• Title/Summary/Keyword: Data Variability

Search Result 1,619, Processing Time 0.029 seconds

Speaker-Independent Korean Digit Recognition Using HCNN with Weighted Distance Measure (가중 거리 개념이 도입된 HCNN을 이용한 화자 독립 숫자음 인식에 관한 연구)

  • 김도석;이수영
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.18 no.10
    • /
    • pp.1422-1432
    • /
    • 1993
  • Nonlinear mapping function of the HCNN( Hidden Control Neural Network ) can change over time to model the temporal variability of a speech signal by combining the nonlinear prediction of conventional neural networks with the segmentation capability of HMM. We have two things in this paper. first, we showed that the performance of the HCNN is better than that of HMM. Second, the HCNN with its prediction error measure given by weighted distance is proposed to use suitable distance measure for the HCNN, and then we showed that the superiority of the proposed system for speaker-independent speech recognition tasks. Weighted distance considers the differences between the variances of each component of the feature vector extraced from the speech data. Speaker-independent Korean digit recognition experiment showed that the recognition rate of 95%was obtained for the HCNN with Euclidean distance. This result is 1.28% higher than HMM, and shows that the HCNN which models the dynamical system is superior to HMM which is based on the statistical restrictions. And we obtained 97.35% for the HCNN with weighted distance, which is 2.35% better than the HCNN with Euclidean distance. The reason why the HCNN with weighted distance shows better performance is as follows : it reduces the variations of the recognition error rate over different speakers by increasing the recognition rate for the speakers who have many misclassified utterances. So we can conclude that the HCNN with weighted distance is more suit-able for speaker-independent speech recognition tasks.

  • PDF

A Case Study on Stochastic Fracture Network Modeling for Rock Slopes of Busan-Ulsan Highway(Reach 5) (부산-울산 고속국도(5공구)에 위치한 암반사면의 추계론적 절리연결구조 모사에 대한 사례연구)

  • Heo, In-Sill;Um, Jeong-Gi;Kim, Yang-Phil;Kim, Kook-Han;Lee, Young-Kyun
    • The Journal of Engineering Geology
    • /
    • v.16 no.4 s.50
    • /
    • pp.337-349
    • /
    • 2006
  • Seven hundred and fifty one fractures of the rhyolitic tuffaceous rock masses were mapped using 6 scanlines placed on rock slope exposures that were within 8.02 km of Busan-Ulsan highway. These data were analyzed to find the number of fracture sets that exist in the rock slopes and the probability distributions of orientation, spacing, trace length and fracture size in 3-D for each of the fracture sets. All the fracture set orientation distributions exhibit high variability. The Fisher distributions were found to be unsuitable to represent the statistical distribution of orientation for most of the fracture sets. The probability distributions, gamma, exponential and lognormal were found to be highly suitable to represent the distribution of spacing and semi-trace length of fracture sets. In obtain-ing these distributions, corrections were applied for sampling biases associated with spacing and trace length. The generated fracture system in 3-D was used to make predictions of fracture traces for each fracture set on 2-D win-dows. Developed stochastic 3-D fracture network for the rock mass was validated by comparing statistical proper-ties of the observed fracture traces on scanlines with the predicted fracture traces on the scanlines. This exercise fumed out to be successful.

Analyzing on the Fluctuation Characteristics of Management Condition of Construction Company (건설업체 경영상태 변동에 대한 특성 분석)

  • Jang, Ho-Myun
    • Journal of the Korea Academia-Industrial cooperation Society
    • /
    • v.15 no.2
    • /
    • pp.1118-1125
    • /
    • 2014
  • The past IMF foreign exchange crisis and subprime financial crisis had a big influence on variability of macroeconomics, even if the origin of its occurrence might be different. This not only had a significant infrequence on the overall industries, but also produced many insolvent companies by being closely linked with a management environment of an individual construction company leading the construction industry. The purpose of this research is to investigate characteristics of management condition of construction company according to the size of construction company using KMV model developed on the basis of the Black & Scholes option pricing theory. This research has set 28 construction companies listed to KOSPI/KOSDAQ for applying the KMV model and measuring the level of the default risk of construction companies. The data was retrieved from TS2000 established by Korea Listed Companies Association (KLCA), Statistics Korea. The analysis period is between first quarter of 2004 and fourth quarter of 2010. This research examine characteristics of the level and fluctuation process of the management condition of construction company according to the size of construction company.

Estimation of Aboveground Forest Biomass Carbon Stock by Satellite Remote Sensing - A Comparison between k-Nearest Neighbor and Regression Tree Analysis - (위성영상을 활용한 지상부 산림바이오매스 탄소량 추정 - k-Nearest Neighbor 및 Regression Tree Analysis 방법의 비교 분석 -)

  • Jung, Jaehoon;Nguyen, Hieu Cong;Heo, Joon;Kim, Kyoungmin;Im, Jungho
    • Korean Journal of Remote Sensing
    • /
    • v.30 no.5
    • /
    • pp.651-664
    • /
    • 2014
  • Recently, the demands of accurate forest carbon stock estimation and mapping are increasing in Korea. This study investigates the feasibility of two methods, k-Nearest Neighbor (kNN) and Regression Tree Analysis (RTA), for carbon stock estimation of pilot areas, Gongju and Sejong cities. The 3rd and 5th ~ 6th NFI data were collected together with Landsat TM acquired in 1992, 2010 and Aster in 2009. Additionally, various vegetation indices and tasseled cap transformation were created for better estimation. Comparison between two methods was conducted by evaluating carbon statistics and visualizing carbon distributions on the map. The comparisons indicated clear strengths and weaknesses of two methods: kNN method has produced more consistent estimates regardless of types of satellite images, but its carbon maps were somewhat smooth to represent the dense carbon areas, particularly for Aster 2009 case. Meanwhile, RTA method has produced better performance on mean bias results and representation of dense carbon areas, but they were more subject to types of satellite images, representing high variability in spatial patterns of carbon maps. Finally, in order to identify the increases in carbon stock of study area, we created the difference maps by subtracting the 1992 carbon map from the 2009 and 2010 carbon maps. Consequently, it was found that the total carbon stock in Gongju and Sejong cities was drastically increased during that period.

Isoflavone Distribution and $\beta$-Glucosidase Activity in Home-made and Factory-produced Doenjang (재래식 및 개량식 된장의 아이소플라본 분포 및 $\beta$-glucosidase 활성 연구)

  • Lee, Seung-Wook;Park, Yong-Woo;Chang, Pahn-Shick;Lee, Jae-Hwan
    • Korean Journal of Food Science and Technology
    • /
    • v.42 no.2
    • /
    • pp.125-129
    • /
    • 2010
  • Isoflavone distribution and $\beta$-glucosidase activity in 16 commercially available doenjang samples were determined. Twelve of the samples were home-made doenjang (HMD) with a relatively long fermentation period and 4 of the samples were factory-produced doenjang (FPD) from four different manufactures. Total isoflavones (TI) in the HMD ranged from $370-723\;{\mu}g/g$, while those in FPD ranged from $179-537\;{\mu}g/g$. The isoflavone distribution in HMD was different from those in FPD. Generally, the TI in HMD was higher than those in FPD. The major isoflavone was aglycones, which ranged from 42.98 to 89.96% in HMD and from 35.51 to 93.48% in FPD. Isoflavones in the $\beta$-glucoside forms were not detected in tested FPD samples. Principal components analysis (PCA) of the isoflavone profiles showed that HMD were differentiated from FPD. First principal component (PC1) and second principal component (PC2) expressed 43.6 and 22.9% of the data variability, respectively. $\beta$-Glucosidase activity in doenjang was lower than that in raw soybeans. The results of this study can be used to understand the differences in the isoflavone distribution in traditionally manufactured and factory produced doenjang.

Quality Improvement in Patient Care Services : Obstacles and Approaches (진료의 질관리에 대한 시론 -장애와 접근-)

  • 한달선
    • Health Policy and Management
    • /
    • v.2 no.2
    • /
    • pp.112-130
    • /
    • 1992
  • Patient care services are provided to individual patients in response to their health needs produced by illnesses or injuries. The services are often addressed to very serious conditions, and also they constitute the most expensive component of health care services. Therefore, the importance of quality is emphasized, but there are many indications that patient care quality is far from a satisfactory state in most of the countries. Based upon this observation, it is attempted to examine obstacles and approaches to quality improvement in patient care services. In doing so, following Taguchi's(1986) definition of product quality, quality of patient care services is conceived of as better when the less is the sociental loss attributalbe to variability of intended function and harmful side effects they emhibit after being delivered. Some distinguishing features of medical care sector pose difficulties in implementing effective quality improvement programs in patient care services. Nevertheless, newly proposed method of quality management, based on industrial quality management approach, seems to have a great deal of potential to effectively cope with such difficulties. This method, unlike the traditional approach to quality assurance, focuses on total organisational processes, not individuals, as the obproach to quality assurance, focuses on total organizational processes, not individuals, as the objects of quality improvement; variation, not comparison with standards, in quality measurement; and continuous improvement, not removing only bad quality care, as an ideal. Prerequisite to a successful use of any quality mangement method is motivating providers to improve quality. Conceivable approaches for such motivation are self-regulation, external controls and promotion of competition. Since these approaches are not mutually exclusive, they may be employed in an appropriate combination. In Korea, medical care providers are now functioning under the circumstances where they have little reason for making efforts to improve quality of their services. Once these circumstantial conditions are changed to exert pressures on providers to improve quality, the use of adequate quality management method becomes an issue. In this connection, much attention shoould be directed to the newly proposed method described above. In all these efforts for improving quality of patient care services, health insurance would be able to play a pivotal role. Poviders of medical care, buth indiciduals and organizations, are usually very responsive to the measures that affect their financing, and thus health insurance can be a strong instrument for motivationg providers to improve quality. Also, the insurance continuously acquires data on patient care, which could be processed to produce information required to effective quality control.

  • PDF

Estimation of Local Surface Temperature from EBM with the Use of GRID/GIS and Remote Sensed data (GRID/GIS 및 RS 자료를 이용한 에너지 평형 모형으로부터의 국지적 지표 온도 산출)

  • 신선희;하경자;김재환;오현미;조명희
    • Korean Journal of Remote Sensing
    • /
    • v.20 no.2
    • /
    • pp.103-116
    • /
    • 2004
  • The mesoscale atmospheric models to produce surface temperature can not generally consider the effect of the sloped terrain for direct solar radiation. These have not showed the regional difference of solar radiation and as a result, have made the big error in the local surface temperature. Therefore, we wished to represent the exact locality of surface temperature by considering the geometric properties of surface as well as the vegetated properties of surface. The purpose of the study is to produce local surface ground temperature in sloped terrain diagnostically using surface Energy Balance Model (EBM) with the use of GRID model in Geographic Information Systems (GIS). In this study, surface inhomogeneity over southeastern part of Korean peninsula are considered in estimation of the absorbed surface solar radiation in terms of the illumination angle, depending on topographical aspect and slope in GRID. Also, the properties of vegetated surface which the major components for the variability of surface temperature are considered in terms of NDVI. The results of our study show the locally changes in the surface ground temperature due to local ground aspect and slope effect and local properties of vegetated surface. The more detailed distribution of local surface temperature may drive the local circulation at lower atmospheric and it may explain better the real local circulation.

Experimental study on peak flow variability using spatially extended data (공간확장자료를 사용한 첨두유량의 변동성에 대한 실험적 연구)

  • Kim, Nam Won;Shin, Mun-Ju
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2017.05a
    • /
    • pp.365-365
    • /
    • 2017
  • 강우의 공간분포는 홍수수문곡선에 큰 영향을 미친다. 이것은 총강우량이 같더라도 강우의 공간분포 특성에 따라 홍수수문곡선 및 첨두유량의 크기가 다를 수 있다는 것을 의미한다. 따라서 강우의 공간분포는 홍수량 산정과 홍수해석은 물론 수자원 설계에 큰 영향을 미칠 수 있다. 본 연구에서는 안동댐유역을 대상으로 물리적 기반의 분포형 모형인 GRM 모형과 공간확장자료 생성방법을 사용하여 47개 미계측유역에 대해 홍수유출 시계열자료를 생성하고 3개 관측유역을 포함한 총 50개 유역에 대해 첨두유량을 추출하여 분석함으로써 강우의 공간분포가 유출에 미치는 영향을 실제유역과 실제사상에 대해 자세히 분석하였다. 공간확장자료 생성방법은 보정 및 검증유역에서 모의유량이 관측유량을 적절히 모의한다면 추정된 매개변수값들을 상류지역에 있는 미계측유역에 적용하여 유량자료를 생성하는 방법이다. 1989년부터 2009년까지의 강우와 유출자료의 질이 좋은 20개의 사상을 추출하고 이 사상들과 3개 관측유역에 대해 GRM 모형의 매개변수들을 보정 및 검증하였다. 그 결과 NSE > 0.5, PBIAS ${\pm}30%$, 그리고 수정상관계수인 $_{mod}>0.6$의 적절한 모형효율 통계결과를 얻었다. 강우의 공간분포가 유출에 미치는 영향을 조사하기 위해 이 추정된 매개변수와 실제강우(강우의 공간분포를 고려한 강우) 및 공간평균강우(실제강우를 공간적으로 평균한 강우)를 사용하여 50개 유역의 홍수유출 시계열자료를 생성하였으며 이 시계열 자료 중 첨두유량을 추출하여 분석하였다. 그 결과 공간평균강우에 의한 첨두유량의 분포는 실제강우에 의한 첨두유량의 분포와 차이가 있었다. 20개 사상중 13개의 사상은 실제강우와 공간평균강우에 의한 첨두유량의 분포가 비슷하거나 공간평균강우에 의한 첨두유량의 분포가 실제강우에 의한 첨두유량의 분포보다 약간 좁아지는 차이가 있었다. 하지만 나머지 7개 사상의 경우에는 공간평균강우에 의한 첨두유량의 분포가 실제강우에 의한 첨두유량의 분포보다 크게 좁아지는 것을 보였다. 이것은 전체사상의 약 35 %에 대해서는 강우의 공간적 변동성을 고려하지 않고 홍수유출을 모의한다면 적절하지 않은 첨두유량 모의결과를 얻을 수 있다는 것을 나타내며 또한 홍수체적에 대해서도 적절하지 않은 모의결과를 얻을 수 있다는 것을 의미한다. 이것은 또한 강우관측소의 밀도가 홍수유출 모의 시 매우 중요하다는 것을 의미한다. 따라서 홍수량 산정 또는 수자원 설계 시 강우의 시간분포 뿐만 아니라 공간분포 또한 고려해야 한다.

  • PDF

Performances of Prognostic Models in Stratifying Patients with Advanced Gastric Cancer Receiving First-line Chemotherapy: a Validation Study in a Chinese Cohort

  • Xu, Hui;Zhang, Xiaopeng;Wu, Zhijun;Feng, Ying;Zhang, Cheng;Xie, Minmin;Yang, Yahui;Zhang, Yi;Feng, Chong;Ma, Tai
    • Journal of Gastric Cancer
    • /
    • v.21 no.3
    • /
    • pp.268-278
    • /
    • 2021
  • Purpose: While several prognostic models for the stratification of death risk have been developed for patients with advanced gastric cancer receiving first-line chemotherapy, they have seldom been tested in the Chinese population. This study investigated the performance of these models and identified the optimal tools for Chinese patients. Materials and Methods: Patients diagnosed with metastatic or recurrent gastric adenocarcinoma who received first-line chemotherapy were eligible for inclusion in the validation cohort. Their clinical data and survival outcomes were retrieved and documented. Time-dependent receiver operating characteristic (ROC) and calibration curves were used to evaluate the predictive ability of the models. Kaplan-Meier curves were plotted for patients in different risk groups divided by 7 published stratification tools. Log-rank tests with pairwise comparisons were used to compare survival differences. Results: The analysis included a total of 346 patients with metastatic or recurrent disease. The median overall survival time was 11.9 months. The patients were different into different risk groups according to the prognostic stratification models, which showed variability in distinguishing mortality risk in these patients. The model proposed by Kim et al. showed relative higher predicting abilities compared to the other models, with the highest χ2 (25.8) value in log-rank tests across subgroups, and areas under the curve values at 6, 12, and 24 months of 0.65 (95% confidence interval [CI]: 0.59-0.72), 0.60 (0.54-0.65), and 0.63 (0.56-0.69), respectively. Conclusions: Among existing prognostic tools, the models constructed by Kim et al., which incorporated performance status score, neutrophil-to-lymphocyte ratio, alkaline phosphatase, albumin, and tumor differentiation, were more effective in stratifying Chinese patients with gastric cancer receiving first-line chemotherapy.

Topographic Variability during Typhoon Events in Udo Rhodoliths Beach, Jeju Island, South Korea (제주 우도 홍조단괴해빈의 태풍 시기 지형변화)

  • Yoon, Woo-Seok;Yoon, Seok-Hoon;Moon, Jae-Hong;Hong, Ji-Seok
    • Ocean and Polar Research
    • /
    • v.43 no.4
    • /
    • pp.307-320
    • /
    • 2021
  • Udo Rhodolith Beach is a small-scale, mixed sand-and-gravel beach embayed on the N-S trending rocky coast of Udo, Jeju Island, South Korea. This study analyzes the short-term topographic changes of the beach during the extreme storm conditions of four typhoons from 2016 to 2020: Chaba (2016), Soulik (2018), Lingling (2019), and Maysak (2020). The analysis uses the topographic data of terrestrial LiDAR scanning and drone photogrammetry, aided by weather and oceanographic datasets of wind, wave, current and tide. The analysis suggests two contrasting features of alongshore topographic change depending on the typhoon pathway, although the intensity and duration of the storm conditions differed in each case. During the Soulik and Lingling events, which moved northward following the western sea of the Jeju Island, the northern part of the beach accreted while the southern part eroded. In contrast, the Chaba and Maysak events passed over the eastern sea of Jeju Island. The central part of the beach was then significantly eroded while sediments accumulated mainly at the northern and southern ends of the beach. Based on the wave and current measurements in the nearshore zone and computer simulations of the wave field, it was inferred that the observed topographic change of the beach after the storm events is related to the directions of the wind-driven current and wave propagation in the nearshore zone. The dominant direction of water movement was southeastward and northeastward when the typhoon pathway lay to the east or west of Jeju Island, respectively. As these enhanced waves and currents approached obliquely to the N-S trending coastline, the beach sediments were reworked and transported southward or northward mainly by longshore currents, which likely acts as a major control mechanism regarding alongshore topographic change with respect to Udo Rhodolith Beach. In contrast to the topographic change, the subaerial volume of the beach overall increased after all storms except for Maysak. The volume increase was attributed to the enhanced transport of onshore sediment under the combined effect of storm-induced long periodic waves and a strong residual component of the near-bottom current. In the Maysak event, the raised sea level during the spring tide probably enhanced the backshore erosion by storm waves, eventually causing sediment loss to the inland area.