• 한국통신학회논문지
• /
• 제32권10C호
• /
• pp.965-974
• /
• 2007

본 논문에서는 내용 기반 음악 범주 분류 시스템에서 다중 범주를 위한 특징벡터 선택 알고리즘을 제안한다. 제안된 특징벡터 선택 알고리즘은 분리 성능을 측정할 때 가우시안 혼합 모델(Gaussian Mixture Model: GMM)을 기반으로 GMM separation score을 측정함으로써 확률분포 및 분리 성능 추정의 정확도를 높였고, sequential forward selection 방법을 개선하여 이전까지 선택된 특징벡터들이 분리를 잘 하지 못하는 범주들을 기준으로 다음 특징벡터를 선택하는 알고리즘을 제안하여 다중 범주 분류의 성능을 높였다. 제안된 알고리즘의 성능 검증을 위해 음색, 리듬, 피치 등 오디오 신호의 특징을 나타내는 다양한 파라미터를 오디오 신호로부터 추출하여 제안된 특징벡터 선택 알고리즘과 기존의 알고리즘으로 특징벡터를 선택한 후 GMM classifier와 k-NN classifier를 이용하여 분류 성능을 평가하였다. 제안된 특징벡터 선택 알고리즘은 기존 알고리즘에 비하여 3%에서 8% 정도의 분류 성능이 향상된 것을 확인할 수 있었고 특히 낮은 차원의 특징벡터의 분류 실험에서는 분류 정확도 측면에서 5%에서 10% 향상된 좋은 성능을 보였다.

• The Plant Pathology Journal
• /
• 제28권2호
• /
• pp.123-136
• /
• 2012

Diseases caused by begomoviruses (family Geminiviridae, genus Begomovirus) constitute a serious constraint to tropical and sub-tropical agro-ecosystems worldwide. In recent years, they have also introduced in temperate regions of the world where they have great impact and are posing a serious threat to a variety of greenhouse crops. Begomoviral diseases can in extreme cases reduce yields to zero leading to catastrophic losses in agriculture. They are still evolving and pose a serious threat to sustainable agriculture across the world, particularly in tropics and sub-tropics. Till recently, there have been no records on the occurrence of begomoviral disease in South Korea, however, the etiology of other plant viral diseases are known since last century. The first begomovirus infected sample was collected from sweet potato plant in 2003 and since then there has been gradual increase in the begomoviral epidemics specially in tomato and sweet potato crops. So far, 48 begomovirus sequences originating from various plant species have been submitted in public sequence data base from different parts of the country. The rapid emergence of begomoviral epidemics might be with some of the factors like evolution of new variants of the viruses, appearance of efficient vectors, changing cropping systems, introduction of susceptible plant varieties, increase in global trade in agricultural products, intercontinental transportation networks, and changes in global climatic conditions. Another concern might be the emergence of a begomovirus complex and satellite DNA molecules. Thorough understanding of the pathosystems is needed for the designing of effective managements. Efforts should also be made towards the integration of the resistant genes for the development of transgenic plants specially tomato and sweet potato as they have been found to be widely infected in South Korea. There should be efficient surveillance for emergence or incursions of other begomoviruses and biotypes of whitefly. This review discusses the general characteristics of begomoviruses, transmission by their vector B. tabaci with an especial emphasis on the occurrence and distribution of begomoviruses in South Korea, and control measures that must be addressed in order to develop more sustainable management strategies.

• 한국음향학회지
• /
• 제21권7호
• /
• pp.632-640
• /
• 2002

본 논문에서는 기존의 RLJ-신뢰도 (RLJ-confidence measure)와 정규화 신뢰도 (normalized CM)의 단점을 보완하기 위해 ACM (Anti-filler CM)을 제안하였고, HCM (hybrid CM)을 이용하여 기존의 NCM과 제안한 ACM을 통합하였다. 제안된 ACM은 기존 신뢰도의 단점 중 하나인 오인증 (FA: false acceptance)의 원인이 반음소 모델의 구성방법에 있다고 보고 음소 인식기를 이용하여 실제 음소 수열을 추정한 다음, 이를 반음소 모델로 정의하고 신뢰도를 계산하였다. 두 가지 신뢰도의 특성을 살펴보면, 기존 NCM(FR: false rejection)에 좋은 성능을 보이고, 제안한 ACM은 FA에서 좋은 성능을 보여 두 신뢰도가 상보적인 특성을 가진다 이를 이용하여 두 가지 신뢰도를 가중치 벡터 α를 이용하여 통합하고 이를 합성 신뢰도 (HCM: Hybrid CM)라고 정의하였다. 실험결과 미검출율 (MDR: missed detection rate) 10%부근에서, HCM 적용시에 0.219 FA/KW/HR (false alarm/keyword/how)로서 NCM 단독사용에 비해 성능이 22% 향상되었다.

• 대한공간정보학회지
• /
• 제10권3호
• /
• pp.89-97
• /
• 2002

45(GNSS, SIIS, GIS, ITS) 기술의 핵심이 되는 공간정보의 상호 공유 극대화를 위하여 원격지 공간 데이터 공유 및 제공을 위하여 45-Van 시스템을 개발해오고 있다. 45-Van 시스템은 GPS/IMU, 레이저, CCD 영상, 무선통신기술을 통합 연계하여 현장에서 GIS용 DB 정보 등과 같은 45 핵심 DB정보 및 정확한 위치 정보를 직접 획득 생산이 가능하다. 즉, 4S-Van은 GPS와 IMU의 통합으로 카메라의 위치 및 자세를 결정하며, 두 대의 CCD카메라로 전방을 촬영하여, 공간전방 교회법(Space Intersection)으로 피사체의 위치해석을 하게 되고 기존의 벡터 DB 체계와 호환됨으로써 데이터베이스의 구축 및 현장활용이 가능하도록 할 수 있는 기술이다. 또한 적외선 카메라 및 무선 통신 기술을 활용한 다양한 응용이 가능하다. 본 논문에서는 GPS, CCD 카메라, IMU의 차량 탑재에 의한 45-Van 설계와 기능에 대하여 살펴본다.

• 전자공학회논문지SC
• /
• 제48권4호
• /
• pp.40-46
• /
• 2011

최근 저전력 경량 운영체계 및 다양한 기능을 수행하는 베터리 기반 소형 노드들의 발전 덕분에 센서네트워크(WSNs)는 많은 응용 가능성을 보고 주고 있다. 센서네트워크 환경에서는 종단 센서 노드에서 수집한 데이터를 싱크 노드까지 적절하게 전송하기위하여 계층 2 및 계층 3 사이 혼성 매체 제어 알고리즘을 개발하는 것이 중요하다. 본 논문에서는 IEEE 802.15.4 표준을 기반으로 UC Berkely에서 제공한 TinyOS 환경에서 동작하는 MAC 계층과 NWK 계층을 혼합한 Cross-layer 기반의 우선 순위 기법을 제안하고자 한다. 제안한 비콘 구간 우선순위 경로 설정(BPR) 기법은 센싱 데이터 전송 시 비콘 구간을 이용한 Wibeem protocol을 이용하였고, 데이터 전송 시 제시한 평가 항목 범위에서 기존의 비컨 전송 방식과 비교하여 상대적으로 우수한 성능을 보여 주고 있다. 평가 항목은 패킷 처리량, 전송률, 지연시간 및 에너지 소비 값이다. TOSSIM 환경에서 모의실험 결과는 제안한 BPR을 이용할 경우 기존 AODV방식과 비교하여 개선된 성능을 보고 주고 있다.

• 전자공학회논문지CI
• /
• 제45권5호
• /
• pp.15-24
• /
• 2008

얼굴인식 등과 같은 고차원 식별문제에서는 샘플패턴의 수가 패턴의 차원보다 작아지게 된다. 이러한 상황에서 차원을 축소하기위해 선형판별분석법을 적용할 경우, 희소성(Small Sample Size: SSS)문제가 발생한다. 최근, SSS 문제를 해결하기 위하여 비유사도에 기반 한 식별법(Dissimilarity-Based Classification: DBC)을 이용하는 방법이 검토되었다. DBC에서는 특징 벡터 대신에 학습 샘플들로부터 추출한 프로토타입들과의 비유사도를 측정하여 입력 패턴을 식별하는 방법이다. 본 논문에서는 비유사도 표현단계와 DBC 학습단계에서 퓨전기법을 중복 적용하는 다단계 퓨전기법(Multi-level Fusion Strategies: MFS)으로 DBCs를 최적화시키는 방법을 제안한다. 제안 방법을 벤취마크 얼굴영상 데이터베이스를 대상으로 실험한 결과, 식별률을 향상시킬 수 있음을 확인하였다.

• Parasites, Hosts and Diseases
• /
• 제56권6호
• /
• pp.559-565
• /
• 2018

The identification and characterization of pathogenic and zoonotic tick-borne diseases like granulocytic anaplasmosis are essential for developing effective control programs. The differential diagnosis of pathogenic Anaplasma phagocytophilum and non-pathogenic A. phagocytophilum-like Anaplasma spp. is important for implementing effective treatment from control programs. The objective of the present study was to investigate the prevalence of Anaplasma spp. in horses in Korea by nucleotide sequencing and restriction enzyme fragment length polymorphism assay. Of the 627 horses included in the study, only 1 (0.2%) was infected with A. phagocytophilum. Co-infection with A. phagocytophilumlike Anaplasma spp. was not detected in the study. The 16S rRNA sequence of A. phagocytophilum was similar (99.5-100%) to A. phagocytophilum 16S rRNA isolated from horses in other countries. PCR adapted to amplify A. phagocytophilum groEL and msp2 genes failed to generate amplicons, suggesting genetic diversity in these genes. This study is the first molecular detection of A. phagocytophilum in horses in Korea. Human granulocytic anaplasmosis and animal infection of A. phagocytophilum have been reported in Korea recently. Because of vector tick distribution, global warming, and the increase of the horse industry, horses should be considered as a potential reservoir for A. phagocytophilum, and cross infectivity should be evaluated even though a low prevalence of infection was detected in this study. Furthermore, continuous surveillance and effective control measures for A. phagocytophilum should be established to prevent disease distribution and possible transmission to humans.

• 한국안전학회지
• /
• 제36권3호
• /
• pp.31-39
• /
• 2021

In Korea, more than half of work-related fatalities have occurred on construction sites. To reduce such occupational accidents, safety inspection by government agencies is essential in construction sites that present a high risk of serious accidents. To address this issue, this study developed risk prediction models of serious accidents in construction sites using five machine learning methods: support vector machine, random forest, XGBoost, LightGBM, and AutoML. To this end, 15 proactive information (e.g., number of stories and period of construction) that are usually available prior to construction were considered and two over-sampling techniques (SMOTE and ADASYN) were used to address the problem of class-imbalanced data. The results showed that all machine learning methods achieved 0.876~0.941 in the F1-score with the adoption of over-sampling techniques. LightGBM with ADASYN yielded the best prediction performance in both the F1-score (0.941) and the area under the ROC curve (0.941). The prediction models revealed four major features: number of stories, period of construction, excavation depth, and height. The prediction models developed in this study can be useful both for government agencies in prioritizing construction sites for safety inspection and for construction companies in establishing pre-construction preventive measures.

• 지질공학
• /
• 제32권3호
• /
• pp.363-376
• /
• 2022

An area under construction for a living facility collapsed around 12:48 KST on 13 January 2021 in Sa-dong, Ansan-si, Gyeonggi-do. There were no casualties due to the rapid evacuation measure, but part of the temporary retaining facility collapsed, and several cracks occurred in the adjacent road on the south side. This study used the potential of synthetic aperture radar (SAR) satellite for surface property changes that lies in backscattering characteristic to map the collapsed structure. The interferometric SAR technique can make a direct measurement of the decorrelation among different acquisition dates by integrating both amplitude and phase information. The damage proxy map (DPM) technique has been employed using four high-resolution Constellation of Small Satellites for Mediterranean basin Observation (COSMO-SkyMed) data spanning from 2020 to 2021 during ascending observation to analyze the collapse of the construction. DPM relies on the difference of pre- and co-event interferometric coherences to depict anomalous changes that indicate collapsed structure in the study area. The DPMs were displayed in a color scale that indicates an increasingly more significant ground surface change in the area covered by the pixels, depicting the collapsed structure. Therefore, the DPM technique with SAR data can be used for damage assessment with accurate and comprehensive detection after an event. In addition, we classify the amplitude information using support vector machine (SVM) and maximum likelihood classification algorithms. An investigation committee was formed to determine the cause of the collapse of the retaining wall and to suggest technical and institutional measures and alternatives to prevent similar incidents from reoccurring. The report from the committee revealed that the incident was caused by a combination of factors that were not carried out properly.

• Journal of Microbiology and Biotechnology
• /
• 제32권9호
• /
• pp.1120-1125
• /
• 2022

Japanese encephalitis virus (JEV), the causative agent of Japanese encephalitis (JE), is an importantly zoonotic, vector-borne virus widely prevalent in Asia. Although JE has been well controlled in China, its prevalence remains a huge threat to the pig industry as well as human health. Herein, we report on our molecular and serological investigations of JEV among pigs from different regions in Hunan Province of China from 2019 to 2021. Collectively, 19.27% (583/3026, 95% Confidential Interval (CI) 17.86-20.68) of sampled pigs were positive for JEV IgG antibody as revealed by indirect enzyme-linked immunosorbent assay, and the seroprevalence of JEV among pigs was significantly associated with the development stage and breeding scale (p < 0.01). Meanwhile, 10.99% (42/382, 95% CI 7.86-14.13) of tissue samples of pigs with suspected clinical symptoms of JE and 23.44% (15/64, 95% CI 13.06-33.82) of mosquito batches were JEV-positive via reverse polymerase chain reaction. In addition, the complete E gene sequences of 14 JEV strains identified in this study were amplified and sequenced. Phylogenetic analysis showed that all 14 JEV strains belonged to genotype I-b and displayed a distinct genetic relationship to the present JEV vaccine strain (SA14-14-2). In conclusion, our results revealed not only the severe prevalence of JEV in Hunan Province, but also that JEV I-b might be the predominant genotype in Hunan Province, suggesting therefore that effective measures for JE control are urgently needed.