• 대한산업공학회지
• /
• 제32권2호
• /
• pp.133-140
• /
• 2006

Large signal type data sets are difficult to classify, especially if the data sets are non-stationary. In this paper, large signal type and non-stationary data sets are wavelet transformed so that distinct features of the data are extracted in wavelet domain rather than time domain. For the classification of the data, a few wavelet coefficients representing class properties are employed for statistical classification methods : Linear Discriminant Analysis, Quadratic Discriminant Analysis, Neural Network etc. The application of our wavelet-based feature selection method to a mass spectrometry data set for ovarian cancer diagnosis resulted in 100% classification accuracy.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.467-469
• /
• 2003

This research aimed to develop forest type classification technique for the mixed forest with coniferous and broad-leaved species using the high resolution satellite data. QuickBird data was used as satellite data. The method of this research was to extract satellite data for every single tree crown using image segmentation technique, then to evaluate the accuracy of classification by changing grouping criteria such as tree species, families, coniferous or broad-leaved species, and timber prices. As a result, the classification of tree species and families level was inaccurate, on the other hand, coniferous or broad-leaved species and timber price level was high accurate.

• 한국의류산업학회지
• /
• 제13권5호
• /
• pp.734-742
• /
• 2011

This study has been undertaken as a basic research for automatic pattern design for baseball uniforms manufactured under custom-MTM system, propose building up of a system whereby various partial patterns are combined under an automatic design system and develop a multi-combination type pattern searching algorithm which allows development of a various designs. As a result of this, type classification based on pattern design elements includes side, open, collar, facing and panel type. Design have been divided into coarse classification ranging from level 1 to 7 according to pattern design elements, based on a design distribution chart. Out of 7 such levels, 3 major types determining design which are, more specifically, level 1 sleeve type, level 2 open type and level 3 collar type, have been taken and combined to determine a total of 12 types to be used for design classification codes. Respective name of style and patterns have been coded using alphabet and numerals. Totally, pattern searching algorithm of multi-combination type has been developed whereby combination of patterns belonging to a specific style can be retrieved automatically once that style name is designated on the automatic pattern design system.

• 한국안전학회지
• /
• 제34권5호
• /
• pp.119-131
• /
• 2019

Among the themes for CSR(Corporate Social Responsibility), consumer issues include protecting the health and safety of consumers who purchase and use the products. In particular, ensuring product safety is a major theme of consumer issues for corporate social responsibility. Currently, the government implements the Electrical Appliances and Consumer Products Safety Control Act for product safety management and selects products that may harmful to consumers as safety control items, and manages the products by designating them as 4 types of safety certification, safety confirmation, supplier conformity verification, and safety standard compliance. In this paper, we propose management plans for the establishment of a more reasonable classification type of safety management target for 48 items of consumer products to be controlled by the act, and confirm the validity of the plan. First, we perform cluster analysis using data for CISS (Consumer Injury Surveillance System) to derive a new classification type of the safety management target. Next, we compare the results of the cluster analysis with the classification type of the act and the existing scenario classification method RAS (Risk Assessment by Scenario) and the causal network method RAMP (Risk Assessment Method based on Probability). Based on these results, we propose two new plans of safety management target classification and verify its validity.

• 치과방사선
• /
• 제24권2호
• /
• pp.227-236
• /
• 1994

The purpose of this study was to observe mandibular condyle shape in an asymptomatic population. In order to carry out this study, 96 temporomandibular joints in 48 adults(22 males, 26 females), who were asymptomatic for temporomandibular disturbances and had no history of prosthodontic or orthodontic treatments, were selected, and radiographed using the Sectograph(Denar Co., U.S.A.) for lateral and frontal individualized corrected TMJ tomograph and submentovertex radiograph. Mandibular condyles were classified morphologically, and measured medioateral and anteroposterior dimensions and condylar angulation. The obtained results were as follows. 1. In the classification of condyle shape on lateral tomographs, 94.8% were convex type and 5.2% were angled type. 2. In the classification of condyle shape on frontal tomographs, 45.3% were convex type, 32.0% were round type, 16.0% were flat type, and 6.7% were angled type. 3. In the classification of condyle shape on submentovertex radiographs, 34.5% were flat-convex type, 22.9% were flat-flat type, 20.8% were concave-convex type, 19.8% were convex-convex type, and 1.0% were concave-flat type and convex-flat type. Concave-concave type, convex-concave type, and flat-concave type were not observed. 4. The average mediolateral legth of the condyle was 19.3㎜ and the average anteroposterior length was 9.4㎜. The average angle between the long axis of condyle and the coronal plane made on submentovertex view was 19.6 degrees.

• Geomechanics and Engineering
• /
• 제37권3호
• /
• pp.263-277
• /
• 2024

Artificial Neural Networks (ANNs) are artificial learning algorithms that provide successful results in solving many machine learning problems such as classification, prediction, object detection, object segmentation, image and video classification. There is an increasing number of studies that use ANNs as a prediction tool in soil classification. The aim of this research was to understand the role of hyperparameter optimization in enhancing the accuracy of ANNs for soil type classification. The research results has shown that the hyperparameter optimization and hyperparamter optimized ANNs can be utilized as an efficient mechanism for increasing the estimation accuracy for this problem. It is observed that the developed hyperparameter tool (HyperNetExplorer) that is utilizing the Covariance Matrix Adaptation Evolution Strategy (CMAES), Genetic Algorithm (GA) and Jaya Algorithm (JA) optimization techniques can be successfully used for the discovery of hyperparameter optimized ANNs, which can accomplish soil classification with 100% accuracy.

• 대한공간정보학회지
• /
• 제17권2호
• /
• pp.55-60
• /
• 2009

본 연구에서는 경주시 내남면을 대상으로 KOMPSAT-2 MSC(Multi Spectral Camera) 영상(2007.06.12)을 기반으로 NDVI(Normalized Difference Vegetation Index)와 TCT(Tasseled-Cap Transformation) 영상 알고리즘을 적용하여 DN 분포도를 작성 하였다. NDVI 및 TCT DN 분포도와 산림 현장 조사 결과와의 비교 분석을 통하여 임상 분류 정확도 향상을 위한 영상 알고리즘 변별력 분석을 수행하고 마지막으로 현장조사 자료와의 중첩 분석을 통하여 임상분류 정확성을 검증 하였다. 본 연구를 통하여 KOMPSAT-2 MSC 영상을 이용하여 임상 분류 자동화 실용성에 대한 검토와 정밀 산림 임상도 제작과정에서 저비용 고효율성을 기대할 수 있으리라 사료된다.

• 한국음향학회지
• /
• 제13권4호
• /
• pp.5-11
• /
• 1994

본 논문에서는 형식별 블럭분할에 기초한 다중신경망과 퍼지추론에 의한 한글 형식분류에 대해 연구하였다. 효과적인 자모분류를 위해 입력문자에 대해서 한글의 각 형식을 구성하는 자모의 영역으로 분할하는 블럭분할방법을 제한하였으며, 분할된 블럭이 형식에 따라 적응적으로 변화할 수 있도록 하였다. 또한 분류율의 향상을 위해 전체신경망과 부분신경망으로 이루어진 다중신경망을 구성하였으며, 퍼지추론에 의해 한글 형식을 판정하였다. 비교, 실험을 통하여 제안된 방법의 타당성을 검증하였으며, $92.6\%$의 분류율을 나타내므로서 유효성을 확인하였다.

• 항공우주기술
• /
• 제6권1호
• /
• pp.213-221
• /
• 2007

본 연구에서는 2002년 4월에 획득된 Hyperion 초분광 원격탐사 자료를 이용하여 산불연료지도 제작을 위한 객체기반 분류 기법을 제시하였으며, 또한 객체기반 분석결과와 화소기반 분석결과를 비교해 보았다. 이를 위해 우선적으로 Hyperion 위성영상에 있는 잡음 화소 보정과 잡음 밴드를 제거하였으며, 또한 정확한 자료 처리를 위해 대기보정을 수행하였다. 산불 연료 지도 제작을 위한 방법은 분광혼합분석(SMA) 처리 결과를 재구성하여 얻었다. 객체 기반 접근 방법은 세그먼트 기반의 endmember 선택방법을 활용하였으며, 화소기반 분석은 표준 분광혼합분석기법을 적용하였다. 검증 및 비교를 위해서는 고해상도 칼라 항공정사영상이 활용되었다.

• 한국도로학회논문집
• /
• 제16권4호
• /
• pp.127-134
• /
• 2014

PURPOSES : To operate more efficient traffic management system, it is utmost important to detect the change in congestion level on a freeway segment rapidly and reliably. This study aims to develop classification method of congestion change type. METHODS: This research proposes two classification methods to capture the change of the congestion level on freeway segments using the dedicated short range communication (DSRC) data and the vehicle detection system (VDS) data. For developing the classification methods, the decision tree models were employed in which the independent variable is the change in congestion level and the covariates are the DSRC and VDS data collected from the freeway segments in Korea. RESULTS : The comparison results show that the decision tree model with DSRC data are better than the decision tree model with VDS data. Specifically, the decision tree model using DSRC data with better fits show approximately 95% accuracies. CONCLUSIONS : It is expected that the congestion change type classified using the decision tree models could play an important role in future freeway traffic management strategy.