• 정보처리학회논문지:컴퓨터 및 통신 시스템
• /
• 제12권4호
• /
• pp.135-142
• /
• 2023

최근에는 디지털 변환이 확대됨에 따라 많은 기업들이 퍼블릭 클라우드 서비스를 이용하거나 자체 데이터센터를 구축하고 있다. 소프트웨어 정의 스토리지는 클라우드 플랫폼에서 데이터를 저장하기 위한 핵심적인 솔루션으로 전세계적으로 이용이 확대되고 있다. 소프트웨어 정의 스토리지는 전체 스토리지 자원을 하나의 저장장치와 같이 가상화하여 사용할 수 있고 유연한 Scale-out을 지원하는 장점이 있는 반면에, 가변 크기의 오브젝트 방식으로 인한 디스크의 이용에 불균형이 발생하고, 장애를 유발할 수 있다. 본 연구에서는 디스크 이용의 불균형 문제를 해결하기 위하여 스토리지의 상태정보를 바탕으로 디스크의 가중치를 최적화하여 오브젝트를 재분배하는 방법에 대하여 제안하고, 그 실험 결과를 제시하였다. 실험을 수행한 결과, 디스크의 최대 이용률이 89%에서 79%로 10%만큼 감소한 것을 확인하였다. 디스크의 이용률을 최적화함으로써 장애를 예방하고, 더 많은 데이터를 균등하게 저장할 수 있어 효율적인 스토리지 이용이 가능할 것으로 기대된다.

• 한국산업정보학회논문지
• /
• 제29권2호
• /
• pp.15-25
• /
• 2024

효과적인 동물 생태계 분석을 위해서는 동물 서식 현황을 자동으로 파악할 수 있는 동물 관제 기술이 중요하다. 특히 울음소리로 종을 판별하는 동물 소리 분류 기술은 영상을 통한 판별이 어려운 환경에서 큰 주목을 받고 있다. 기존 연구들은 단일 딥러닝 모델을 사용하여 동물 소리를 분류하였으나, 야외 환경에서 수집된 동물 소리는 많은 배경 잡음을 포함하여 단일 모델의 판별력을 악화시키며, 종에 따른 데이터 불균형으로 인해 모델의 편향된 학습을 야기한다. 이에, 본 논문에서는 클래스의 데이터 수를 고려하여 페널티를 부여하는 Focal Loss를 사용한 여러 분류 모델의 예측결과를 앙상블을 통해 결합하여 잡음이 많은 동물 소리를 효과적으로 분류할 수 있는 기법을 제안한다. 공개 데이터 셋을 사용한 실험에서, 제안된 기법은 단일 모델의 평균 성능에 비해 Recall 기준으로 최대 22.6%의 성능 개선을 달성하였다.

• 대한임베디드공학회논문지
• /
• 제19권1호
• /
• pp.47-55
• /
• 2024

We spend a lot of time in indoor space, and the space has a huge impact on our lives. Interior design plays a significant role to make an indoor space attractive and functional. However, it should consider a lot of complex elements such as color, pattern, and material etc. With the increasing demand for interior design, there is a growing need for technologies that analyze these design elements accurately and efficiently. To address this need, this study suggests a deep learning-based design analysis system. The proposed system consists of a semantic segmentation model that classifies spatial components and an image classification model that classifies attributes such as color, pattern, and material from the segmented components. Semantic segmentation model was trained using a dataset of 30000 personal indoor interior images collected for research, and during inference, the model separate the input image pixel into 34 categories. And experiments were conducted with various backbones in order to obtain the optimal performance of the deep learning model for the collected interior dataset. Finally, the model achieved good performance of 89.05% and 0.5768 in terms of accuracy and mean intersection over union (mIoU). In classification part convolutional neural network (CNN) model which has recorded high performance in other image recognition tasks was used. To improve the performance of the classification model we suggests an approach that how to handle data that has data imbalance and vulnerable to light intensity. Using our methods, we achieve satisfactory results in classifying interior design component attributes. In this paper, we propose indoor space design analysis system that automatically analyzes and classifies the attributes of indoor images using a deep learning-based model. This analysis system, used as a core module in the A.I interior recommendation service, can help users pursuing self-interior design to complete their designs more easily and efficiently.

• 혜화의학회지
• /
• 제18권2호
• /
• pp.13-20
• /
• 2009

Objectives : In order to investigate change of isokinetic trunk muscle strength according to decrease of body composition analysis parameter after obesity treatment. Methods : 2 obese patients have been treated with oriental medical obese treatment for 1 month. One patient got the exercise treatment, another didn't. Before and after treatment, the segmental bioelectrical impedance analysis, isokinetic trunk muscle strength test were performed. Then we analyzed the relationship of data. Results : After obesity treatment, BMI(Body Mass Index), PBF(Percentage of Body Fat), WHR(Waist Hip Ratio) were decreased in all patient and LBM(Lean Body Mass) was increased. In non-exercise patient, Ext.PT(extension Peak Torque) was decreased and Flex.PT(flexion Peak Torque) was increased. In exercise patient showed the opposite results. E/F ratio became more imbalance. Conclusions: Ext.PT was decreased in non-exercise patient but increased in exercise patient. And the trunk muscle strength became imbalance in both patients, right after the treatment. So trunk muscle exercise should be carried out and it is necessary to do long term study.

• Journal of Nutrition and Health
• /
• 제30권3호
• /
• pp.326-335
• /
• 1997

In the Health Promotion Law proclaimed on January 1995, nutritional improvement at national level was emphasized and designated as one of the jobs to be carried out by local governmnets. With such a situational necessity, we conducted a dietary survey along with an anthropometric measurement, biochemical assessment and questionnaire analysis on general characteristics of the students from 3 junior higher schools in Seoul area. About 300 students had participated in the study and the data from only 139 students, 28 boys and 111 girls, with complete report of dietary intake were subjected to analysis, comparison and discussion. Fasting blood samples were drawn and analyzed for hemoglobin, hematocrit and total cholesterol. Dietary intake was monitored by 1-day 24hr recall +2-day food record. In general, the average intake of nutrients for most of the subjects were above RDA for korean of that age except for 2 nutrients namely, vitamin A and calcium, of which average intake corresponded to 46-69% of RDA. In addition to this kind of nutritional imbalance, there were several other factors of nutritional problems such as skipping breakfast, overeating at dinner and frequent eating of snacks. As the best countermeasure for these kinds of nutritional problems, more detailed campaign and prractical nutrition education for these adolescents are necessary. Only through proper education and guidance for them, the healthy and intellectual man power could be guaranteed for the forthcoming 21st century.

• 전력전자학회논문지
• /
• 제20권1호
• /
• pp.19-30
• /
• 2015

A static synchronous compensator (STATCOM) applied to rapidly changing, highly unbalanced loads such as electric arc furnaces (EAFs), requires both positive-sequence and negative-sequence current control, which indicates fast response characteristics and can be controlled independently. Furthermore, a delta-connected STATCOM with cascaded H-bridge configuration accompanying multiple separate DC-sides, should have high performance zero-sequence current control to suppress a phase-to-phase imbalance in DC-side voltages when compensating for unbalanced load. In this paper, actual EAF data is analyzed to reflect on the design of current controllers and a pioneering zero-sequence current controller with a superb transient performance is devised, which generates an imaginary -axis component from the presumed response of forwarded reference. Via simulation and experiments, the performance of the positive, negative, and zero-sequence current control of a cascaded H-bridge STATCOM for EAF is verified.

• 기술혁신연구
• /
• 제16권2호
• /
• pp.201-223
• /
• 2008

This paper studies the supply and demand of Industrial Technology Labor-force(ITL) and analyzes the determinate of ITL shortage in Korean manufacturing. We classified the industry into four categories-high technology industries, medium-high technology industries, medium-low technology industries and low technology industries-based on its R&D intensity like OECD. For the empirical analyses we use a survey data collected from 5,703 enterprises. The key findings are as follows: Firstly, a large majority of ITL is engaged in more technology-intensive industries but the categories that are exposed to more serious labor-force shortage problem are medium-high technology industries and low technology industries. Secondly, in the terms of supply factor, the ITL shortage problems are mainly due to the avoidance of ITL jobs. And the demand point, the reason is that the most of ITL are not researchers but production managers. Thirdly, the cause of imbalance between supply and demand of ITL are different by the technological categories. For example, in the high technology industries, the supply factors, such as average wage and turnover rate played more important role in the imbalance. But in the low technology industries the demand factors, such as per capita sales and the ratio of ITL in all employees were relatively much more important. Based on the findings, we discovered some political meanings such as the necessity to plan various policies to resolve the shortage problem of ITL according to the technological categories, etc.

• 가정∙방문간호학회지
• /
• 제27권2호
• /
• pp.137-145
• /
• 2020

Purpose: This study aimed to identify the training system and current status of home health nurses and to examine ways to retain sufficient number of advanced practice nurses (APNs) in the home health nursing field. Methods: This study analyzed the adequacy of the supply of home health nurse by reviewing the existing research literature and statistical data. Results: Discussions on how to revitalize the home care business have been ongoing since the beginning of 2001. However, despite home health nurses being oversupplied, discussions on the adequacy of supply have always been excluded from the focus of revitalization. The recent expansion of the home care business has resulted in a shortage of workforce, which can be inked not only to the continuous reduction of the designated quota of programs but also to the regional imbalance of educational institutions. The serious imbalance between supply and demand has caused fears that the home care business would drastically reduce. Conclusion: It is necessary to not only increase designated quotas for APNs programs but also integrate those programs of the similar curricula, thus lowering supply shortages in home health nurses.

• Journal of Information Technology Applications and Management
• /
• 제28권3호
• /
• pp.109-125
• /
• 2021

For the government of president Moon's AI SW HRD policy, he proclaimed AI democracy that anyone can utilize artificial intelligence technology to spread AI education for the people of the country. Through cognitive map analysis, this study presents expected policy outcomes due to the input of policy factors to overcome crisis factors and utilize opportunity factors. According to the cognitive guidance analysis, first, the opportunity factor is recognized as accelerating the digital transformation to Covid 19 if AI SW HRD is well nurtured. Second, the crisis factor refers to the rapid paradigm shift caused by the intelligence information society, resulting in job losses in the manufacturing sector and deepening imbalance in manpower supply and demand, especially in the artificial intelligence sector. Third, the comprehensive cognitive map shows a circular process for creating an AI SW ecosystem in response to threats caused by untact caused by Corona and a circular process for securing AI talent in response to threats caused by deepening imbalance in manpower supply and demand in the AI sector. Fourth, in order to accelerate the digital circulation that has been accelerated by Corona, we found a circular process to succeed in the Korean version of digital new deal by strengthening national and corporate competitiveness through AI-utilized capacity and industrial and regional AI education. Finally, the AI utilization empowerment strengthening rotation process is the most dominant of the four mechanisms, and we also found a relatively controllable feedback loop to obtain policy outputs.

• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.