• Journal of Internet Computing and Services
• /
• v.19 no.4
• /
• pp.35-44
• /
• 2018

Monitoring through Synthesis Aperture Radar (SAR) is responsible for marine safety from floating icebergs. However, there are limits to distinguishing between icebergs and ships in SAR images. Convolutional Neural Network (CNN) is used to distinguish the iceberg from the ship. The goal of this paper is to increase the accuracy of identifying icebergs from SAR images. The metrics for performance evaluation uses the log loss. The two-layer CNN model proposed in research of C.Bentes et al.[1] is used as a benchmark model and compared with the four-layer CNN model using data augmentation. Finally, the performance of the final CNN model using the VGG-16 pre-trained model is compared with the previous model. This paper shows how to improve the benchmark model and propose the final CNN model.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10a
• /
• pp.141-143
• /
• 1999

빙산 질의란 대용량의 데이터들에 대해 집단 함수를 수행하여 임계값 이상인 데이터를 결과로 반환하는 연산을 의미한다. 빙산 질의는 도메인의 크기가 대단히 큰 대용량의 데이터에 대해 적용되므로 집단 함수의 수행을 위한 카운터를 전부 메모리에 적재할 수 없는 상황이 발생한다. 지난 연구에서는 이러한 빙산 질의를 표본 추출과 해싱을 사용하여 수행하였다. 하지만 많은 수행시간을 필요로 하는 질의를 대용량의 데이터 전체에 대해 매번 수행하여야 하며, 결과를 얻기 위한 후보 수가 커질 수 있다는 문제점이 있다. 이 논문에서는 빙산 질의에 대한 저장뷰를 통해, 사용자의 질의 중 많은 부분을 차지하는 큰 임계값으로 구성된 질의에 대해서는 즉각적인 결과를 돌려주고, 빙산 질의 결과 후보의 수를 감소시키며, 별도의 표준 추출 과정을 생략할 수 있는 방법을 제시한다. 또한 빙산 질의 히스토그램을 통해 빙산 순위 질의를 수행하는 방법을 제시하여 사용자로 하여금 보다 직관적인 질의를 작성할 수 있도록 하였다.

• Tuberculosis and Respiratory Diseases
• /
• v.87 no.2
• /
• pp.200-201
• /
• 2024

• Tuberculosis and Respiratory Diseases
• /
• v.87 no.3
• /
• pp.409-410
• /
• 2024

• Journal of Bio-Environment Control
• /
• v.19 no.4
• /
• pp.343-350
• /
• 2010

This study was conducted to investigate the effect of harvest seasons on quality and microbial population at different steps of production chain of fresh-cut iceberg lettuce. Iceberg lettuces harvested in May, June, July, October, and December were processed following industrial practices, and stored at $5^{\circ}C$ for 9 days. For microbial measurement, samples were taken from each of the following steps: harvest, transport, pretreatment, cutting, 1st-washing, 2nd-washing, and day 3, 6, and 9 of storage. Iceberg lettuce cultivated in protect house and harvested in May and October showed higher $CO_2$ levels in the packages and electrolyte leakages than lettuce harvested in June, July and December. Microbial population of raw materials harvested in July was highest (6.76 log), and microbial growth rate during storage was highest in samples harvested in May. Lettuce harvested in June had better quality and microbial safety compared to other lettuces. Although lettuce harvested in October and December had less microbial population in either raw materials or processed products, those samples had inferior quality due to off-odor development and severe browning. Therefore, it is required to maintain quality and ensure microbial safety to distribute fresh-cut lettuce with high quality and safety throughout the year.

• The Korean Journal of Pain
• /
• v.28 no.3
• /
• pp.169-176
• /
• 2015

This review has focused on important but less visible aspects of fibromyalgia (FM) with respect to the high impact of this disorder on patients and societies. FM is a common but challengeable illness. It is characterized by chronic widespread pain, which can be accompanied by other symptoms including fatigue, sleep disturbances, cognitive dysfunction, anxiety and depressive episodes. While our understanding of this debilitating disorder is limited, diagnosis and treatment of this condition is very difficult, even in the hands of experts. Due to the nature of disease, where patients experience invalidation by medical services, their families and societies regarding the recognition and management of disease, direct, indirect and immeasurable costs are considerable. These clinical and economic costs are comparable with other common diseases, such as diabetes, hypertension and osteoarthritis, but the latter usually receives much more attention from healthcare and non-healthcare resources. Present alarming data shows the grave and "iceberg-like" burden of FM despite the benign appearance of this disorder and highlights the urgent need both for greater awareness of the disease among medical services and societies, as well as for more research focused on easily used diagnostic methods and target specific treatment.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.05a
• /
• pp.34-37
• /
• 2007

간 및 낮은 메모리 사용량을 요구한다. 이러한 데이터 스트림에서의 데이터 마이닝은 전체 데이터에 대한 분석 보다는 사용자가 관심을 갖는 영역에 대한 마이닝에 초점이 맞추어져 있어, 사용자 관심영역에 대한 분석 데이터 탐색을 필요로 한다. 이에 본 논문에서는 기존의 분석 데이터 탐색 기법인 빙산 질의 및 상위-k 질의에 대하여 알아보고, 이를 보완하기 위한 확률에 기반한 데이터 탐색법인 확률기반 빙산 질의를 제안한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10a
• /
• pp.126-128
• /
• 1999

평균 빙산 질의란 대용량의 데이터들에 의해 avg 집단 함수를 수행한 뒤 임계값 이상인 데이터들을 결과로 출력하는 연산을 의미한다. 이 때 데이터 도메인의 크기가 메모리에 생성할 수 있는 카운터의 수보다 크기 때문에 연산 처리가 어렵다. 지난 연구에서 빙산 질의에 대해 제안한 해시 카운터는 avg 연사의 경우 착오누락이 발생한다는 문제점이 존재한다. 그래서 이런 문제점들을 해결하며 효율적으로 연산을 수행하기 위해, 데이터베이스를 분할하며 카운터를 관리하는 '메모리 Full 분할', '후보 Full 분할'의 두 알고리즘을 제안한다. 실험결과 두 알고리즘은 메모리크기, 데이터 분포, 데이터 순서에 영향을 받았는데, 데이터들이 정렬이 되어 있거나 데이터분포가 정규분포를 이룰 때 우수한 성능을 보였다.

• Journal of the Korean Society for information Management
• /
• v.24 no.2
• /
• pp.45-69
• /
• 2007

This paper is divided into two parts. The first part elaborates on four Information Retrieval (IR) models: a traditional IR model and three more recent, user-oriented models of It interaction presented by Belkin, Ingwersen, and Saracevic. The strengths and limitations of each model are discussed. The second part, based on an analysis of the previous models, presents the author's interactive model, namely, the Iceberg Model. The rationales that are given to explain the design of this model are associated with the following: a greater specificity of system attributes; more concrete interplays among different components of IR interaction; and, the increased role of the Human Information Intermediary (HII). In sum, the new model presents a framework that can evolve in varying information-seeking contexts.

• Environmental Analysis Health and Toxicology
• /
• v.31
• /
• pp.4.1-4.2
• /
• 2016

Once released into the air, humidifier disinfectants became tiny nano-size particles, and resulted in chemical bronchoalveolitis. Families had lost their most beloved members, and even some of them became broken. Based on an estimate of two million potential victims who had experienced adverse effects from the use of humidifier disinfectants, we can say that what we have observed was only the tip of the iceberg. Problems of entire airways, as well as other systemic effects, should be examined, as we know these nano-size particles can irritate cell membranes and migrate into systemic circulation. The story of humidifier disinfectant is not finished yet.