• 산업공학
• /
• 제25권2호
• /
• pp.170-177
• /
• 2012

Process control is essential to operate the semiconductor process efficiently. This paper consider fault classification of semiconductor based cyclic signal for process control. In general, process signal usually take the different pattern depending on some different cause of fault. If faults can be classified by cause of faults, it could improve the process control through a definite and rapid diagnosis. One of the most important thing is a finding definite diagnosis in fault classification, even-though it is classified several times. This paper proposes the method that one-class classifier classify fault causes as each classes. Hotelling T2 chart, kNNDD(k-Nearest Neighbor Data Description), Distance based Novelty Detection are used to perform the one-class classifier. PCA(Principal Component Analysis) is also used to reduce the data dimension because the length of process signal is too long generally. In experiment, it generates the data based real signal patterns from semiconductor process. The objective of this experiment is to compare between the proposed method and SVM(Support Vector Machine). Most of the experiments' results show that proposed method using Distance based Novelty Detection has a good performance in classification and diagnosis problems.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2006년도 한국컴퓨터종합학술대회 논문집 Vol.33 No.1 (B)
• /
• pp.304-306
• /
• 2006

본 논문에서는 GWT(Gabor Wavelet Transform) 계수 에너지와 원 영상간의 영상 결합을 수행한 영상을 주성분 분석법(Principal Component Analysis)에 적용하여 얼굴 인식을 하는 방법을 제안한다. GWT는 가버 함수의 크기 변화와 방향 변화에 의해 생성된다. 따라서 GWT는 다양한 크기 변화와 방향 변화를 가지는 변환으로 특정 주파수 성분과 방향성을 가지는 영상 구조가 어디에 있는지의 지역적 정보를 효과적으로 표현할 수 있는 변환으로 알려져 있다. GWT를 통해 나온 계수 에너지를 추출하고 원 영상에 더하여 지역적 특성을 크게 만든 후에 통계적 방법 중 가장 많이 사용되어지고 검증을 받은 PCA를 사용하여 인식한다. GWT 계수의 에너지는 얼굴 윤곽선, 눈과 입, 얼굴과 머리의 경계 등 색감의 급격한 변화를 나타내는 곳의 정보를 표현을 해주기 때문에 특징점 추출에 사용되고 있지만 이를 전역적으로 이용하여 인식하는 방법에 관한 연구가 이루어지지 않고 있다. 본 논문에서는 에너지 값만으로 전체 얼굴 영상의 세부적 표현을 할 수 없기 때문에 원 영상과의 l:l 비율의 영상 결항을 한 후 얼굴 인식 처리에 사용한다. 이 영상을 얼굴인식에 사용하였을 때원본 영상을 사용하였을 때보다 오인식이 줄었다.

• 센서학회지
• /
• 제17권6호
• /
• pp.406-413
• /
• 2008

We have been implemented simple statistical pattern recognition methods for harmful gases classification using gas sensors array fabricated by MEMS (Micro Electro Mechanical System) technology. The performance of pattern recognition method as a gas classifier is highly dependent on the choice of pre-processing techniques for sensor and sensors array signals and optimal classification algorithms among the various classification techniques. We carried out pre-processing for each sensor's signal as well as sensors array signals to extract features for each gas. We adapted simple statistical pattern recognition algorithms, which were PCA (Principal Component Analysis) for visualization of patterns clustering and MLR (Multi-Linear Regression) for real-time system implementation, to classify harmful gases. Experimental results of adapted pattern recognition methods with pre-processing techniques have been shown good clustering performance and expected easy implementation for real-time sensing system.

• 한국농공학회논문집
• /
• 제55권4호
• /
• pp.73-82
• /
• 2013

The main objective of this study was to evaluate Representative Concentration Pathways (RCP) scenarios-based flood risk at a Si-Gun level. A bias correction using a quantile mapping method with the Generalized Extreme Value (GEV) distribution was performed to correct future precipitation data provided by the Korea Meteorological Administration (KMA). A series of proxy variables including CN80 (Number of days over 80 mm) and CX3h (Maximum precipitation during 3-hr) etc. were used to carry out flood risk assessment. Indicators were normalized by a Z-score method and weighted by factors estimated by principal component analysis (PCA). Flood risk evaluation was conducted for the four different time periods, i.e. 1990s, 2025s, 2055s, and 2085s, which correspond to 1976~2005, 2011~2040, 2041~2070, and 2071~2100. The average flood risk indices based on RCP4.5 scenario were 0.08, 0.16, 0.22, and 0.13 for the corresponding periods in the order of time, which increased steadily up to 2055s period and decreased. The average indices based on RCP8.5 scenario were 0.08, 0.23, 0.11, and 0.21, which decreased in the 2055s period and then increased again. Considering the average index during entire period of the future, RCP8.5 scenario resulted in greater risk than RCP4.5 scenario.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2017년도 추계학술대회
• /
• pp.674-677
• /
• 2017

허프변환은 에지 화소를 대상으로 직선을 검출하는 가장 대표적인 알고리즘이다. 허프변환은 단순한 직선 영상에서는 우수한 성능을 보이나 잡음이 있거나 복잡한 영상에서는 상당한 계산량을 요구하고 쉽게 의사직선을 검출하는 문제를 갖고 있다. 본 논문은 기존 허프변환의 문제를 개선하기 위한 직선 검출 알고리즘을 제안한다. 제안된 알고리즘은 허프변환을 수행하기 전에 주성분 분석을 이용해 에지 화소의 직선 정보를 검출한다. 에지 화소의 직선 정보를 근거로 유효 에지 화소에서 제한된 기울기 영역의 허프변환을 수행한다. 모의실험 결과들은 제안된 알고리즘이 계산량을 크게 줄이는 것은 물론 의사직선도 제거하는 것을 보여주고 있다.

• 한국식생활문화학회지
• /
• 제34권5호
• /
• pp.629-636
• /
• 2019

This study examined the quality characteristics of jelly prepared with green grape juice (GJ). The pH, $^{\circ}Brix$ value, color, texture, and sensory evaluation of the jelly were measured. The pH of the jelly significantly decreased with increasing amount of GJ over the range of 3.25-5.27. The $^{\circ}Brix$ value of the jelly showed a significantly higher result as the amount of GJ increased (p<0.001). Lightness (L) and redness (a) decreased with increasing amount of GJ, and yellowness (b) increased. In the texture measurement, the GJ-100 sample group with a high substitute rate of GJ showed high hardness, gumminess, and chewiness (p<0.001). The results of principal component analysis (PCA) showed that the sample groups with high GJ content were classified as having relatively strong yellowness, sweet aroma, metallic aroma, grassy aroma, sweetness, sourness, green grape skin taste, and astringency. In the acceptance test, the GJ-50 sample group was evaluated to be high in flavor (p<0.001) and overall acceptance (p<0.01). However, sample groups consisting of 50% or more GJ were evaluated to be significantly strong in terms of astringency. Therefore, further study needs to be conducted about improving astringency in the future.

• Current Optics and Photonics
• /
• 제3권2호
• /
• pp.150-153
• /
• 2019

Larynx cancer is a potentially terminal and severe type of neck and head cancer in which malignant cells start to grow and spread upwards in the larynx, or voice box. Smoking tobacco, drinking hot beverages and drinking alcohol are the main risk factors for these tumors. In this study, we aimed to develop a precise, accurate and rapid chemometrics assisted Raman spectroscopy method for diagnosis of larynx cancer in deparaffinized tissue samples. In the proposed method, samples were deparaffinized and 20 microns of each tissue were located on a coverslip. Both healthy (n = 13) and cancerous tissues (n = 13) were exposed to a Raman laser (785 nm) and excitations were recorded between wavenumbers of $50{\sim}1500cm^{-1}$. An Orthogonal Partial Least Square algorithm was applied to evaluate the Raman spectrum obtained. Sensitivity and specificity of the proposed method is high enough with the aid of Principal Component Analysis (PCA) to test the whole model. Healthy and cancerous tissues were accurately and precisely clustered. A rapid, easy and precise diagnosis algorithm was developed for larynx cancer. By this method, some useful data about differences in biomolecules of each group (phospholipids, amides, tyrosine, phenylalanine collagen etc.) was also obtained from the spectra. It is claimed that the optimized method has a great potential for clustering and separating tumor tissues from healthy ones. This novel, rapid, precise and objective diagnosis method may be an alternative for the conventional methods in literature for diagnosis of larynx cancer.

• 신재생에너지
• /
• 제16권4호
• /
• pp.33-40
• /
• 2020

The research topics of more than 1,900 wind energy papers registered in the Korean Journal Citation Index (KCI) were modeled into 25 topics using latent directory allocation (LDA), and their consistency was cross-validated through principal component analysis (PCA) of the document word matrix. Key research topics in the wind energy field were identified as "offshore, wind farm," "blade, design," "generator, voltage, control," 'dynamic, load, noise," and "performance test." As a new method to determine the similarity between research topics in journals, a systematic evaluation method was proposed to analyze the correlation between topics by constructing a journal-topic matrix (JTM) and clustering them based on topic similarity between journals. By evaluating 24 journals that published more than 20 wind energy papers, it was confirmed that they were classified into meaningful clusters of mechanical engineering, electrical engineering, marine engineering, and renewable energy. It is expected that the proposed systematic method can be applied to the evaluation of the specificity of subsequent journals.

• 한국멀티미디어학회논문지
• /
• 제24권5호
• /
• pp.667-675
• /
• 2021

The treatments for symptoms of Alzheimer's disease are being provided and for the early diagnosis several researches are undergoing. In this regard, by using T1-weighted images several classification techniques had been proposed to distinguish among AD, MCI, and Healthy Control (HC) patients. In this paper, we also used some traditional Machine Learning (ML) approaches in order to diagnose the AD. This paper consists of an improvised feature selection method which is used to reduce the model complexity which accounted an issue while utilizing the ML approaches. In our presented work, combination of subcortical and cortical features of 308 subjects of ADNI dataset has been used to diagnose AD using structural magnetic resonance (sMRI) images. Three classification experiments were performed: binary classification. i.e., AD vs eMCI, AD vs lMCI, and AD vs HC. Proposed Feature Selection method consist of a combination of Principal Component Analysis and Recursive Feature Elimination method that has been used to reduce the dimension size and selection of best features simultaneously. Experiment on the dataset demonstrated that SVM is best suited for the AD vs lMCI, AD vs HC, and AD vs eMCI classification with the accuracy of 95.83%, 97.83%, and 97.87% respectively.

• Journal of Ecology and Environment
• /
• 제45권1호
• /
• pp.24-39
• /
• 2021

Background: To address the lack of evidence supporting invasion by three invasive plant species (Imperata cylindrica, Lantana camara, and Chromolaena odorata) in tropical ecosystems, we compared the ecophysiological and leaf anatomical traits of these three invasive alien species with those of species native to Sempu Island, Indonesia. Data on four plant traits were obtained from the TRY Plant Trait Database, and leaf anatomical traits were measured using transverse leaf sections. Results: Two ecophysiological traits including specific leaf area (SLA) and seed dry weight showed significant association with plant invasion in the Sempu Island Nature Reserve. Invasive species showed higher SLA and lower seed dry weight than non-invasive species. Moreover, invasive species showed superior leaf anatomical traits including sclerenchymatous tissue thickness, vascular bundle area, chlorophyll content, and bundle sheath area. Principal component analysis (PCA) showed that leaf anatomical traits strongly influenced with cumulative variances (100% in grass and 88.92% in shrubs), where I. cylindrica and C. odorata outperformed non-invasive species in these traits. Conclusions: These data suggest that the traits studied are important for plant invasiveness since ecophysiological traits influence of light capture, plant growth, and reproduction while leaf anatomical traits affect herbivory, photosynthetic assimilate transport, and photosynthetic activity.