• 전자공학회논문지B
• /
• 제32B권5호
• /
• pp.765-776
• /
• 1995

This paper describes a speaker adaptation method to improve the recognition performance of MLP(multiLayer Perceptron) based HMM(Hidden Markov Model) speech recognizer. In this method, we use lst-order linear transformation network to fit data of a new speaker to the MLP. Transformation parameters are adjusted by back-propagating classification error to the transformation network while leaving the MLP classifier fixed. The recognition system is based on semicontinuous HMM's which use the MLP as a fuzzy vector quantizer. The experimental results show that rapid speaker adaptation resulting in high recognition performance can be accomplished by this method. Namely, for supervised adaptation, the error rate is signifecantly reduced from 9.2% for the baseline system to 5.6% after speaker adaptation. And for unsupervised adaptation, the error rate is reduced to 5.1%, without any information from new speakers.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2008년도 제39회 하계학술대회
• /
• pp.243-244
• /
• 2008

기존의 고정 부하 중심점 방식의 ULTC 운전제어전략은 계절별 부하 변화에 따른 전압보상을 효율적으로 실현하기 어렵다. 따라서 가변 부하 중심점 방식의 ULTC 운전제어전략들이 제안되는데, ANN이나 Fuzzy 멤버쉽 함수를 설계하는 문제는 고도의 전문적 설계경험과 상당한 시간을 요구한다. 따라서 본 연구에서는 적용이 쉬운 최소거리 판별기법을 적용, 부하변화에 따라 ULTC의 운전 부하 중심점을 이동시켜, 전압보상 범위를 개선시킬 수 있는 디지털 ULTC 운전 제어전략을 제안한다.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2005년도 춘계학술발표대회
• /
• pp.443-446
• /
• 2005

본 논문에서는 패턴 분류문제에서 특징의 분포와 빈도를 고려하는 FMM(Fuzzy Min-Max) 신경망 구조와 이를 이용한 특징 분석 기법을 소개한다. 이는 기존의 모델에서 균일한 가중치를 고려했을때 비정상적 학습데이터에 학습 효과가 민감하게 왜곡되는 현상을 방지한다. 또한 학습된 신경망으로부터 각 특징의 중요도를 분석할 수 있게 한다. 본 연구에서는 제안된 모델의 특성을 소개하고 특징 값과 하이퍼박스 간의 관계로부터 특징의 연관도 요소, 중요도 평가 및 특징의 서열화 기법을 제시한다. 이는 패턴 분류 신경망의 노드수를 최적화 함으로써 학습 및 분류 과정에서 연산의 효율성을 증대시킨다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제2권3호
• /
• pp.227-230
• /
• 2002

A new iris segmentation and recognition method is described. Combining a statistical classification and elastic boundary fitting, the iris is first segmented robustly and accurately. Once the iris is segmented, one-dimensional signals are computed in the iris and decomposed into multiple frequency bands. Each decomposed signal is approximated by a piecewise linear curve connecting a small set of node points. The node points represent features of each signal. The similarity measture between two iris images is the normalized cross-correlation coefficients between simplified signals.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제4권2호
• /
• pp.249-253
• /
• 2004

When we decide the software quality on the basis of the software measurement, the transitive property which is a requirement for an equivalence relation is not always satisfied. Therefore, we propose a scheme for classifying the software quality that employs a tolerance relation instead of an equivalence relation. Given the experimental data set, the proposed scheme generates the tolerant classes for elements in the experiment data set, and generates the tolerant ranges for classifying the software quality by clustering the means of the tolerance classes. Through the experiment, we showed that the proposed scheme could product very useful and valid results. That is, it has no problems that we use as the criteria for classifying the software quality the tolerant ranges generated by the proposed scheme.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제3권2호
• /
• pp.227-232
• /
• 2003

Recently, support vector learning attracts an enormous amount of interest in the areas of function approximation, pattern classification, and novelty detection. One of the main reasons for the success of the support vector machines(SVMs) seems to be the availability of global and sparse solutions. Among the approaches sharing the same reasons for success and exhibiting a similarly good performance, we have KFD(kernel Fisher discriminant) approach. In this paper, we consider the problem of function approximation utilizing both predetermined basis functions and the KFD approach for regression. After reviewing support vector regression, semi-parametric approach for including predetermined basis functions, and the KFD regression, this paper presents an extension of the conventional KFD approach for regression toward the direction that can utilize predetermined basis functions. The applicability of the presented method is illustrated via a regression example.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제3권2호
• /
• pp.178-186
• /
• 2003

Recently, many researchers have been interested in biometric systems such as fingerprint, handwriting, key-stroke patterns and human iris. From the viewpoint of reliability and robustness, iris recognition is the most attractive biometric system. Moreover, the iris recognition system is a comfortable biometric system, since the video image of an eye can be taken at a distance. In this paper, we discuss human iris recognition, which is based on accurate iris localization, robust feature extraction, and Neural Network classification. The iris region is accurately localized in the eye image using a multiresolution active snake model. For the feature representation, the localized iris image is decomposed using wavelet transform based on dyadic Haar wavelet. Experimental results show the usefulness of wavelet transform in comparison to conventional Gabor transform. In addition, we present a new method for setting initial weight vectors in competitive learning. The proposed initialization method yields better accuracy than the conventional method.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권3호
• /
• pp.173-180
• /
• 2016

Fault detection and diagnosis is a task to monitor the occurrence of faults and pinpoint the exact location of faults in the system. Fault detection and diagnosis is gaining importance in development of efficient, advanced and safe industrial systems. Three phase inverter is one of the most common and excessively used power electronic system in industries. A fault diagnosis system is essential for safe and efficient usage of these inverters. This paper presents a fault detection technique and fault classification algorithm. A new feature extraction approach is proposed by using three-phase load current in three-dimensional space and neural network is used to diagnose the fault. Neural network is responsible of pinpointing the fault location. Proposed method and experiment results are presented in detail.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권3호
• /
• pp.157-162
• /
• 2016

In this paper, a yield prediction model based on relevance vector machine (RVM) and a granular computing model (quotient space theory) is presented. With a granular computing model, massive and complex meteorological data can be analyzed at different layers of different grain sizes, and new meteorological feature data sets can be formed in this way. In order to forecast the crop yield, a grey model is introduced to label the training sample data sets, which also can be used for computing the tendency yield. An RVM algorithm is introduced as the classification model for meteorological data mining. Experiments on data sets from the real world using this model show an advantage in terms of yield prediction compared with other models.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2005년도 가을 학술발표논문집 Vol.32 No.2 (2)
• /
• pp.769-771
• /
• 2005

데이터 특성이 연속적이고 애매할 때 퍼지규칙으로 분류 규칙을 표현하는 것은 매우 유용하고 효과적이다. 그러나 일반적으로 정확하지 않은 데이터 특성에 대해서 소속함수를 결정한다는 것은 어려운 일이다. 본 논문에서는 진화알고리즘을 이용하여 효과적인 퍼지 분류 규칙을 자동으로 생성하는 방법을 제안한다. 제안한 방법에서 규칙의 정확성과 이해성을 고려하여 최적화된 소속함수를 생성하기 위해 진화알고리즘을 사용한다. 먼저 지도 군집화로 진화를 위한 초기 소속함수를 생성한다. 진화알고리즘은 전역적 최적 해를 찾는데 효과적이다. 그러나 시간에 대한 효율성이 낮다. 특히 모델 최적화 문제에서는 개체 평가 단계에서 많은 시간이 소요된다. 따라서 본 논문에서는 전체 데이터를 여러 개의 부분 데이터들로 나누고 개체들은 전체 데이터 대신 매번 부분 데이터를 임의적으로 선택하여 개체를 평가함으로써 수행 시간을 단축시킬 수 있는 진화 방법을 제안한다. 제안한 퍼지 분류 규칙 생성 방법의 타당성을 검증하기 위한 실험 데이터로 UCI에서 제공하는 데이터들을 사용하였으며, 실험 결과는 기존 방법에 비해 평균적으로 더 효과적임을 확인하였다.