• 비파괴검사학회지
• /
• 제15권4호
• /
• pp.531-539
• /
• 1996

본 연구에서는 초음파 신호형상인식법을 이용하여 용접부의 인공 결함을 분류하기 위한 연구를 실시하였다. 이를 위해 신호처리 및 특징 변수를 추출할 때에 많은 사용자 정의 변수를 가지는 신호 형상 인식 패키지를 개발하였으며 디지탈 신호처리, 특징 변수 추출, 특징 변수의 선택, 분류기 선정 등의 과정을 일괄적으로 처리하였다. 특히, 선형 분류기, 경험적 Bayesian 분류기 등의 통계적 분류기와 신경회로망 분류기를 함께 사용하여 비교, 검토하였다. 이에 관한 적용 연구로 노치와 구멍으로 이루어진 인공 결함을 분류하였다. 그 결과 인공결함 분류에서 높은 인식률을 얻었으며, 특히 적절히 학습 시켰을 경우 신경회로망 분류기가 통계적 분류기에 비하여 인식률 면에서 유리하였다.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제15권5호
• /
• pp.437-444
• /
• 2015

In simulation-based circuit optimization, many simulation runs may be wasted while evaluating infeasible designs, i.e. the designs that do not meet the constraints. To avoid such a waste, this paper investigates the use of support vector machine (SVM) classifiers in predicting the design's feasibility prior to simulation and the optimal selection of the SVM parameters, namely, the Gaussian kernel shape parameter ${\gamma}$ and the misclassification penalty parameter C. These parameters affect the complexity as well as the accuracy of the model that SVM represents. For instance, the higher ${\gamma}$ is good for detailed modeling and the higher C is good for rejecting noise in the training set. However, our empirical study shows that a low ${\gamma}$ value is preferable due to the high spatial correlation among the circuit design candidates while C has negligible impacts due to the smooth and clean constraint boundaries of most circuit designs. The experimental results with an LC-tank oscillator example show that an optimal selection of these parameters can improve the prediction accuracy from 80 to 98% and model complexity by $10{\times}$.

• 비파괴검사학회지
• /
• 제15권2호
• /
• pp.395-406
• /
• 1995

초음파탐상시험을 통해 용접 결함의 종류를 정확히 구분하는 것은 정량적 비파괴시험을 위한 기본적인 단계로서 매우 중요한 문제인데, 이 문제는 최근 활발한 연구가 진행중인 초음파 형상 인식 기법의 적용에 의해 해결할 수 있다. 여기에서는 특징 추출, 특징 선택 그리고 결함 분류 등 초음파 형상 인식 기법의 세부 기술과 함께, 특히 최근 효율적인 분류기로 관심을 모으고 있는 확률 신경 회로망의 적용에 대해 논의하였다. 그리고 강 용접부 내부에 존재하는 결함을 균열, 기공, 슬래그 혼입의 3 종류로 분류하는 문제에 확률 신경 회로망을 적용한 예를 통하여, 초음파 형상 인식 기법의 효용성을 검증하였다. 또한 민감한 특징을 효율적으로 선별하는데 널리 사용되는 전방 특징 선택법과 그 적용에 대해서도 논의하였다.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2008년도 춘계종합학술대회 A
• /
• pp.877-881
• /
• 2008

In this paper, we propose a real-time people tracking system with multiple CCD cameras for security inside the building. The camera is mounted from the ceiling of the laboratory so that the image data of the passing people are fully overlapped. The implemented system recognizes people movement along various directions. To track people even when their images are partially overlapped, the proposed system estimates and tracks a bounding box enclosing each person in the tracking region. The approximated convex hull of each individual in the tracking area is obtained to provide more accurate tracking information. To achieve this goal, we propose a method for 3D walking human tracking based on the IMPRESARIO framework incorporating cascaded classifiers into hypothesis evaluation. The efficiency of adaptive selection of cascaded classifiers have been also presented. We have shown the improvement of reliability for likelihood calculation by using cascaded classifiers. Experimental results show that the proposed method can smoothly and effectively detect and track walking humans through environments such as dense forests.

• Journal of Electrical Engineering and Technology
• /
• 제10권2호
• /
• pp.670-675
• /
• 2015

Medical imaging is one of the most powerful tools for gaining information about internal organs and tissues. It is a challenging task to develop sophisticated image analysis methods in order to improve the accuracy of diagnosis. The objective of this paper is to develop a Computer Aided Diagnostics (CAD) scheme for Brain Tumour detection from Magnetic Resonance Image (MRI) using active contour models and to investigate with several approaches for improving CAD performances. The problem in clinical medicine is the automatic detection of brain Tumours with maximum accuracy and in less time. This work involves the following steps: i) Segmentation performed by Fuzzy Clustering with Level Set Method (FCMLSM) and performance is compared with snake models based on Balloon force and Gradient Vector Force (GVF), Distance Regularized Level Set Method (DRLSE). ii) Feature extraction done by Shape and Texture based features. iii) Brain Tumour detection performed by various tree classifiers. Based on investigation FCMLSM is well suited segmentation method and Random Forest is the most optimum classifier for this problem. This method gives accuracy of 97% and with minimum classification error. The time taken to detect Tumour is approximately 2 mins for an examination (30 slices).

• Journal of Information Processing Systems
• /
• 제14권2호
• /
• pp.418-434
• /
• 2018

With the increasing use of the Internet and electronic documents, automatic text categorization becomes imperative. Several machine learning algorithms have been proposed for text categorization. The k-nearest neighbor algorithm (kNN) is known to be one of the best state of the art classifiers when used for text categorization. However, kNN suffers from limitations such as high computation when classifying new instances. Instance selection techniques have emerged as highly competitive methods to improve kNN through data reduction. However previous works have evaluated those approaches only on structured datasets. In addition, their performance has not been examined over the text categorization domain where the dimensionality and size of the dataset is very high. Motivated by these observations, this paper investigates and analyzes the impact of instance selection on kNN-based text categorization in terms of various aspects such as classification accuracy, classification efficiency, and data reduction.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제13권1호
• /
• pp.73-82
• /
• 2013

In pattern classification, feature selection is an important factor in the performance of classifiers. In particular, when classifying a large number of features or variables, the accuracy and computational time of the classifier can be improved by using the relevant feature subset to remove the irrelevant, redundant, or noisy data. The proposed method consists of two parts: a wrapper part with an improved genetic algorithm(GA) using a new reproduction method and a filter part using mutual information. We also considered feature selection methods based on mutual information(MI) to improve computational complexity. Experimental results show that this method can achieve better performance in pattern recognition problems than other conventional solutions.

• 대한원격탐사학회지
• /
• 제28권5호
• /
• pp.489-499
• /
• 2012

이 논문에서는 원격탐사 자료의 분류를 목적으로 서로 다른 훈련 집단들과 분류자들로부터 생성된 분류 결과들을 결합하는 분류 틀을 제안하였다. 제안 분류 틀의 핵심 부분은 서로 다른 훈련 집단과 분류자들을 이용함으로써 분류 결과 사이의 다양성을 증가시켜서 결과적으로 분류 정확도를 향상시키는데 있다. 제안 분류 틀에서는 우선 서로 다른 샘플링 밀도를 가지는 서로 다른 훈련 집단들을 생성한 후에, 이들을 서로 다른 구분 능력을 나타내는 분류자들의 입력 훈련 자료로 사용한다. 그리고 초기 분류 결과들에 다수결 규칙을 적용하여 최종 분류 결과를 얻게 된다. 다중 시기 ENVISAT ASAR 자료를 이용한 토지 피복 분류사례 연구를 통해 제안 방법론의 적용 가능성을 검토하였다. 사례 연구에서 3개의 훈련 집단과 최대우도 분류자, 다층 퍼셉트론 분류자, support vector machine 등과 같은 3개의 분류자를 이용한 9개의 분류 결과를 결합하였다. 사례 연구 결과, 제안 분류 틀 안에서 토지 피복 구분에 관한 상호 보완적인 정보의 이용이 가능해져서 가장 높은 분류 정확도를 나타내었다. 서로 다른 결합들을 비교하였을 때, 다양성이 크지 않은 분류 결과들을 결합한 경우에는 분류 정확도의 향상이 나타나지 않았다. 따라서 다중 분류 시스템의 설계시 분류자들의 다양성을 확보하는 것이 중요함을 확인할 수 있었다.

• International Journal of Control, Automation, and Systems
• /
• 제4권2호
• /
• pp.187-196
• /
• 2006

A new GA-based methodology using information granules is suggested for the construction of fuzzy classifiers. The proposed scheme consists of three steps: selection of information granules, construction of the associated fuzzy sets, and tuning of the fuzzy rules. First, the genetic algorithm (GA) is applied to the development of the adequate information granules. The fuzzy sets are then constructed from the analysis of the developed information granules. An interpretable fuzzy classifier is designed by using the constructed fuzzy sets. Finally, the GA is utilized for tuning of the fuzzy rules, which can enhance the classification performance on the misclassified data (e.g., data with the strange pattern or on the boundaries of the classes). To show the effectiveness of the proposed method, an example, the classification of the Iris data, is provided.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2001년도 하계종합학술대회 논문집(3)
• /
• pp.9-12
• /
• 2001

Learning vector quantization (LVQ) is a supervised learning technique that uses class information to move the vector quantizer slightly, so as to improve the quality of the classifier decision regions. In this paper we propose a selection method of initial codebook vectors for a teaming vector quantization (LVQ3) using support vector machines (SVM). The method is experimented with artificial and real design data sets and compared with conventional methods of the condensed nearest neighbor (CNN) and its modifications (mCNN). From the experiments, it is discovered that the proposed method produces higher performance than the conventional ones and then it could be used efficiently for designing nonparametric classifiers.