• Journal of Preventive Medicine and Public Health
• /
• 제33권3호
• /
• pp.343-348
• /
• 2000

Objectives : This study was conducted to integrate the results of studies assessing the association between chronic noise exposure and blood pressure. Methods : Using a MEDLINE search with noise exposure, blood pressure and hypertension as key words, we retrieved articles from the literature that were published from 1980 to December 1999. The criteria for quality evaluation were as follows: 1) the study subjects must have been workers employed at a high noise level area 2) The paper should use average and cumulative noise exposure as method for exposure evaluation. 3) Blood pressure in each article should be reported in a continuous scale Among the 77 retrieved articles, six studies were selected for quantitative meta-analysis. Before the integration of the regression coefficients for the association between blood pressure and noise level, homogeneity tests were conducted. Results : All studies were a cross-sectional design and the study subjects were industrial workers. Five papers used a time-weighted average for noise exposure and only one paper calculated the cumulative noise exposure level. The measurement of blood pressure in the majority of studios were accomplished in a resting stale, and used an average of two or more readings. The homogeneity of studies was rejected in a fixed effect model, so we used the results in a random effect model. The results of the quantitative meta-analysis, the weighted regression coefficient of noise associated with systolic blood pressure and diastolic blood pressure were 0.05 (95% confidence interval [CI]: -0.03, 0.13) and 0.06 (95% CI: -0.01, 0.13), respectively. Conclusions : Our results suggested that chronic exposure to industrial noise does not cause elevated blood pressure.

• 한국산림과학회지
• /
• 제101권2호
• /
• pp.213-219
• /
• 2012

제5차 국가산림자원조사는 다양한 산림자원의 현황을 평가하고 시간경과에 따른 산림자원의 변화를 모니터링하기 위하여 연년조사체계로 개편되었다. 본 연구는 충청북도를 대상으로 연년조사체계에서 수집된 현지조사 표본점 자료를 이용하여 일정시점의 평균임목축적을 추정하기 위한 방법을 모색하기 위해 수행되었다. 연년통계량의 산출을 위하여 임상구분의 표본층을 고려하지 않은 단순임의추출법과 표본층을 고려한 사후층화이중추출법의 추정식을 이용한 추정치를 비교한 결과, 사후층화이중추출법에 의한 추정치의 정확도가 높은 것으로 나타났다. 최근 5년간 수집된 현지조사 자료를 통합하기 위하여 조사년도의 차이를 고려하지 않은 시차 무시법(Temporally Indifferent Method), 조사년도별 추정치를 산출한 후 통합하는 단순이동평균법, 그리고 연도별 표본개수에 의해 가중치를 부여하는 가중이동평균법에 의한 평균임목축적과 추정분산을 비교하였다. 평균임목축적은 시차 무시법과 가중이동평균법에서 동일한 것으로 나타났지만, 추정치의 정도를 나타내는 추정분산은 가중이동평균법을 이용한 것이 약간 향상되었으며, 결과적으로 연도별 변이를 반영할 수 있는 가중이동평균법이 보다 적합한 것으로 평가되었다.

• 한국컴퓨터정보학회논문지
• /
• 제28권1호
• /
• pp.17-26
• /
• 2023

본 연구논문에서는 LSTM 기반의 학습 모델 적용과 그 효용성을 높일 수 있도록 몇 가지 평활 기법을 비교, 적용하고자 한다. 적용된 평활 기법은 Savitky-Golay, 지수 평활법, 가중치 이동 평균 등이다. 본 연구를 통해 비트코인 데이터에 LSTM모델 적용 시 보여준 결과 값보다 전처리 과정에서 적용된 Savitky-Golay 필터가 적용된 LSTM 알고리즘이 예측 성능에 유의미한 좋은 결과를 보였다. 예측 성능 결과를 확인하기 위해 비트코인 가격 예측에 따른 복잡 요인을 제거하는데 사용된 LSTM의 경우와 Savitzky-Golay LSTM 모델에 따른 학습 손실율과 검증 손실율을 비교하고 그 신뢰성을 높일 수 있도록 20회 평균값으로 실험하였다. 그 결과 (3.0556, 0.00005), (1.4659, 0.00002)의 값을 얻을 수 있었다. 결과적으로는 비트코인과 같은 암호화폐가 주식보다 더한 변동성을 가지는 만큼 데이터 전처리 과정에서 평활 기법(Savitzky-Golay)을 적용하여 잡음(Noise)을 제거하였으며, 전처리 후의 데이터는 LSTM 신경망 학습을 통해서 비트코인 예측률을 높이는데 가장 유의미한 결과를 얻을 수 있었다.

• 대한토목학회논문집
• /
• 제34권6호
• /
• pp.1873-1879
• /
• 2014

실시간 자료를 반영한 통행시간 예측 기법은 다양하지만 관련 연구 검토 결과 과거이력데이터가 충분하다면 타 모형에 비해 K 최대근접이웃(K-Nearest Neighbors)의 정확도가 우수하므로 본 연구에서는 이에 대한 적용 방법 도출 및 가능성 평가를 목적으로 한다. 본 연구에서는 KNN의 입력 자료로 TCS 교통량 및 DSRC 구간통행시간의 실시간 및 과거 이력자료, 경로통행시간 이력자료를 활용하였다. 통행시간 예측치는 TCS 교통량 및 DSRC 구간통행시간의 실시간 자료와 유사한 경로통행시간을 탐색한 후 이를 가중평균하여 산출하였다. 예측 기법을 적용한 결과 DSRC 구간통행시간의 가중치가 증가할수록 정확도는 증가하였으며, 이는 실시간 교통상황 변화를 DSRC 구간통행시간이 잘 반영하기 때문이다. 그러나 TCS 교통량을 기반으로 한 경우 역시 정확도의 차이가 크지 않으며, 변화 추이도 유사하게 나타났다. 이러한 결과를 볼 때 향후 대용량의 과거이력자료가 축적될 경우 예측오차는 더욱 감소될 것으로 기대된다.

• 대한예방의학회:학술대회논문집
• /
• 대한예방의학회 1994년도 교수 연수회(환경)
• /
• pp.681-691
• /
• 1994

The American Conference of Governmental Industrial Hygienists (ACGIH) has proposed a threshold limit value (TLV) for benzene of 0.1 ppm. Individuals representing the American Petroleum Institute (API)and the Chemical Manufacturers Association (CMA) have argued that 1) the risk assessment by Rinsky .et al. which ACGIH partially relied upon for its proposed TLV overestimates the risk; however, at the exposures levels of interest - (e.g., 0.1 to 1.0 ppm) for establishing a benzene TLV, the Rinsky et al. assessment provides lower estimates of leukemia risk than most others; 2) ACGIH should not use the Dow study for direct observational evidence of leukemia risk associated with low-level benzene exposure because of confounding exposure; however, it is unlikely that confounding exposures played a role in the excess of leukemia demonstrated in the study, and the Dow cohort was exposed to an average benzene concentration of about 5.5 ppm benzene for 7.11 years (31:1.5 ppm-years), while some of the individuals in the study who died from leukemia were exposed to an average of only 1.0 ppm without the opportunity for highpeak exposures; 3) the Occupational Safety and Health Administration (OSHA) established an 8-hour time-weighted average (TWA) of 1.0 ppm in 1987, and there is no new evidence that would justify reducing the TWA below that level; however, the OSHA TWA of 1.0 ppm was based on economic feasibility and the level of excess risk remaining at 1.0 ppm, i.e., 10 excess leukemia deaths per 1000 workers over an occupational lifetime (45 years) according to OSHA's preferred estimate leaves behind I risk considered significant by OSHA. In addition, chromosomal studies among workers and in animals exposed to benzene indicate that low-level exposure, i.e., 1.0 ppm, is associated with elevated Cytogenetic damage. On the basis of adverse health effects data alone, in this author's opinion, it would be poor science and poor public health policy to establish a benzene TLV greater than 0.1 ppm.

• 한국지역사회생활과학회지
• /
• 제14권3호
• /
• pp.99-109
• /
• 2003

This study was conducted to collect basic marketing and management data from businesses run by farmers producing traditional Kanjang and Doenjang, fermented soybean products. The actual conditions of the fermentation processing sites at farmhouses participating in the production of soybean fermentation products were investigated. The subjects of this survey were 130 small size farmhouse Kanjang and Doenjang processing sites nationwide. The frequency, percentage, t-value, chi-square, and ANOVA were used for statistical analysis. The farmhouse business surveyed were generally operated by rural women for non-farming business income. The percentage of co-worked sites was 71.2 while the percentage of sites operating with permits was 39.2. Generally, the the facilities, size, number of working people, and output were very small. The areas in which the products were sold, site-located regions and region metropolises, were equally weighted. Sales volumes in region metropolises for sites with permits were a little higher than sites without permits. Without regard to operation type, the percentage of sales was highest in cases of direct sale by customer order. Co-worked sites have been found to have more experience in publicity than individually operated sites. As for methods of publicity, co-worked sites use mass media such as newspapers and broadcasting. Individually operated sites usually use social organizations and acquaintances. It was found that the average sales of each site totaled 25 million Won. The average income of each site was 12 million Won, and average income per participant was 3 million Won. Total sales income for sites with permits was significantly higher than sites without permits. But personal income was much higher at individually operated sites without regard to whether the site had a permit or not. This kind of business was found to contribute to an individual's time management skills as well as instill a sense of pride.

• 한국정보통신학회논문지
• /
• 제23권5호
• /
• pp.533-539
• /
• 2019

본 연구는 본 연구는 Microsoft Malware Classification Challenge 데이터 셋을 사용해 임의의 길이 입력 데이터에 대응할 수 있는 멀웨어 분류 모델을 제안한다. 우리는 기존 연구의 멜웨어 데이터를 이미지화 시키는 것을 기반으로 한다. 제안 모델은 멀웨어 데이터가 큰 경우는 많은 이미지를 생성하고, 작은 데이터는 적은 이미지를 생성한다. 생성된 이미지를 시계열 데이터로 Dynamic RNN으로 학습시킨다. RNN의 출력 값은 Attention 기법을 응용해 가장 가중치가 높은 출력만 사용하고, RNN 출력값을 다시 Residual CNN으로 학습시켜 최종적으로 멀웨어를 분류한다. 제안모델을 실험한 결과 검증 데이터 셋에서 Micro-average F1 score 92%를 기록하였다. 실험 결과 특별한 특징 추출 및 차원 축소 없이 임의 길이의 데이터를 학습 및 분류할 수 있는 모델의 성능을 검증할 수 있었다.

• 한국산업보건학회지
• /
• 제30권4호
• /
• pp.342-352
• /
• 2020

Objectives: The characteristics of research workers are different from those working in the manufacturing industry. Furthermore, the reagents used change according to the research due to the characteristics of the laboratory, and the amounts used vary. In addition, since the working time changes almost every day, it is difficult to adjust the time according to exposure standards. There are also difficulties in setting standards as in the manufacturing industry since laboratory environments and the types of experiments performed are all different. For these reasons, the measurement of the working environment of research workers is not realistically carried out within the legal framework, there is a concern that the accuracy of measurement results may be degraded, and there are difficulties in securing data. The exposure evaluation based on an eight-hour time-weighted average used for measuring the working environment to be studied in this study may not be appropriate, but it was judged and consequently applied as the most suitable method among the recognized test methods. Methods: The investigation of the use of chemical substances in the research laboratory, which is the subject of this study, was conducted in the order of carrying out work environment measurement, sample analysis, and result analysis. In the case of the use of chemical substances, after organizing the substances to be measured in the working environment, the research workers were asked to write down the status, frequency, and period of use. Work environment measurement and sample analysis were conducted by a recognized test method, and the results were compared with the exposure standards (TWA: time weighted average value) for chemical substances and physical factors. Results: For the substances subject to work environment measurement, the department of chemical engineering was the most exposed, followed by the department of chemistry. This can lead to exposure to a variety of chemicals in departmental laboratories that primarily deal with chemicals, including acetone, hydrogen peroxide, nitric acid, sodium hydroxide, and normal hexane. Hydrogen chloride was measured higher than the average level of domestic work environment measurements. This can suggest that researchers in research activities should also be managed within the work environment measurement system. As a result of a comparison between the professional science and technology service industry and the education service industry, which are the most similar business types to university research laboratories among the domestic work environment measurements provided by the Korea Safety and Health Agency, acetone, dichloromethane, hydrogen peroxide, sodium hydroxide, nitric acid, normal hexane, and hydrogen chloride are items that appear higher than the average level. This can also be expressed as a basis for supporting management within the work environment measurement system. Conclusions: In the case of research activity workers' work environment measurement and management, specific details can be presented as follows. When changing projects and research, work environment measurement is carried out, and work environment measurement targets and methods are determined by the measurement and analysis method determined by the Ministry of Employment and Labor. The measurement results and exposure standards apply exposure standards for chemical substances and physical factors by the Ministry of Employment and Labor. Implementation costs include safety management expenses and submission of improvement plans when exposure standards are exceeded. The results of this study were presented only for the measurement of the working environment among the minimum health management measures for research workers, but it is necessary to prepare a system to improve the level of safety and health.

• 한국콘텐츠학회논문지
• /
• 제11권3호
• /
• pp.271-280
• /
• 2011

심박 변이도 (HRV) 분석은 심근허혈 (MI)를 평가하기 위한 편리한 도구이다. HRV에 대한 분석법은 시간 영역과 주파수 영역 분석으로 나눠질 수 있다. 본 논문은 단기간의 HRV 분석에 있어서 웨이블릿 변환을 주파수 영역 분석과 시간 영역 분석 비교하기 위하여 사용하였다. ST-T와 정상 에피소드는 각각 European ST-T 데이터베이스와 MIT-BIH Normal Sinus Rhythm 데이터베이스에서 각각 수집되었다. 한 에피소드는 32개 연속하는 RR 간격으로 나눠질 수 있다. 18개 HRV 특징은 시간과 주파수 영역 분석을 통하여 추출된다. 가종 퍼지소속함수 신경망 (NEWFM)은 추출된 18개의 특징을 이용하여 심근허혈을 진단하였다. 결과는 보여주는 평균 정확도로부터 시간영역과 주파수영역의 특징은 각각 75.29%와 80.93%이다.

• 대한교통학회지
• /
• 제28권6호
• /
• pp.7-15
• /
• 2010

본 연구에서는 과거의 교통패턴과 실시간 교통데이터와의 차이값을 이용하여 돌발상황을 판정하는 알고리즘을 개발하고자 한다. 이를 통해 운영자의 측면에서 이해하기 쉽고 운영 및 수정 보완이 용이한 돌발상황 감지알고리즘을 개발하는 것이 목적이다. 본 연구에서 제안한 알고리즘은 교통패턴 구축을 위하여 30초 주기 원시데이터를 바탕으로 동일한 지점의 동일한 요일 및 시간대의 교통량과 속도를 이용한 가중이동평균법을 사용하였다. 모형은 오류자료 보정처리, 소통상황 판정, 패턴자료와의 비교, 돌발상황 판정, 지속성 검사의 단계로 이루어졌으며, 적정 파라메타 선정을 위하여 다양한 파라메타값을 적용하였다. 알고리즘의 적용 결과 검지율은 평균 94.7%, 오보율은 0.8%, 평균 검지시간은 1.6분으로 기존 모형과의 비교분석 결과에서도 우수한 편에 속하는 것을 확인할 수 있다. 교통패턴이라는 개념을 사용하여 복잡하지 않은 과정을 통해 우수한 결과를 얻었으며, 운영자의 측면에서 실제 운영자들이 돌발상황을 판단하는 과정을 알고리즘으로 완성하였다는 측면에서 본 연구의 의의가 있다.