• 제목/요약/키워드: DenseNet

검색결과 149건 처리시간 0.024초

한반도 일원의 중력측정 및 보정의 정밀화와 고밀도 부우게이상 (On the Improvement of Precision in Gravity Surveying and Correction, and a Dense Bouguer Anomaly in and Around the Korean Peninsula)

  • 신영홍;양철수;옥수석;최광선
    • 한국지구과학회지
    • /
    • 제24권3호
    • /
    • pp.205-215
    • /
    • 2003
  • 정밀하고 밀도 높은 부우게이상자료는 지구물리학과 물리측지학적인 측면에서 중요한 기초자료가 된다. 이를 위해서 정밀한 절대중력기준점망의 설치외에도 두 가지 측면을 고려해야 하는데, 하나는 중력측정과 보정의 정밀도를 향상시키는 것이며, 다른 하나는 측점의 밀도를 높이는 일이다. 중력측정에서 GPS를 어떻게 적절히 이용할 것인가에 대해서 알아보았는데, 40km이내의 기선거리에서 두 대의 측지용 GPS수신기를 이용하면, 5분간의 GPS측정으로도 충분하다고 여겨진다. 이 경우 PNU95 지오이드모델과 같은 정밀한 지오이드모델을 사용하여야 한다, 이 방법을 적용함으로써, 비용 및 시간과 인력을 줄일 수 있을 뿐만 아니라 질적인 측면에서도 향상을 가져온다. 또 교점오차를 보정하고 지형효과를 정밀하게 계산하기 위해 컴퓨터 프로그램을 개발했다. 중력탐사에서 같은 지점에서 중복 측정하는 것은 스프링의 변이를 보정하는 것뿐만 아니라 망조정을 적용함으로써 결과들에 대해 통계적으로 접근하는 것을 가능하게 한다. 그럼으로써 다양한 원인에 기인한 큰 오차를 쉽게 찾아내고 보정하여 자료의 질을 향상시킬 수 있으며, 또한 측정의 질을 평가할 수 있다. 최근의 컴퓨터 기술과 디지털 고도자료의 발달도 좀더 정밀한 지형보정을 사용하여 부우게이상의 질을 향상하도록 우리를 고무하고 있다. 본 연구에서는 육상중력, 선상중력, 북한의 부우게이상도, 일본의 중력자료, 고도위성자료, EGM96 지오포텐셜 자료 등, 다양한 중력자료를 입수하였다. 따라서, 이들 자료를 계산하고 편집하여 한반도 일원의 정밀 부우게이상도를 작성할 수 있었는데, 이것은 지구물리학과 물리측지학의 응용에 유용한 자료가 될 것이다.

사회연결망분석과 인공신경망을 이용한 추천시스템 성능 예측 (Predicting the Performance of Recommender Systems through Social Network Analysis and Artificial Neural Network)

  • 조윤호;김인환
    • 지능정보연구
    • /
    • 제16권4호
    • /
    • pp.159-172
    • /
    • 2010
  • 협업필터링 추천은 다양한 분야에서 활용되고 있지만 트랜잭션 데이터의 성격에 따라 추천 성능에 현저한 차이를 보이고 있다. 기존 연구에서는 이러한 추천 성능의 차이가 나타나는 이유에 대한 설명을 구체적으로 제시하지 못하고 있고 이에 따라 추천 성능의 예측 또한 연구된 바가 없다. 본 연구는 사회네트워크분석과 인공신경망 모형을 이용하여 협업필터링 추천시스템의 성능을 예측하고자 한다. 본 연구의 목적을 달성하기 위해 국내 백화점의 트랜잭션 데이터를 기반으로 형성되는 고객간 사회 네트워크의 구조적 지표를 측정한 후 이를 기반으로 인공신경망 모형을 구축하고 검증한다. 본 연구는 협업필터링 추천 성능을 예측할 수 있는 새로운 모형을 제시하였다는 점에서 그 의의가 있으며 이를 통해 기업들의 협업필터링 추천시스템 도입에 대한 의사결정에 도움을 줄 수 있을 것으로 기대된다.

동중국해에 있어서 어족생물의 일주기적 유영행동특성에 관한 연구-I-획야 전이시간대의 유영행동특성- (Hydroacoustic Observations on the Diel Distribution and Activity Patterns of Fishes in the East China Sea II -Activity Patterns during the Evening and Morning Transition Periods -)

  • 이대재
    • 수산해양기술연구
    • /
    • 제30권4호
    • /
    • pp.239-250
    • /
    • 1994
  • The vertical distribution and activity patterns of fishes during the evening and morning transitions between day and night were studied acoustically and by bottom trawling in November 1990-1992 in thermally stratified waters of the East China Sea. The acoustic data were collected from six stations with a scientific echo-sounder operating at two frequencies of 25 and 100kHz, and the echograms were used to determine the vertical distributions of fish. Biological sampling was accomplished by bottom trawling to identify fish species recorded on the echograms, and the species and length compositions were determined. At each station, vertical profiles of water temperature, salinity and dissolved oxygen were taken with a CTD system and were related to the diel movements and the depth distributions of fish. During the day most fish were within several meters above bottom, but began to migrate upwards just before sunset, and during the night they were dispersed in midwater. Prior to sunrise with a thermocline present, one group of the fish aggregation occurred in dense schools slightly above the thermocline, while the other group occurred with the numerous single fish-traces bellow it. These groups of aggregations rapidly began to migrate toward the bottom across the thermocline from about 40 min before sunrise. Trawl hauls in the bottom strata below the thermocline with the characteristic single fish traces yieled invariably catches dominated by snailfish and fishing frog with minor quantities of other species in all stations. Hence, the results indicate that snailfish and fishing frog were the dominated scatterers in the depth strata below the thermocline, and the single-fish recordings were mainly snailfish. The fish species such as anchovy and juvenile mackerel in bottom trawl catches is poorly represented in relation to the mesh selectivity of the trawl net, but their occurrence suggest that the fish-school recording above the thermocline were due to these species which migrated vertically across the thermocline, with a temperature gradient of about 8$^{\circ}C$, from the water layers near the bottom at night. Accordingly, we conclude that the vertical distribution and activity patterns of snailfish were strongly temperature dependent and in the termally stratified waters, the upper limit to diel activity was closely linked to the position of the thermocline.

  • PDF

레가시 시스템의 컴포넌트화 방법론 개발을 위한 프로그램 분석 활동 (Program Analysis Activities for Development of Componentization Methodology for of Legacy System)

  • 차정은;김철홍;양영종
    • 한국정보처리학회:학술대회논문집
    • /
    • 한국정보처리학회 2002년도 추계학술발표논문집 (하)
    • /
    • pp.2119-2122
    • /
    • 2002
  • 기업의 비즈니스 프로세스가 복잡, 다양해짐에 따라, 현재 운영 시스템에 대한 급격한 기술적 변화를 수용하고 이를 조직적 측면의 기업 프로세스로 적용하기 위해 레가시 시스템의 현대화가 요구된다. 따라서, 현재의 기업들은 다양한 사용자들이 각자 그들의 관점에서 필요한 비즈니스 요구들을 웹 상에서 처리시킬 수 있도록 J2EE, .NET 등으로 대표되는 컴포넌트 및 웹 서비스 기술을 적용한 새로운 e-business 환경을 수용해야만 한다. 하지만, 기업 조직의 중요한 지식과 프로세스들을 처리하는 시스템들은 대부분 과거(Legacy)의 기술에 의해 개발되어졌으며, 이러한 시스템들을 새로운 비즈니스 환경에 적용하기에는 웹 환경을 위한 분산 아키텍쳐의 결여와 개방성과 표준화 미흡으로 시스템의 유지보수에 많은 어려움을 가진다. 또한, 방법론 차원에서 재공학의 절차와 기법을 체계적으로 정의하고 지원하기 위한 노력이 매우 부족한 실정이다. 본 논문에서는 레거시 시스템을 새로운 시스템 환경으로의 변환 및 통합을 위한 재공학 방법론 개발을 목적으로 프로그램 분석 활동을 설명한다. 본 논문에서 개발하고자 하는 방법론은 다양한 추상화 수준에서 역공학 정보를 복구하고, 컴포넌트화 단계를 통해 새로운 시스템으로 진화할 수 있는 절차 및 기법들을 제공한다. 레거시 시스템 컴포넌트화 방법론은 변환계획 단계, 역공학 단계, 컴포넌트화 단계, 인도 단계의 4 단계로 구성되어 있으며, 본 논문에서는 전체 단계 중 가장 기초가 되고 중요한 단계인 역공학 단계에 초점을 두고 프로그램 분석을 위한 절차 및 과정의 주요 지침들을 제시한다.0보다 유의적으로 우수하였으며, 맛, 다즙성 및 전체적인 기호도는 TMR-0 및 TMR-1 사이에 유의성이 없었다(p<0.05).能性)을 알아보고자 본(本) 실험(實驗)을 실시(實施)한 결과(結果), 유의성(有意性) 있는 결과(結果)를 얻었기에 보고(報告)하는 바이다.또한 이들 상피세포(上皮細胞)들을 투과전자현미경적(透過電子顯微鏡的)으로 관찰(觀察)하였을 때 초미세구조(超微細構造)가 잘 보존(保存)되었으나 Langer-hans cell내(內)의 Birbeck granule은 유리전(遊離前) 피부상피조직내(皮膚上皮組織內)의 Langerhans cell내(內)의 Birbeck granule에 비(比)해 수적(數的)으로 현저히 감소(減少)되어 있었다. 그러나 Thy-1 양성(陽性) dendritic cell에서 볼 수 있는 dense-core 과립(顆粒)은 별변화(別變化)없이 쉽게 관찰(觀察)될 수 있었다. 조직배양(組織培養)을 한 견(犬)의 keratinocyte에 대(對)해 사람 pemphigus vulgaris의 항체(抗體)로 반응(反應)시킨 후 protein-A gold(15 nm)로 표식(標識)시킨 바 제일 바깥 상층(上層)의 keratinocyte에 있어서 세포막표면(細胞膜表面)을 따라 표식(標識)되어 세포막항원(細胞膜抗元)을 나타내었으며, 이와 같은 소견(所見)으로 미루어 정상피부(正常皮膚) 중층편평상피세포(重層扁平上皮細胞)에서도 동일(同一)한 소견(所見)을 관찰(觀察)할 수 있다고 본다.al remnants, Resorption of fetus로 관찰된 것이다. Fetal death는 수정후

  • PDF

Field Programmable Stateful Logic Array 패브릭 매핑 및 배치 (Fabric Mapping and Placement of Field Programmable Stateful Logic Array)

  • 김교선
    • 전자공학회논문지
    • /
    • 제49권12호
    • /
    • pp.209-218
    • /
    • 2012
  • 최근 무어의 법칙을 연장시킬 시스템 집적 기술로서 Field Programmable Stateful Logic Array (FPSLA)가 제안되었다. 본 논문은 FPSLA의 설계 자동화 절차를 확립하고 논리 합성, 동기화, 물리적 매핑, 자동 배치 등의 접근 방법을 최초로 제시한다. 특히, 동기화를 통해 배치를 1차원 문제로 축소한 후 비선형 최적화 기법을 개량한 개략 배치 모델 및 하향식 계층적 2분법을 이용한 배치 적법화 알고리즘을 제안하였다. 또한, 제안된 모델 및 알고리즘을 소프트웨어로 구현하여 ACM/SIGDA 벤치 마크 예제에 적용함으로써 그 유효성을 입증하였다. 이 소프트웨어에는 Fanout 수만큼 출력 상태를 같은 단의 멤리스터성 스위치에 복사해야 하는 FPSLA의 특성을 고려하여 최적화 단계 별로 넷을 하이퍼에지로 통합했다가 다시 에지로 분리하는 기법이 제안되었으며 약 18.4%의 추가적 최적화를 이룩했다. FPSLA의 출력 상태 복사는 논리 단 일부에 셀 밀도가 집중되는 문제를 노출했으며 단위 논리 게이트의 Fanin을 제한하는 기법으로 18.5% 감소 효과를 얻었다. FPSLA의 실용성 확보를 위해서는 우선 논리 합성 시 Fanin의 수가 일부 단에 집중되지 않도록 제약하는 방안을 개발하여야 한다. 또한, FPSLA 패브릭 구조를 이식하기 위해 대칭성이 감소된 나노와이어 크로스바가 형성하는 복잡한 그래프 상에서 수행되어야 하는 자동 배선의 효율성 연구도 필요하다. 이러한 툴 개발은 설계 자동화 자체뿐만 아니라 FPSLA의 패브릭 구조 개선에 필요한 실험에 유용한 평가 도구로서도 큰 역할을 할 것이다.

Dry matter and grain production of a near-isogenic line carrying a 'Takanari' (high yielding, Indica) allele for increased leaf inclination angle in rice with the 'Koshihikari' (Japonica) genetic background

  • San, Nan Su;Otsuki, Yosuke;Adachi, Shunsuke;Yamamoto, Toshio;Ueda, Tadamasa;Tanabata, Takanari;Ookawa, Taiichiro;Hirasawa, Tadashi
    • 한국작물학회:학술대회논문집
    • /
    • 한국작물학회 2017년도 9th Asian Crop Science Association conference
    • /
    • pp.32-32
    • /
    • 2017
  • To increase rice production, manipulating plant architecture, especially developing new high-yielding cultivars with erect leaves, is crucial in rice breeding programs. Leaf inclination angle determines the light extinction coefficient (k) of the canopy. Erect leaves increase light penetration into the canopy and enable dense plantings with a high leaf area index, thus increasing biomass production and grain yield. Because of erect leaves, the high-yielding indica rice cultivar 'Takanari' has smaller k during ripening than 'Koshihikari', a japonica cultivar with good eating quality. In our previous study, using chromosome segment substitution lines (CSSLs) derived from a cross between 'Takanari' and 'Koshihikari', we detected seven quantitative trait loci (QTLs) for leaf inclination angle on chromosomes 1 (two QTLs), 2, 3, 4, 7, and 12. In this study, we developed a near-isogenic line (NIL-3) carrying a 'Takanari' allele for increased leaf inclination angle on chromosome 3 in the 'Koshihikari' genetic background. We compared k, dry matter production, and grain yield of NIL-3 with those of 'Koshihikari' in the field from 2013 to 2016. NIL-3 had higher inclination angles of the flag, second, and third leaves at full heading and 3 (- 4) weeks after full heading and smaller k of the canopy at the ripening stage. Biomass at full heading and leaf area index at full heading and at harvest did not significantly differ between NIL-3 and 'Koshihikari'. However, biomass at harvest was significantly greater in NIL-3 than in 'Koshihikari' due to a higher net assimilation rate at the ripening stage. The photosynthetic rates of the flag and third leaves did not differ between NIL-3 and Koshihikari at ripening. Grain yield was higher in NIL-3 than 'Koshihikari'. Higher panicle number per square meter in NIL-3 contributed to the higher grain yield of NIL-3. We conclude that the QTL on chromosome 3 increases dry matter and grain production in rice by increasing leaf inclination angle.

  • PDF

A REVIEW ON THE ODSCC OF STEAM GENERATOR TUBES IN KOREAN NPPS

  • Chung, Hansub;Kim, Hong-Deok;Oh, Seungjin;Boo, Myung Hwan;Na, Kyung-Hwan;Yun, Eunsup;Kang, Yong-Seok;Kim, Wang-Bae;Lee, Jae Gon;Kim, Dong-Jin;Kim, Hong Pyo
    • Nuclear Engineering and Technology
    • /
    • 제45권4호
    • /
    • pp.513-522
    • /
    • 2013
  • The ODSCC detected in the TSP position of Ulchin 3&4 SGs are typical ODSCC of Alloy 600MA tubes. The causative chemical environment is formed by concentration of impurities inside the occluded region formed by the tube surface, egg crate strips, and sludge deposit there. Most cracks are detected at or near the line contacts between the tube surface and the egg crate strips. The region of dense crack population, as defined as between $4^{th}$ and $9^{th}$ TSPs, and near the center of hot leg hemisphere plane, coincided well with the region of preferential sludge deposition as defined by thermal hydraulics calculation using SGAP computer code. The cracks developed homogeneously in a wide range of SGs, so that the number of cracks detected each outage increased very rapidly since the first detection in the $8^{th}$ refueling outage. The root cause assessment focused on investigation of the difference in microstructure and manufacturing residual stress in order to reveal the cause of different susceptibilities to ODSCC among identical six units. The manufacturing residual stress as measured by XRD on OD surface and by split tube method indicated that the high residual stress of Alloy 600MA tube played a critical role in developing ODSCC. The level of residual stress showed substantial variations among the six units depending on details of straightening and OD grinding processes. Youngwang 3&4 tubes are less susceptible to ODSCC than U3 and U4 tubes because semi-continuous coarse chromium carbides are formed along the grain boundary of Y3&4 tubes, while there are finer less continuous chromium carbides in U3 and U4. The different carbide morphology is caused by the difference in cooling rate after mill anneal. There is a possibility that high chromium content in the Y3&4 tubes, still within the allowable range of Alloy 600, has made some contribution to the improved resistance to ODSCC. It is anticipated that ODSCC in Y5&6 SGs will be retarded more considerably than U3 SGs since the manufacturing residual stress in Y5&6 tubes is substantially lower than in U3 tubes, while the microstructure is similar with each other.

Accuracy of one-step automated orthodontic diagnosis model using a convolutional neural network and lateral cephalogram images with different qualities obtained from nationwide multi-hospitals

  • Yim, Sunjin;Kim, Sungchul;Kim, Inhwan;Park, Jae-Woo;Cho, Jin-Hyoung;Hong, Mihee;Kang, Kyung-Hwa;Kim, Minji;Kim, Su-Jung;Kim, Yoon-Ji;Kim, Young Ho;Lim, Sung-Hoon;Sung, Sang Jin;Kim, Namkug;Baek, Seung-Hak
    • 대한치과교정학회지
    • /
    • 제52권1호
    • /
    • pp.3-19
    • /
    • 2022
  • Objective: The purpose of this study was to investigate the accuracy of one-step automated orthodontic diagnosis of skeletodental discrepancies using a convolutional neural network (CNN) and lateral cephalogram images with different qualities from nationwide multi-hospitals. Methods: Among 2,174 lateral cephalograms, 1,993 cephalograms from two hospitals were used for training and internal test sets and 181 cephalograms from eight other hospitals were used for an external test set. They were divided into three classification groups according to anteroposterior skeletal discrepancies (Class I, II, and III), vertical skeletal discrepancies (normodivergent, hypodivergent, and hyperdivergent patterns), and vertical dental discrepancies (normal overbite, deep bite, and open bite) as a gold standard. Pre-trained DenseNet-169 was used as a CNN classifier model. Diagnostic performance was evaluated by receiver operating characteristic (ROC) analysis, t-stochastic neighbor embedding (t-SNE), and gradient-weighted class activation mapping (Grad-CAM). Results: In the ROC analysis, the mean area under the curve and the mean accuracy of all classifications were high with both internal and external test sets (all, > 0.89 and > 0.80). In the t-SNE analysis, our model succeeded in creating good separation between three classification groups. Grad-CAM figures showed differences in the location and size of the focus areas between three classification groups in each diagnosis. Conclusions: Since the accuracy of our model was validated with both internal and external test sets, it shows the possible usefulness of a one-step automated orthodontic diagnosis tool using a CNN model. However, it still needs technical improvement in terms of classifying vertical dental discrepancies.

COATED PARTICLE FUEL FOR HIGH TEMPERATURE GAS COOLED REACTORS

  • Verfondern, Karl;Nabielek, Heinz;Kendall, James M.
    • Nuclear Engineering and Technology
    • /
    • 제39권5호
    • /
    • pp.603-616
    • /
    • 2007
  • Roy Huddle, having invented the coated particle in Harwell 1957, stated in the early 1970s that we know now everything about particles and coatings and should be going over to deal with other problems. This was on the occasion of the Dragon fuel performance information meeting London 1973: How wrong a genius be! It took until 1978 that really good particles were made in Germany, then during the Japanese HTTR production in the 1990s and finally the Chinese 2000-2001 campaign for HTR-10. Here, we present a review of history and present status. Today, good fuel is measured by different standards from the seventies: where $9*10^{-4}$ initial free heavy metal fraction was typical for early AVR carbide fuel and $3*10^{-4}$ initial free heavy metal fraction was acceptable for oxide fuel in THTR, we insist on values more than an order of magnitude below this value today. Half a percent of particle failure at the end-of-irradiation, another ancient standard, is not even acceptable today, even for the most severe accidents. While legislation and licensing has not changed, one of the reasons we insist on these improvements is the preference for passive systems rather than active controls of earlier times. After renewed HTGR interest, we are reporting about the start of new or reactivated coated particle work in several parts of the world, considering the aspects of designs/ traditional and new materials, manufacturing technologies/ quality control quality assurance, irradiation and accident performance, modeling and performance predictions, and fuel cycle aspects and spent fuel treatment. In very general terms, the coated particle should be strong, reliable, retentive, and affordable. These properties have to be quantified and will be eventually optimized for a specific application system. Results obtained so far indicate that the same particle can be used for steam cycle applications with $700-750^{\circ}C$ helium coolant gas exit, for gas turbine applications at $850-900^{\circ}C$ and for process heat/hydrogen generation applications with $950^{\circ}C$ outlet temperatures. There is a clear set of standards for modem high quality fuel in terms of low levels of heavy metal contamination, manufacture-induced particle defects during fuel body and fuel element making, irradiation/accident induced particle failures and limits on fission product release from intact particles. While gas-cooled reactor design is still open-ended with blocks for the prismatic and spherical fuel elements for the pebble-bed design, there is near worldwide agreement on high quality fuel: a $500{\mu}m$ diameter $UO_2$ kernel of 10% enrichment is surrounded by a $100{\mu}m$ thick sacrificial buffer layer to be followed by a dense inner pyrocarbon layer, a high quality silicon carbide layer of $35{\mu}m$ thickness and theoretical density and another outer pyrocarbon layer. Good performance has been demonstrated both under operational and under accident conditions, i.e. to 10% FIMA and maximum $1600^{\circ}C$ afterwards. And it is the wide-ranging demonstration experience that makes this particle superior. Recommendations are made for further work: 1. Generation of data for presently manufactured materials, e.g. SiC strength and strength distribution, PyC creep and shrinkage and many more material data sets. 2. Renewed start of irradiation and accident testing of modem coated particle fuel. 3. Analysis of existing and newly created data with a view to demonstrate satisfactory performance at burnups beyond 10% FIMA and complete fission product retention even in accidents that go beyond $1600^{\circ}C$ for a short period of time. This work should proceed at both national and international level.