• Journal of the Korean Data and Information Science Society
• /
• 제26권1호
• /
• pp.65-75
• /
• 2015

본 논문에서는 전문대 및 4년제 대졸 신규취업자의 임금을 결정하는 요인을 분석하기 위해 개인수준의 인적특성 변수들과 업종수준의 특성변수들을 사용하였다. 본 논문은 개인수준의 임금이 개인수준의 인적특성 (1-수준)과 업종 수준의 산업특성 (2-수준)에 의해 영향을 받는 다층구조 (multi-level)를 가지게 된다는 점을 주목하였다. 이와 같이 위계적 자료 특성을 가지는 복수의 분석단위 구조가 되면, 전통적인 회귀분석에서와 같이 개인수준의 임금이 독립이라는 가정을 할 수 없게 된다. 따라서 본 논문에서는 개인수준의 임금에 영향을 미치는 다층구조의 특성을 가진 변수들의 영향력을 보다 타당하게 분석하기 위한 방법으로 위계선형모형 (HLM; hierarchical linear model)을 이용하였다. 주요 결과는 다음과 같다. 첫째, 개인수준과 업종수준 변수들 모두를 포함한 다중대응분석의 결과에 의하면, 개인수준의 임금이 서로 다른 그룹에 대응되는 개인 특성 변수값과 업종 특성 변수값이 그룹별로 서로 상이하여 개인특성 변수만이 아니라 업종특성 변수도 개인수준의 임금에 영향을 미치고 있다는 점이 발견되었다. 둘째, 개인수준과 업종수준 변수들 모두를 포함한 의사결정나무분석의 결과에 의하면, 개인수준의 임금에 가장 많은 영향을 미치는 변수가 업종별 임금이고, 그 다음으로는 업종별 근로시간, 연령, 성별 등의 순으로 나타났다. 이와 같이 개인수준의 임금을 결정하는 데 있어서 업종의 특성이 매우 중요한 것으로 나타났다는 점은 위계적 선형모형의 활용이 타당하다는 것을 시사하는 것이다. 셋째, 개인수준의 인적특성과 업종 수준의 산업특성 변수들을 모두 포함한 모형이 다른 모형들에 비해서 모형 적합도가 가장 개선되어 위계적 선형모형이 적합한 것으로 나타났다.

• 한국농공학회논문집
• /
• 제57권6호
• /
• pp.1-7
• /
• 2015

Most of chloride diffusion models based on finite difference method (FDM) could not express the diffusion in horizontal direction at each elevation. To overcome these weakness, two dimensional chloride ion penetration model based on finite element method (FEM) to be able to combine various multi-physics simultaneously was suggested by introducing moving mesh technique. To avoid the generation of mesh being able to be distorted depending on the relative movement of water level to static concrete, a rectangular type of mesh was intentionally adopted and the total number of meshes was empirically selected. The simulated results showed that the contents of surface chloride decreased following to the increase of elevation in the top part of low sea level, whereas there were no changes in the bottom part of low level. In the DuraCrete model, the diffusion coefficient of splashed zone is generally smaller than submerged zone, whereas the trend of Life365 model is reverse. Therefore, it could be understood that the developed model using moving mesh technique effectively reflects $DuraCrete^{TM}$ model rather than $Life365^{TM}$ model. In the future, the model will be easily expanded to be combined with various multi-physics models considering water evaporation, heat of hydration, irradiation effect of sun and so on because it is based on FEM.

• 보건행정학회지
• /
• 제25권1호
• /
• pp.31-39
• /
• 2015

Background: The purpose of this study is comparison of the results between regression and multi-level analysis to find out factors influencing outcome indicators (in-hospital death, length of stay, and medical charges) of stroke patients. Methods: By using patient sample data of Health Insurance Review & Assessment Service, patients admitted with stroke were selected as survey target and 15,864 patients and 762 hospitals were surveyed. Results: For the results of existing regression analysis and multi-level analysis, models were assessed through model suitability index value and as a result, the value of results of multi-level analysis decreased compared to the results of regression, showing it is a better model. Conclusion: Factors influencing in-hospital death of stroke patients were analyzed and as a result, intra-class correlation (ICC) was 13.6%. In factors influencing length of stay, ICC was 11.4%, and medical charges, ICC was 17.7%. It was found that factors influencing the outcome indicators of stroke patients may vary in every hospital. This study could carry out more accurate analysis than existing research findings through analysis of reflecting structure at patient level and hospital level factors and analysis on random effect.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권10호
• /
• pp.3230-3255
• /
• 2022

Causality mining in NLP is a significant area of interest, which benefits in many daily life applications, including decision making, business risk management, question answering, future event prediction, scenario generation, and information retrieval. Mining those causalities was a challenging and open problem for the prior non-statistical and statistical techniques using web sources that required hand-crafted linguistics patterns for feature engineering, which were subject to domain knowledge and required much human effort. Those studies overlooked implicit, ambiguous, and heterogeneous causality and focused on explicit causality mining. In contrast to statistical and non-statistical approaches, we present Bidirectional Encoder Representations from Transformers (BERT) integrated with Multi-level Feature Networks (MFN) for causality recognition, called BERT+MFN for causality recognition in noisy and informal web datasets without human-designed features. In our model, MFN consists of a three-column knowledge-oriented network (TC-KN), bi-LSTM, and Relation Network (RN) that mine causality information at the segment level. BERT captures semantic features at the word level. We perform experiments on Alternative Lexicalization (AltLexes) datasets. The experimental outcomes show that our model outperforms baseline causality and text mining techniques.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.786-791
• /
• 2005

This paper is concerned with the examination and evaluation concerning a tuning method of multivariable PID controllers based on partial model matching on frequency domain proposed by authors from practical view point. In this case, PID controller parameters are determined by minimizing the loss function defined by the difference between frequency response of ideal model transfer function and actual frequency response on several frequency points. The purpose of the paper is to examine and evaluate the performance of the method through actual experiments of MIMO liquid level experimental process control equipment.

• 산업경영시스템학회지
• /
• 제23권60호
• /
• pp.11-22
• /
• 2000

The main objective of this research is to analyze an order point and an order quantity of a distribution center and each branch to attain a target service level in multi-level inventory distribution system. In case of product item, we use the item with low volume of average monthly demand. Under the continuous review method, the distribution center places a particular order quantity to an outside supplier whenever the level of inventory reaches an order point, and receives the order quantity after elapsing a certain lead time. Also, each branch places an order quantity to the distribution center whenever the level of inventory reaches an order point, and receives the quantity after elapsing a particular lead time. When an out of stock condition occurs, we assume that the item is backordered. For considering more realistic situations, we use generic type of probability distribution of lead times. In the variable lead time model, the actually achieved service level is estimated as the expected service level. Therefore, this study focuses on the analysis of deciding the optimal order point and order quantity to achieve a target service level at each depot as a expected service level, while the system-wide inventory level is minimized. In addition, we analyze the order level as a maximum level of inventory to suggest more efficient way to develop the low demand item model.

• 디지털콘텐츠학회 논문지
• /
• 제9권3호
• /
• pp.509-516
• /
• 2008

컴퓨터 그래픽의 다중 해상도를 지원하는 하이브리드 렌더링 방법을 제안한다. 기본적으로 단말기 환경의 성능과 사용자 요구조건에 따른 그래픽을 위한 다중해상도 방법은 메시를 이용하거나 렌더링 부분에서 텍스쳐의 밉매핑이나 옥트리를 이용한 레이 트레이싱들의 적응 방법이 사용되었다. 본 연구에서는 제안한 하이브리드 렌더링 방법은 지역조명 모델에서 기존의 고로 음영과 평면 음영 라이브러리를 개선한 방법으로 하나의 물체를 이루는 여러 개의 폴리곤에 각각 다른 음영법을 적절하게 적용한다. 실시간 렌더링 시간을 줄일 수 있는 효율적인 대안 방법이 될 수 있으며 이러한 장점이 유비쿼터스 환경에서 다양한 단말기 환경의 그래픽 콘텐츠의 실시간 적응 서비스에 매우 적절하게 사용될 수 있다.

• International Journal of Contents
• /
• 제12권1호
• /
• pp.49-53
• /
• 2016

"Hahoe Village" in Andong region is an UNESCO World Heritage Site. It should be protected against various disasters such as fire, flooding, earthquake, etc. Among these disasters, flooding has drastic impact on the lives and properties in a wide area. Since "Hahoe Village" is adjacent to Nakdong River, it is important to monitor the water level near the village. In this paper, we developed a hydrological modelling using multi-layer perceptron (MLP) to predict the water level of Nakdong River near "Hahoe Village". To develop the prediction model, error back-propagation (EBP) algorithm was used to train the MLP with water level data near the village and rainfall data at the upper reaches of the village. After training with data in 2012 and 2013, we verified the prediction performance of MLP with untrained data in 2014.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권4호
• /
• pp.246-253
• /
• 2016

Infrared (IR) imaging has been researched for various applications such as surveillance. IR radiation has the capability to detect thermal characteristics of objects under low-light conditions. However, automatic segmentation for finding the object of interest would be challenging since the IR detector often provides the low spatial and contrast resolution image without color and texture information. Another hindrance is that the image can be degraded by noise and clutters. This paper proposes multi-level segmentation for extracting regions of interest (ROIs) and objects of interest (OOIs) in the IR scene. Each level of the multi-level segmentation is composed of a k-means clustering algorithm, an expectation-maximization (EM) algorithm, and a decision process. The k-means clustering initializes the parameters of the Gaussian mixture model (GMM), and the EM algorithm estimates those parameters iteratively. During the multi-level segmentation, the area extracted at one level becomes the input to the next level segmentation. Thus, the segmentation is consecutively performed narrowing the area to be processed. The foreground objects are individually extracted from the final ROI windows. In the experiments, the effectiveness of the proposed method is demonstrated using several IR images, in which human subjects are captured at a long distance. The average probability of error is shown to be lower than that obtained from other conventional methods such as Gonzalez, Otsu, k-means, and EM methods.

• 스마트미디어저널
• /
• 제5권1호
• /
• pp.137-142
• /
• 2016

In this paper, a multi-scale model is applied to the simulation of thrombus growth. This model includes macroscale model and microscale model. The former is used to model the plasma flow with Navier-Stokes equations, and the latter is used to model the platelets adhesion and aggregation, thrombus motion, and the surface expansion of thrombus. The force acting on platelets and thrombus from plasma is modeled by the drag force, and the forces from biochemical reactions are modeled by the adhesion force and the aggregation force. As more platelets are merged into the thrombus, the thrombus surface expands. We proposed a thrombus growth model for simulating the expansion of thrombus surface and tracking the surface by Level Set Methods. We implemented the computational model. The model performs well, and the experimental results show that the shape of thrombus in level set expansion form is similar with the thrombus in clinical test.