• 한국지리정보학회지
• /
• 제18권1호
• /
• pp.156-169
• /
• 2015

본 연구는 SWAT(Soil and Water Assessment Tool) 모형을 이용하여 임하댐 유역($1,355.5km^2$)을 대상으로 RCP(AR5) 기후변화 시나리오에 따른 미래 수문순환 영향을 평가하였다. 3지점의 실측된 유출량을 활용하여 모형의 보정(2002~2007년) 및 검증(2008~2013년)을 실시하였다. 검 보정 결과 결정계수($R^2$)는 0.70~0.85로, Nash-Sutcliffe 모형 효율(NSE)은 0.67~0.82로 분석되어 신뢰성 있는 유출량 모의 결과를 나타내었다. 기후변화 시나리오는 기상청에서 제공하는 HadGEM3-RA 모형의 RCP 4.5 및 8.5 시나리오를 수집하여, 과거 34년(1980-2013, baseline period)의 기상자료를 기준으로 편이보정(Bias Correction)하여 SWAT모형에 적용하였다. 기후변화 분석 결과 강수량과 평균기온이 10.8%, $4.9^{\circ}C$ 증가하였으며, 강수량과 기온의 증가로 증발산 11.2%, 토양수분 1.9%, 지표유출 10.0%, 중간유출 12.1%, 회귀유출 18.2%이 각각 증가함에 따라 총 하천유출량이 11.2% 증가하였다.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제11권8호
• /
• pp.325-330
• /
• 2022

본 논문은 한국 수어에 대하여 수어 인식, 수어 번역, 수어 영상 시분할과 같은 수어에 관한 딥러닝 연구를 위한 데이터셋의 수집 및 실험을 진행하였다. 수어 연구를 위한 어려움은 2가지로 볼 수 있다. 첫째, 손의 움직임과 손의 방향, 표정 등의 종합적인 정보를 가지는 수어의 특성에 따른 인식의 어려움이 있다. 둘째, 딥러닝 연구를 진행하기 위한 학습데이터의 절대적 부재이다. 현재 알려진 문장 단위의 한국 수어 데이터셋은 KETI 데이터셋이 유일하다. 해외의 수어 딥러닝 연구를 위한 데이터셋은 Isolated 수어와 Continuous 수어 두 가지로 분류되어 수집되며 시간이 지날수록 더 많은 양의 수어 데이터가 수집되고 있다. 하지만 이러한 해외의 수어 데이터셋도 방대한 데이터셋을 필요로 하는 딥러닝 연구를 위해서는 부족한 상황이다. 본 연구에서는 한국 수어 딥러닝 연구를 진행하기 위한 대규모의 한국어-수어 데이터셋을 수집을 시도하였으며 베이스라인 모델을 이용하여 수어 번역 모델의 성능 평가 실험을 진행하였다. 본 논문을 위해 수집된 데이터셋은 총 11,402개의 영상과 텍스트로 구성되었다. 이를 이용하여 학습을 진행할 베이스라인 모델로는 수어 번역 분야에서 SOTA의 성능을 가지고 있는 TSPNet 모델을 이용하였다. 본 논문의 실험에서 수집된 데이터셋에 대한 특성을 정량적으로 보이고, 베이스라인 모델의 실험 결과로는 BLEU-4 score 3.63을 보였다. 또한, 향후 연구에서 보다 정확하게 데이터셋을 수집할 수 있도록, 한국어-수어 데이터셋 수집에 있어서 고려할 점을 평가 결과에 대한 고찰로 제시한다.

• Journal of Astronomy and Space Sciences
• /
• 제25권3호
• /
• pp.283-290
• /
• 2008

이 논문에서는 간섭계 구성의 기본이 되는 위성 간 상대위치 추정에 관한 알고리즘을 개발하고 검증하였다. 편대위성 간 상대위치 추정을 실시간으로 수행하기 위해 확장칼만필터 (EKF, Extended Kalman Filter)와 Unscented 칼만필터 (UKF, Unscented Kalman Filter) 를 사용하였다. 칼만 필터를 이용한 상태벡터 (state-vector)의 갱신(update)을 위한 관측 데이터는 시뮬레이션을 통해 얻어진 GPS 위성 신호의 단일차분 (Single Difference)에 대한 값을 사용하였다. 이 연구에서 개발한 알고리즘으로 추정된 편대위성 간 상대위치는 확장칼만필터와 Unscented 칼만필터 모두 참 값으로 가정한 STK(Satellite Tool Kit) 의 시뮬레이션된 관측 값에 대해 ${\pm}1m$ 이내의 오차로 수렴함을 확인하였다. 또한 두 종류의 칼만필터를 이용하여 상대위치 결정을 수행함으로써 비선 형성을 가지는 경우 Unscented 칼만필터의 성능이 상대적으로 우수함을 확인할 수 있었다.

• 한국전자통신학회논문지
• /
• 제15권3호
• /
• pp.465-472
• /
• 2020

최근 들어, 오디오 이벤트 검출을 위하여 다양한 딥뉴럴네트워크 기반의 방법들이 제안되어 왔다. 본 연구에서는 베이스라인 CRNN(Convolutional Recurrent Neural Network) 구조에 attention 방식을 도입함으로서 오디오 이벤트 검출의 성능을 향상시키고자 하였다. 베이스라인 CRNN의 입력단에 context gating을 적용하고 출력단에 attention layer을 추가하였다. 또한, 프레임(frame) 단위의 강전사 레이블(strong label)정보 뿐만 아니라 클립(clip) 단위의 약전사 레이블(weakly label) 오디오 데이터를 이용한 학습을 통하여 보다 나은 성능을 이루고자 하였다. DCASE 2018/2019 Challenge Task 4 데이터를 이용한 오디오 이벤트 검출 실험에서 제안된 attention 기반의 CRNN을 통하여 기존의 CRNN 방식에 비해서 최대 66%의 상대적 F-score 향상을 얻을 수 있었다.

• Journal of Preventive Medicine and Public Health
• /
• 제44권4호
• /
• pp.167-175
• /
• 2011

Objective: This study was designed to describe the trends in body mass index (BMI) during 6 years (2002-2008) and to identify associations between these trends and the amount of physical activity of South Korean career soldiers. Method: This study targeted the 40 993 (38 857 men and 2136 women) of the 58 657 career soldiers who had undergone four (2002, 2004, 2006, and 2008) biennial medical examinations conducted by the National Health Insurance Corporation; 17 664 soldiers with missing data on height, weight, and physical activity were excluded. A linear mixedregression model was used to categorize changes in BMI due to age versus those due to amount of physical activity. Results: Career soldiers experienced significant increases in BMI compared with baseline data gathered in 2002. The increases in each age group were as follows: men aged 20-29: 1.16, men aged 30-39: 0.61, men aged 40-49: 0.05, women aged 20-29: 0.35, women aged 30-39: 0.30, women aged 40 -49: 0.26, and women aged 50-59: 0.21. However, men aged 50 or older showed significant decreases (as high as 0.5) in BMI compared with baseline data obtained in 2002. They also experienced significant decreases in BMI compared with those who reported no physical activity. The differences between baseline and final BMIs were: 0.02 for men exercising 1-2 times per week, -0.07 for men exercising 3-4 times per week, -0.19 for men exercising 5-6 times per week, -0.21 for men exercising seven times per week, -0.05 for women exercising 1-2 times per week, -0.19 for women exercising 3-4 times per week, -0.30 for women exercising 5-6 times per week, and -0.30 for women exercising seven times per week. Conclusions: Obesity in South Korean career soldiers increased markedly between 2002 and 2008, and our data showed that the amount of physical activity was inversely related to increases in BMI. Policies to prevent obesity are needed to reduce this trend.

• Nutrition Research and Practice
• /
• 제8권1호
• /
• pp.103-111
• /
• 2014

This study aims to determine the effect of a trained dedicated dietitian on clinical outcomes among Lebanese hemodialysis (HD) patients: and thus demonstrate a viable developing country model. This paper describes the study protocol and baseline data. The study was a multicenter randomized controlled trial with parallel-group design involving 12 HD units: assigned to cluster A (n = 6) or B (n = 6). A total of 570 patients met the inclusion criteria. Patients in cluster A were randomly assigned as per dialysis shift to the following: Dedicated Dietitian (DD) (n = 133) and Existing Practice (EP) (n = 138) protocols. Cluster B patients (n = 299) received Trained Hospital Dietitian (THD) protocol. Dietitians of the DD and THD groups were trained by the research team on Kidney Disease Outcomes Quality Initiative nutrition guidelines. DD protocol included: individualized nutrition education for 2 hours/month/HD patient for 6 months focusing on renal osteodystrophy and using the Trans-theoretical theory for behavioral change. EP protocol included nutrition education given to patients by hospital dietitians who were blinded to the study. The THD protocol included nutrition education to patients given by hospital dietitian as per the training received but within hospital responsibilities, with no set educational protocol or tools. Baseline data revealed that 40% of patients were hyperphosphatemics (> 5.5 mg/dl) with low dietary adherence and knowledge of dietary P restriction in addition to inadequate daily protein intake ($58.86%{\pm}33.87%$ of needs) yet adequate dietary P intake ($795.52{\pm}366.94$ mg/day). Quality of life (QOL) ranged from 48-75% of full health. Baseline differences between the 3 groups revealed significant differences in serum P, malnutrition status, adherence to diet and P chelators and in 2 factors of the QOL: physical and social functioning. The data show room for improvement in the nutritional status of the patients. The NEMO trial may be able to demonstrate a better nutritional management of HD patients.

• 대한간호학회지
• /
• 제35권4호
• /
• pp.745-753
• /
• 2005

Purpose. The purposes of this study were to 1) describe the type and frequency of aggressive behavior of cognitively impaired nursing home resident, 2) develop a caregiver training program on prevention and management of aggressive behavior, 3) examine the effects of caregiver training program on the incidence of aggressive behavior of cognitively impaired nursing home resident, and 4) examine the effects of caregiver training program on nursing staff's aggressive behavior management skills. Methods. One-group, time series, quasi-experimental design with a pre-test and two post- tests was used. Data were collected from cognitively impaired home residents (N = 32) and nursing staff (N = 36) in a proprietary nursing home using Ryden Aggression Scale I, II, and Aggressive Behavior Management Scale. Data were entered and analyzed by descriptive statistics and repeated measures ANOVA. Results. Incidence of aggressive behavior was high with a mean score of 3.09 (SD = 3.11) at baseline. Caregiver training program was developed based on Progressively Lowered Stress Threshold (PLST) model and gerontological and psychiatric literature. The mean scores of aggressive behavior at baseline, Post I, and II did not differ significantly although the difference approached to the significant level (F = 2.925, p = .066). Nursing staff's aggressive behavior management skills increased at Post I, and at Post II when compared to baseline, and the difference was significant (F=12.736, p=<.00l). Conclusion. Caregiver training program showed potential impact on reduction of aggressive behavior in elders with cognitive impairment and was effective in increasing nursing staff's aggressive behavior management skills.

• Diabetes and Metabolism Journal
• /
• 제42권6호
• /
• pp.488-495
• /
• 2018

Background: Skeletal muscle plays a major role in glucose metabolism. We investigated the association between thigh muscle mass, insulin resistance, and incident type 2 diabetes mellitus (T2DM) risk. In addition, we examined the role of body mass index (BMI) as a potential effect modifier in this association. Methods: This prospective study included 399 Japanese Americans without diabetes (mean age 51.6 years) who at baseline had an estimation of thigh muscle mass by computed tomography and at baseline and after 10 years of follow-up a 75-g oral glucose tolerance test and determination of homeostasis model assessment of insulin resistance (HOMA-IR). We fit regression models to examine the association between thigh muscle area and incidence of T2DM and change in HOMA-IR, both measured over 10 years. Results: Thigh muscle area was inversely associated with future HOMA-IR after adjustment for age, sex, BMI, HOMA-IR, fasting plasma glucose, total abdominal fat area, and thigh subcutaneous fat area at baseline (P=0.033). The 10-year cumulative incidence of T2DM was 22.1%. A statistically significant interaction between thigh muscle area and BMI was observed, i.e., greater thigh muscle area was associated with lower risk of incident T2DM for subjects at lower levels of BMI, but this association diminished at higher BMI levels. Conclusion: Thigh muscle mass area was inversely associated with future insulin resistance. Greater thigh muscle area predicts a lower risk of incident T2DM for leaner Japanese Americans.

• 대한지역사회영양학회지
• /
• 제28권2호
• /
• pp.141-159
• /
• 2023

Objectives: This study aimed to determine whether a mobile health (mhealth) intervention is effective in reducing weight and changing dietary behavior among employees with overweight and obesity. The study also investigated whether engagement with the intervention affected its effectiveness. Methods: The intervention involved the use of a dietary coaching app, a wearable device for monitoring physical activity and body composition, and a messenger app for communicating with participants and an intervention manager. A total of 235 employees were recruited for a 12-week intervention from eight workplaces in Korea. Questionnaire surveys, anthropometric measurements, and 24-h dietary recalls were conducted at baseline and after the intervention. Results: After the intervention, significant decreases in the mean body weight, body mass index, body fat percentage, and waist circumference were observed. Furthermore, the consumption frequencies of multigrain rice and legumes significantly increased, whereas those of pork belly, instant noodles, processed meat, carbonated beverages, and fast food significantly decreased compared with those at baseline. The mean dietary intake of energy and most nutrients also decreased after the intervention. When the participants were categorized into three groups according to their engagement level, significant differences in anthropometric data, dietary behaviors, and energy intake were observed following the intervention, although there were no differences at baseline, indicating that higher engagement level led to greater improvements in weight loss and dietary behavior. Conclusions: The intervention had positive effects on weight loss and dietary behavior changes, particularly among employees with higher engagement levels. These results indicate the importance of increasing the level of engagement in the intervention to enhance its effectiveness. The mhealth intervention is a promising model for health promotion for busy workers with limited time.

• Structural Monitoring and Maintenance
• /
• 제3권4호
• /
• pp.349-375
• /
• 2016

The tension of an arch bridge hanger is estimated using a number of experimentally identified modal frequencies. The hanger is connected through metallic plates to the bridge deck and arch. Two different categories of model classes are considered to simulate the vibrations of the hanger: an analytical model based on the Euler-Bernoulli beam theory, and a high-fidelity finite element (FE) model. A Bayesian parameter estimation and model selection method is used to discriminate between models, select the best model, and estimate the hanger tension and its uncertainty. It is demonstrated that the end plate connections and boundary conditions of the hanger due to the flexibility of the deck/arch significantly affect the estimate of the axial load and its uncertainty. A fixed-end high fidelity FE model of the hanger underestimates the hanger tension by more than 20 compared to a baseline FE model with flexible supports. Simplified beam models can give fairly accurate results, close to the ones obtained from the high fidelity FE model with flexible support conditions, provided that the concept of equivalent length is introduced and/or end rotational springs are included to simulate the flexibility of the hanger ends. The effect of the number of experimentally identified modal frequencies on the estimates of the hanger tension and its uncertainty is investigated.