• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.81-84
• /
• 2022

제조 산업에서의 이상치 검출은 생산품의 품질과 운영비용을 절감하기 위한 중요한 요소로 최근 딥러닝을 사용하여 자동화되고 있다. 이상치 검출을 위한 딥러닝 기법에는 CNN이 있으며, CNN을 계층적으로 구성할 경우 단일 CNN 모델에 비해 상대적으로 성능의 향상을 보일 수 있다는 것이 많은 선행 연구에서 나타났다. 이에 MVTec-AD 데이터셋을 이용하여 계층 CNN이 다중 클래스 이상치 판별 문제에 대해 효과적인지를 탐구하고자 하였다. 실험 결과 단일 CNN의 정확도는 0.7715, 계층 CNN의 정확도는 0.7838로 다중 클래스 이상치 판별 문제에 있어 계층 CNN 방식 접근이 다중 클래스 이상치 탐지 문제에서 알고리즘의 성능을 향상할 수 있음을 확인할 수 있었다. 계층 CNN은 모델과 파라미터의 개수와 리소스의 사용이 단일 CNN에 비하여 기하급수적으로 증가한다는 단점이 존재한다. 이에 계층 CNN의 장점을 유지하며 사용 리소스를 절약하고자 하였고 K-means, GMM, 계층적 클러스터링 알고리즘을 통해 제작한 새로운 클래스를 이용해 계층 CNN을 구성하여 각각 정확도 0.7930, 0.7891, 0.7936의 결과를 얻을 수 있었다. 이를 통해 Clustering 알고리즘을 사용하여 적절히 물체를 분류할 경우 물체에 따른 개별 상태 판단 모델을 제작하는 것과 비슷하거나 더 좋은 성능을 내며 리소스 사용을 줄일 수 있음을 확인할 수 있었다.

• Journal of the Economic Geographical Society of Korea
• /
• v.17 no.2
• /
• pp.402-419
• /
• 2014

This study attempts to investigate the spatial characteristics of private tutoring markets for adults which have been expanded rapidly with labor market changes in Korea. In particular, For the purpose, we examine thoroughly various indies of labor markets and private tutoring markets for adults in Korea in first and then analyze the spatial characteristics. We classify private tutoring institutes for adults into two categories by job-statuses and education levels, and analyze the spatial distribution patterns of the attendants of the classes. In order to understand the spatial characteristic of their distributions, we distinguish whether there exist the spatial autocorrelation or not by applying Moran's I values for each categories in first. We also examine the spatial cluster patterns by Hot spots analysis utilizing $G^*$ statistics. Multiple linear regression models are developed for each category to explain the relationships between the spatial distributions of private tutoring institutes and geographical variables.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.7
• /
• pp.24-30
• /
• 1993

In this paper, a new method is proposed to recognize a character from its similar characters, which are selected by pattern matching method in Korean character recognition. This new method, which couples the merits of already suggested methods, can choose the character to be in the candidate set and discriminate it from the others correctly. To evaluate performance of this algorithm, we used 15 kinds of different laser printer fonts and obtained about 97% of recognition rate.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.460-463
• /
• 2022

세계에서 가장 중요한 온대 과일 작물 중 하나인 사과의 생산성과 품질은 병해충 여부에 큰 영향을 받는다. 이를 진단하기 위해서는 효율적이고 많은 전문 지식과 상당한 시간이 필요하다. 그러므로 이를 해결하기 위해 효율적이고 정확하게 다양한 병해충을 진단하는 시스템이 필요하다. 본 논문에서는 이미지 분석에 큰 효율을 보인 딥러닝 기반 CNN 들을 비교 분석하여 사과의 병해충 여부를 판별하고 최적의 모델을 제시한다. 딥러닝 기반 CNN 구조를 가진 AlexNet, VGGNet, Inception-ResNet-v2, DenseNet 을 채택해 사과 병해충 분류 성능 평가를 진행했다. 그 결과 DenseNet 이 가장 우수한 성능을 보여주었다.

• Composites Research
• /
• v.14 no.4
• /
• pp.8-14
• /
• 2001

The satellite system experiences severe mechanical loads during the launch period. Therefore, the positive margin of safety of the satellite system must be demonstrated for every possible mechanical loading conditions during the launch period. This paper presents modal and stress analysis results due to quasi-static loads for the satellite antenna system. The failure tendency fur the sandwich construction of the satellite antenna system has been studied with various lamination angles of unidirectional prepreg.

• Journal of the Korean Chemical Society
• /
• v.47 no.4
• /
• pp.376-390
• /
• 2003

The contents of chemical reaction in the science textbooks, used by starting from elementary school to college level, were investigated how much closely connected with them. For this purpose, nineteen subjects of chemistry contents were selected and the articulation analysis, classified into 5 criteria, was carried out. The results indicated that the articulation from elementary school to middle school looked quite acceptable, but there was a big gap from middle school to high school and the articulation from high school to college exhibited a big overlap and gap. Finally, the flow map of chemical reaction was suggested.

• Economic and Environmental Geology
• /
• v.34 no.1
• /
• pp.13-23
• /
• 2001

Bentonite layers are intercalated within the basal conglomerates in the Tertiary sedimentary basins of Kampo, Janggi and Pohang, southeastern Korea. Eighteen samples of the bentonites went through X-ray diffraction, scanning electron microscopy, heavy mineral analyses, chemical analyses and oxygen, hydrogen stable isotope analyses to define the mineralogical characters of the bentonites. Heavy minerals such as zircons, apatites, amphiboles and biotites separated from bentonites show clean and euhedral surfaces, which are the characteristic features of volcanic origin. But biotites from the Chunbook Conglomerate are found as altered and heavily broken flakes which implies longer transportation of these bentonites. $TiO_{2}/Al_{2}O_{3} ratios of <2 $\mu$m particle fractions (the Chunbook Conglomerate 0.031; Janggi 0.029; Kampo 0.025) suggest that those are originated from volcanic tuffs. That is, the higher the value is, the more mafic in chemical compositions of the original tuffs. Authigenic montmorillonite and zeolite minerals were observed by SEM, which indicates diagenesis origin of bentonites. But the samples from the Chunbook Conglomerate showed only chaotically packed clay flakes in the matrix of sands or conglomerates, which implies detrital influence, not authigenic origin. The structural formulae of montmorillonite from these basins reflects their environment of formation. Fe (Ⅵ) can show the redox condition of its past environment and much lower $Fe^{2+}(Ⅵ)/Fe^{3+}(Ⅵ)$ ratios in montmorillonite of the Chunbook Conglomerate imply the greater oxidizing influence. Calculated burial depths from oxygen stable isotope data of the samples from the Chunbook Conglomerate generally fall to the range of 929~963 m whereas the real burial depth of this area is only 530~580 m. This could be explained as the bentonites of the Chunbook conglomerate had not been formed in situ. Discriminant analyses with the data from chemical analyses and structural formulae of montmorillonites show that bentonites from three different basins could definitely be distinguished with each other. This result arises from the different chemical compositions of original volcanic ashes and the difference of sedimentary environments.

• Journal of the Korean earth science society
• /
• v.22 no.3
• /
• pp.163-178
• /
• 2001

Granite gneiss, pophyroblastic gneiss and leucocratic gneiss are widely distributed in the Seungju-Suncheon area, the southwestern part of the Sobacksan Massif, Korea. These orthogneisses show intrusive relationships in outcrops of the study area. This study focuses on the geochemical properties and the tectonic environments for the original rocks of these orthogneisses. The pophyroblastic gneiss is plotted in diorite and granodiorite domain, and granite gneiss and leucocratic gneiss are plotted in both of granodiorite and granite domains on lUGS silica-alkali diagram. Geochemical properies of major elements suggest that these rocks are sub-alkali rock series, and were formed from S-type magma which generated in syn-collision tectonic environment. Discrimination diagrams using HFS elements suggest that original rocks of the three orthogneisses were granitoid of calc-alkali rock series, and were formed in syn-collision environment.

• Journal of Digital Convergence
• /
• v.13 no.12
• /
• pp.117-124
• /
• 2015

As the acceptance of statistical analysis has been increased because of Big Data, the needs for an advanced second generation of statistical analysis method like Structural Equation Model are also increasing. This study suggests how R-Program, as open software, can be utilized when Partial Least Square Model, one of the SEMs, is applied to statistical analysis. R is a free software as a part of GNU projects as well as a powerful and useful tool for statistical analysis including Big Data. The study utilized R and SmartPLS, a representative statistical package of PLS-SEM, and analyzed internal consistency reliability, convergent validity, and discriminant validity of the measurement model. The study also analyzed path coefficients and moderator effects of the structural model and compared the results, respectively. The results indicated that R showed the same results with SmartPLS on the measurement model and the structural model. Therefore, the study confirmed that R could be a powerful tool that is alternative to a commercial statistical package in the future.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.4
• /
• pp.308-317
• /
• 2009

In the present work, a sensor system composed of an oscillator sensor and a Lamb wave sensor is proposed for the purpose of structural health monitoring. The oscillator sensing system detecting the shift of a structural resonant frequency in proportion to the amount of defects in the structure is a pretty sensitive and simple device, but its detectable range is limited to its local zone. The Lamb wave sensor system, however, is applicable to global detection of the defects. This study is aimed at investigating the feasible combination of the two systems to exploits their merits simultaneously. The scheme to use PZT patches as the oscillator sensor as well as the Lamb wave sensor was proposed to identify the position, length and number of cracks by means of TOF and amplitude of signals, and its validity was confirmed through experiments.