• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.13 no.4
• /
• pp.405-415
• /
• 2000

Time series of wind-induced pressure on a structure are modeled using autoregressive moving average (ARMA) model. In an AR process, the current value of the time series is expressed in terms of a finite, linear combination of the previous values and a white noise. In a MA process, the value of the time series is linearly dependent on a finite number of the previous white noises. The ARMA process is a combination of the AR and MA processes. In this paper, the ARMA models with several different combinations of the AR and MA orders are fitted to the wind-induced pressure time series, and the procedure to select the most appropriate ARMA model to represent the data is described. The maximum likelihood method is used to estimate the model parameters, and the AICC model selection criterion is employed in the optimization of the model order, which is assumed to be a measure of the temporal complexity of the pressure time series. The goodness of fit of the model is examined using the LBP test. It is shown that AR processes adequately fit wind pressure time series.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.34 no.4
• /
• pp.399-405
• /
• 2010

The grasping motion of a person's hand for a simplified degree of freedom was modeled by using the photographic motion measured by a high-speed camera. The mathematical expression of distal interphalangeal (DIP) motion was developed by using relation models of the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) motions to reduce the degree of freedom. The mathematical expression for humanoid-hand operation obtained using a learning algorithm with a neural network and using a kernel regression model were compared. A feasible model of hand operation was obtained on the basis of comparative data analysis by using the kernel regression model.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2002.12a
• /
• pp.203-206
• /
• 2002

동적 시스템의 동정은 시스템의 관측된 데이터를 가지고 동적 모델의 수학적 모델을 찾는 문제를 다루는 것이다. 기존의 고전적인 방법으로는 차분 방정식(ARX 또는 ARMAX) 또는 상태 공간 표현에 관한 계수들을 추정하기 위해서 회귀 기법 등을 사용하였다. 그러나 이러한 고전적인 방법들은 파라미터가 비선형이고, 실세계 문제에서 모델링 오차나 측정 잡음을 수반하게 되면 탐색의 어려움을 가지게 된다. 따라서 이러한 문제점을 극복하고자 퍼지 이론이나 신경망 이론 둥이 이용되었으나 본 논문에서는 비선형 동적 시스템의 파라미터 동정에 최근 복잡한 최적화 문제를 해결하는 도구로 점점 관심을 받고 있는 유전 알고리즘을 동정 알고리즘으로 제안하고, 비선형 동적 시스템의 파라미터 동정에 유전 알고리즘을 응용한 몇 가지 예를 제시하고자 한다.

• Journal of Advanced Navigation Technology
• /
• v.23 no.6
• /
• pp.561-569
• /
• 2019

Solar photovoltaic can provide electrical energy with only radiation, and its use is expanding rapidly as a new energy source. This study predicts the short and long-term PV power generation using actual converter output data of photovoltaic system. The prediction algorithm uses multiple linear regression, support vector machine (SVM), and deep learning such as deep neural network (DNN) and long short-term memory (LSTM). In addition, three models are used according to the input and output structure of the weather element. Long-term forecasts are made monthly, seasonally and annually, and short-term forecasts are made for 7 days. As a result, the deep learning network is better in prediction accuracy than multiple linear regression and SVM. In addition, LSTM, which is a better model for time series prediction than DNN, is somewhat superior in terms of prediction accuracy. The experiment results according to the input and output structure appear Model 2 has less error than Model 1, and Model 3 has less error than Model 2.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1995.06a
• /
• pp.175-178
• /
• 1995

기존이 연속 출력 분포형 HMM은 시계열의 과도적 변화에 대하여 표현 능력이 부족하다는 단점이 있다. 이것을 보완하기 위해 본 논문에서는 음성의 동적 변화를 반영하기 위한 특징 파라메타로서 여러 개의 프레임을 결합하여 세그멘트를 구성하여 각각에 대해 한 개의 벡터를 만들었다. 이것을 그대로 이용하면 세그멘트의 프레임수에 대응하는 파라메타의 차원수가 증가하기 때문에 학습 데이터가 불충분한 경우 모델의 파라메타를 잘 추정할 수 없으므로 K-L 전개로서 파라메타의 차원을 압축하여 파라메타수를 감소시켰다. 인식실험은 한국어 단음절에 대하여 멜켑스트럼ㅇ르 K-L 전개로 압축한 벡터를 이용한 결과와 멜켑스트럼, 멜켑스트럼 선형회귀계수를 파라메타로 이용한 경우를 비교하였다. 실험결과 K-L 전개로 압축한 벡터만을 이용한 경우는 멜켑스트럼 + 선형회귀계수를 파라메타로 이용한 경우보다 인식율이 낮앗으나 멜켑스트럼 + K-L 전개로 압축한 경우와 거의 동등한 결과를 얻을 수 있었다.

• Korea Journal of Hospital Management
• /
• v.20 no.1
• /
• pp.53-64
• /
• 2015

Our objective was to evaluate differences in linear regression versus multilevel(cross-level interaction model) modeling for affecting factors lumbar discectomy. The data were used in 2011 patients with HIRA sample data. Total number of analysis is 3,641 patients and 248 hospitals. The results of research model showed that the type and location of the hospital-level factors were significant. However, all factors of patient-level were similar in the two models. Therefore, it requires the selection of an appropriate model for a more accurate analysis of the influencing factors in the daily medical charge.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2019.11a
• /
• pp.16-17
• /
• 2019

In order to measure the amount of carbon dioxide emitted from the construction sites, many literature which have been conducted have proposed methodologies for calculating coefficients based on actual data collections for estimating the emission formula. The existing data collected under controlled conditions not on site measurement were too limited to apply in actual sites. The purpose of this study is to conduct analysis based on the data measured in fields and to present predictive models using artificial neural network and nonlinear regression analysis for appropriate predictions and practical applications.

• The KIPS Transactions:PartD
• /
• v.10D no.7
• /
• pp.1137-1144
• /
• 2003

The existing Function Point method to estimate the software size has been utilized frequently with the management information system. Due to the expanding usage of the real-time and embedded system, the Full Function Point method is being proposed. However, despite many research is being carried out relation to the software size, the research on the model to estimate the development cost from the measured software size is inadequate. This paper analyzed the linear regression model and power regression model which estimate the development cost from the software FFP The power model is selected, which shows its estimation is most adequate.

• Journal of the Korea Computer Industry Society
• /
• v.2 no.6
• /
• pp.859-866
• /
• 2001

Successful project planning relies on a good estimation of the effort required to complete a project, together with the schedule options that may be available. Despite the extensive research done developing new and better models, existing software effort estimation models are present only the total effort and effort (or manpower： people per unit time) function for the software life-cycle. Also, Putnam presents constant effort rate in each subcycles. However, the size of total efforts are variable according to the software projects under the influence of its size, complexity and operational environment. As a result, the allocated effort in subcycle also differ from project to project. This paper suggests the linear and polynomial effort estimation models in specifying, building and testing phase followed by the project total effort. These models are derived from 128 different projects. This result can be considered as a practical guideline in management of project schedule and effort allocation.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.11
• /
• pp.41-49
• /
• 2019

We are to build an unsupervised machine learning-based language model which can estimate the amount of information that are in need to process words consisting of subword-level morphemes and syllables. We are then to investigate whether the reading times of words reflecting their morphemic and syllabic structures are predicted by an information-theoretic measure such as surprisal. Specifically, the proposed Morfessor-based unsupervised machine learning model is first to be trained on the large dataset of sentences on Sejong Corpus and is then to be applied to estimate the information-theoretic measure on each word in the test data of Korean words. The reading times of the words in the test data are to be recruited from Korean Lexicon Project (KLP) Database. A comparison between the information-theoretic measures of the words in point and the corresponding reading times by using a linear mixed effect model reveals a reliable correlation between surprisal and reading time. We conclude that surprisal is positively related to the processing effort (i.e. reading time), confirming the surprisal hypothesis.