approximation scheme using the Bayesian formalism, and the high recognition rates were obtained by them.

• Journal of KIISE:Computer Systems and Theory
• /
• v.36 no.4
• /
• pp.274-282
• /
• 2009

Various location tracking sensors; such as GPS, INS, radar, and optical equipment; are used for tracking moving targets. In order to effectively track moving targets, it is necessary to develop an effective fusion method for these heterogeneous devices. There have been studies in which the estimated values of each sensors were regarded as different models and fused together, considering the different error characteristics of the sensors for the improvement of tracking performance using heterogeneous multi-sensor. However, the rate of errors for the estimated values of other sensors has increased, in that there has been a sharp increase in sensor errors and the attempts to change the estimated sensor values for the Sensor Probability could not be applied in real time. In this study, the Sensor Probability is obtained by comparing the RMSE (Root Mean Square Error) for the difference between the updated and measured values of the Kalman filter for each sensor. The process of substituting the new combined values for the Kalman filter input values for each sensor is excluded. There are improvements in both the real-time application of estimated sensor values, and the tracking performance for the areas in which the sensor performance has rapidly decreased. The proposed algorithm adds the error characteristic of each sensor as a conditional probability value, and ensures greater accuracy by performing the track fusion with the sensors with the most reliable performance. The trajectory of a UAV is generated in an experiment and a performance analysis is conducted with other fusion algorithms.

• Journal of Korea Water Resources Association
• /
• v.49 no.4
• /
• pp.275-282
• /
• 2016

The objective of this study is to propose a new method to determine the drought event and the design drought severity. In order to define a drought event from precipitation data, theory of run was applied with the cumulative rainfall deficit. When we have a large amount of rainfall over the threshold level, in this study, we compare with the previous cumulative rainfall deficit to determine whether the drought is relieved or not. The recurrence characteristics of the drought severity on the specific duration was analyzed by the conditional bivariate copula function and confidence intervals were estimated to quantify uncertainties. The methodology was applied to Seoul station with the historical dataset (1909~2015). It was observed that the past droughts considered as extreme hydrological events had from 10 to 50 years of return period. On the other hand, the current on-going drought event started from 2013 showed the significantly higher return period. It is expected that the result of this study may be utilized as the reliable criteria based on the concept of return period for the drought contingency plan.