다구찌 디자인을 이용한 데이터 퓨전 및 군집분석 분류 성능 비교

신형원,손소영 한국경영과학회 2000 한국경영과학회 학술대회논문집 Vol.- No.1

In this paper, we compare the clasification performace of both data fusion and clustering algorithms (Data Bagging, Variable Selection Bagging, Parameter Combining, Clustering) to logistic regression in consideration of various characteristics of input data. Four factors used to simulate the logistic model are (1) correlation among input variables (2) variance of observation (3) training data size and (4) input-output function. Since the relationship between input & output is not typically known, we use Taguchi design to improve the practicality of out study results by letting it as a noise factor. Experimental study results indicate the following: Clustering based logistic regression truns out to provide the highest classification accuracy when input variables are weakly correlated and the variance of data is high. When there is high correlation among input variables, variable bagging performs better than logistic regression. When there is strong correlation among input variables and high variance between observations, bagging appears to be marginally better than logistic regression but was not significant

Metal과 Metal Oxidefh 구성된 복합구조의 Peel Strength

신형원,정택균,이효수,정승부,Shin, Hyeong-Won,Jung, Taek-Kyun,Lee, Hyo-Soo,Jung, Seung-Boo 한국마이크로전자및패키징학회 2013 마이크로전자 및 패키징학회지 Vol.20 No.4

양극산화(anodization)공정으로 제작된 규칙성 나노구조의 다공성 산화알루미늄(Aluminum Anodic Oxide, AAO)는 공정이 적용된 LED 모듈은 비교적 쉽고 경제적이므로 최근 LED용 방열소재로 응용하기 위하여 다양하게 연구가 진행되고 있다. 일반적으로 LED 모듈은 알루미늄/폴리머/구리 회로층으로 구성되며 절연체 역할을 하는 폴리머는 히트스프레더로 구성되어있다. 그러나 열전도도가 낮은 폴리머로 인하여 LED부품의 열 방출이 원활하지 못하므로 LED의 수명단축 및 오작동에 영향을 미친다. 따라서, 본 연구에서는 폴리머 대신 상대적으로 열전도도가 우수한 AAO를 양극산화 공정으로 제작하여 히트스프레더(heat spread)로 사용하였다. 이때, AAO와 금속인 구리 회로층간의 접착력을 향상시키기 위하여 스퍼터링 DBC(direct bonding copper)법으로 시드층(seed layer)을 형성한 뒤 최종적으로 전해도금공정으로 구리회로층을 형성하였다. 본 연구에서는 양극 산화공정으로 AAO와 금속간의 접착강도를 개선하여 1.18~1.45 kgf/cm와 같은 우수한 peel strength 값을 얻었다. A lot of various researches have been going on to use heat spreader for LED module. Nano porous aluminum anodic oxide (AAO) applied LED, which is produced from anodization, is easy and economically advantageous. Convensional LED module is consist of aluminum/adhesive/copper circuit. The polymer adhesive in this module is used as heat spreader. However the thermal emission of LED component is degraded because of low heat conductivity of polymer and also reliability of LED component is reduced. Therefore, AAO in this work was applied to heat spreader of LED module which has higher heat conductivity compare to polymer. Bonding strength between AAO and copper circuit was improved with Ti/Cu seed layer by copper sputtering process (DBC) before the bonding. And this copper circuit has been fabricated by electro plating method. Peel strength of AAO and copper circuit in this work showed range between 1.18~1.45 kgf/cm with anodizing process which is very suitable for high power LED application.

범주형 자료의 결측치 추정방법 성능 비교

신형원,손소영 한국통계학회 2002 응용통계연구 Vol.15 No.1

Various kinds of estimation methods have been developed for imputation of categorical missing data. They include category method, logistic regression, and association rule. In this study, we propose two fusions algorithms based on both neural network and voting scheme that combine the results of individual imputation methods. A Mont-Carlo simulation is used to compare the performance of these methods. Five factors used to simulate the missing data pattern are (1) input-output function, (2) data size, (3) noise of input-output function (4) proportion of missing data, and (5) pattern of missing data. Experimental study results indicate the following: when the data size is small and missing data proportion is large, modal category method, association rule, and neural network based fusion have better performances than the other methods. However, when the data size is small and correlation between input and missing output is strong, logistic regression and neural network barred fusion algorithm appear better than the others. When data size is large with low missing data proportion, a large noise, and strong correlation between input and missing output, neural networks based fusion algorithm turns out to be the best choice. 범주형 데이터의 결측치 추정을 위하여 최빈 범주법, 로지스틱 회귀분석, 연관규칙과 같은 다양한 방법이 연구되어 왔다. 본 연구에서는 이러한 방법의 추정 값을 결합하는 신경망 융합과 투표융합 방법을 제안하고 이의 성능을 시뮬레이션을 이용하여 비교하였다. 실험에 사용된 데이터의 특성을 나타내는 인자로는 (1) 입출력 변수간의 연결함수, (2) 데이터의 크기, (3) 노이즈의 크기 (4) 결측치의 비율, (5) 결측발생 함수를 사용하였다. 분석결과는 다음과 같다. 데이터의 크기가 작고 결측 발생 비율이 높으면 최빈 범주법, 연관규칙, 신경망 융합의 성능이 높게 나타났으며 데이터의 크기가 작고 결측발생 확률이 결측이 안된 나머지 변수에 높은 의존관계가 있으면 로지스틱 회귀분석, 신경망 융합의 성능이 높게 나타났다. 데이터의 크기가 크고, 결측치의 비율이 낮으면서, 노이즈가 크고 결측발생 확률이 결측이 안된 나머지 변수에 높은 의존관계가 있으면 신경망 융합의 성능이 높게 나타났다.

범주형 자료의 결측치 추정방법 성능 비교

신형원,손소영 대한산업공학회 2001 대한산업공학회 춘계학술대회논문집 Vol.2001 No.4

Thermal Via 구조 LED 모듈의 열저항 변화

신형원,이효수,방제오,유세훈,정승부,김강동,Shin, Hyeong-Won,Lee, Hyo-Soo,Bang, Jae-Oh,Yoo, Se-Hoon,Jung, Seung-Boo,Kim, Kang-Dong 한국마이크로전자및패키징학회 2010 마이크로전자 및 패키징학회지 Vol.17 No.4

LED (Light Emitting Diode)는 인가된 에너지 대비 15%가 빛으로, 나머지 85%가 열로 변환되는 것은 이미 잘 알려져 있다. 최근 LED칩 용량이 증가함에 따라서 LED칩으로부터 방출되는 열은 더욱 증가하게 되고 이는 LED 제품의 성능저하와 수명단축에 직접적인 영향을 미친다. 따라서, 산업계에서는 고출력 LED 칩에서 발생하는 열을 제어하기 위해 제품설계구조 연구가 진행 중에 있으며 또한, 부가적으로, 기존의 알루미늄, 접착제 및 구리를 사용하는 MCL(Metal Clad Laminate)구조에서 저가형 FR4 및 구리를 사용하는 CCL (Copper Clad Laminate)로 변경하여 원가절감을 하고자 하는 대체 소재연구가 진행되고 있다. 본 연구에서는 저가형 CCL에 열방출 극대화를 위하여 열비아(thermal via)를 디자인별로 형성한 후 1 W급 LED 칩을 실장하여 열저항(thermal resistance) 변화를 분석하였으며, 최적의 열방출을 위한 열비아 구조를 제안하고자 하였다. LED (Light Emitting Diode) is 85% of the applied energy is converted into heat that is already well known. Lately, LED chips increasing the capacity as result delivered to increase the heat of the LED products and module that directly related to life span and degradation. Thus, in industry the high-power LED chip to control the heat generated during the course of the study and the existing aluminum, copper adhesives, and uses MLC (Metal clad laminate) structures using low-cost FR4 and copper CCL (Copper Clad Laminate) to reduce costs by changing to a study being carried out. In this study, using low-cost CCL Class, mounted 1W LED chip to analyze changes in the thermal resistance. In addition, heat dissipation in the CCL to facilitate a variety of thermal via design outside of the heat generated by the LED chip to control and facilitate the optimal structure of the heat dissipation is suggested.

간아세포종 1 례

신형원,조수철,최삼임 의과학연구소 1989 全北醫大論文集 Vol.13 No.1

Primary neoplasm of the liver constitutes between 0.5% and 2.0% of pediatric tumors. Two major forms are hepatocarcinoma and hepatoblastoma, the latter is classified pure epithelial and mixed type. The mixed type of hepatoblatoma contains mesenchymal tissue in addition to epithelial elements. Authors experienced a case of mixed hepatoblastoma in a 25-month old male patient, who complained of abdominal distension and weight loss. A brief review of literature was made.

Analysis of Peel Strength of Consisting of an Aluminum Sheet, Anodic Aluminum Oxide and a Copper Foil Laminate Composite

신형원,이효수,정승부 대한금속·재료학회 2017 METALS AND MATERIALS International Vol.23 No.1

Laminate composites consisting of an aluminum sheet, anodic aluminum oxide, and copper foil have been used as heat-spreader materials for high-power light-emitting diodes (LEDs). These composites are comparable to the conventional structure comprising an aluminum sheet, epoxy adhesives, and copper foil. The peel strength between the copper foil and anodic aluminum oxide should be more than 1.0 kgf/cm in order to be applied in high-power LED products. We investigated the effect of the anodic aluminum oxide morphology and heattreatment conditions on the peel strength of the composites. We formed an anodic aluminum oxide layer on a 99.999% pure aluminum sheet using electrochemical anodization. A Ti/Cu seed layer was formed using the sputtering direct bonding copper process in order to form a copper circuit layer on the anodic aluminum oxide layer by electroplating. The developed heat spreader, composed of an aluminum layer, anodic aluminum oxide, and a copper circuit layer, showed peel strengths ranging from 1.05 to 3.45 kgf/cm, which is very suitable for high-power LED applications.

범주형 자료의 결측치 추정방법 성능 비교

신형원,손소영 한국경영과학회 2001 한국경영과학회 학술대회논문집 Vol.- No.1(2)

Various kinds of estimation methods have been developed for imputation of categorical missing data. They include category method, logistic regression, and association rule. In this study, we propose two fusions algorithms based on both neural network and voting scheme that combine the results of individual imputation methods. A Monte-Carlo simulation is used to compare the performance of these methods. Five factor s used to simulate the missing data pattern are (1) input-output function, (2) data size, (3) noise of input-output function (4) proportion of missing data, and (5) pattern of missing data. Experimental study results indicate the following: when the data size is small and missing data proportion is large, modal category method, association rule, and neural network based fusion have better performances than the other methods. However, when the data size is small and correlation between input and missing output is strong, logistic regression and neural network based fusion algorithm appear better than the others. When data size is large with low missing data proportion, a large noise, and strong correlation between input and missing output, neural networks based fusion algorithm turns out to be the best choice.

Combined structure 형성과정에 이용된 Cu-Fe-P-C 합금과 plate된 Cu/Ag와의 계면분석 (초)

신형원,이해중,박소현,이효수,임성철,권혁천,정승부 대한금속ㆍ재료학회 2009 대한금속재료학회 학술대회 개요집 Vol.2009 No.-

Study of emotional color of Copper alloys with its surface morphology and reflectivity of the wavelength

신형원,이효수,정승부 한국감성과학회 2016 춘계학술대회 Vol.2016 No.-