• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.134-138
• /
• 1998

Unification of Kohone SOM(Self-Organizing Maps) neural network with the branch-and-bound algorithm is presented for clustering large set of patterns. The branch-and-bound search technique is employed for designing coarse neural network learning paradaim. Those unification can be use for clustering or calssfication of large patterns. For classfication purposes further usefulness is possible, since only two clusters exists in the SOM neural network of each nodes. The result of experiments show the fast learning time, the fast recognition time and the compactness of clustering.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.9C
• /
• pp.853-858
• /
• 2006

TMHMM(Tied Mixture Hidden Markov Model) is an important approach to reduce the number of free parameters in speech recognition. However, this model suffers from a degradation in recognition accuracy due to its GPDF (Gaussian Probability Density Function) clustering error. This paper proposes a clustering algorithm, called HCNN(Homogeneous Centroid Neural network), to cluster acoustic feature vectors in TMHMM. Moreover, the HCNN uses the heterogeneous distance measure to allocate more code vectors in the heterogeneous areas where probability densities of different states overlap each other. When applied to Korean digit isolated word recognition, the HCNN reduces the error rate by 9.39% over CNN clustering, and 14.63% over the traditional K-means clustering.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.10a
• /
• pp.153-157
• /
• 2004

본 논문은 패턴인식을 위해 사용할 수 있는 감독학습을 이용한 supervised IAFC neural network 1과 supervised IAFC neural network 2를 제안하였다 Supervised IAFC neural network 1과 supervised IAFC neural network 2는 LVQ(Learning Vector Quantization)를 퍼지화한 새로운 퍼지 학습법칙을 사용하고 있다. 이 새로운 퍼지 학습 법칙은 기존의 학습률 대신에 퍼지화된 학습률을 사용하고 있는데, 이 퍼지화된 학습률은 조건 확률을 퍼지화 한 것에 근간을 두고 있다. Supervised IAFC neural network 1과 supervised IAFC neural network 2의 성능과 오류역전파 신경회로망의 성능을 비교하기 위하여 iris 데이터를 사용하였는데, 실험결과 supervised IAFC neural network 2 의 성능이 오류역전파 신경회로망의 성능보다 우수함이 입증되었다.

• ETRI Journal
• /
• v.43 no.1
• /
• pp.74-81
• /
• 2021

The mixture model is a very powerful and flexible tool in clustering analysis. Based on the Dirichlet process and parsimonious Gaussian distribution, we propose a new nonparametric mixture framework for solving challenging clustering problems. Meanwhile, the inference of the model depends on the efficient online variational Bayesian approach, which enhances the information exchange between the whole and the part to a certain extent and applies to scalable datasets. The experiments on the scene database indicate that the novel clustering framework, when combined with a convolutional neural network for feature extraction, has meaningful advantages over other models.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.1 no.3
• /
• pp.69-76
• /
• 2008

This paper concerns Fuzzy Radial Basis Function Neural Network (FRBFNN) and automatic rule generation of extraction of the FRBFNN by means of mountain clustering. In the proposed network, the membership functions of the premise part of fuzzy rules do not assume any explicit functional forms such as Gaussian, ellipsoidal, triangular, etc., so its resulting fitness values (degree of membership) directly rely on the computation of the relevant distance between data points. Also, we consider high-order polynomial as the consequent part of fuzzy rules which represent input-output characteristic of sup-space. The number of clusters and the centers of clusters are automatically generated by using mountain clustering method based on the density of data. The centers of cluster which are obtained by using mountain clustering are used to determine a degree of membership and weighted least square estimator (WLSE) is adopted to estimate the coefficients of the consequent polynomial of fuzzy rules. The effectiveness of the proposed model have been investigated and analyzed in detail for the representative nonlinear function.

• Proceedings of the KIEE Conference
• /
• 2008.10b
• /
• pp.239-240
• /
• 2008

본 논문에서는 Subtractive clustering 알고리즘을 이용한 Fuzzy Radial Basis Function Neural Network (FRBFNN)의 규칙 수를 자동적으로 생성하는 방법을 제시한다. FRBFNN은 멤버쉽 함수로써 기존 RBFNN에서 가우시안이나 타원형 형태의 특정 RBF를 사용하는 구조와 달리 Fuzzy C-Means clustering 알고리즘에서 사용하는 거리에 기한 멤버쉽 함수를 사용하여 전반부의 공간 분할 및 활성화 레벨을 결정하는 구조이다. 본 논문에서는 데이터의 밀집도에 기반을 두어 클러스터링을 하는 Subtractive clustering 알고리즘을 사용하여 퍼지 규칙의 수와 같은 의미를 갖는 분할할 입력공간의 수와 분할된 입력공간의 중심값을 동정하며, Least Square Estimator (LSE) 알고리즘을 사용하여 후반부 다항식의 계수를 추정 한다.

• Journal of Electrical Engineering and Technology
• /
• v.2 no.2
• /
• pp.157-164
• /
• 2007

Diagnosis techniques based on the dissolved gas analysis(DGA) have been developed to detect incipient faults in power transformers. Various methods exist based on DGA such as IEC, Roger, Dornenburg, and etc. However, these methods have been applied to different problems with different standards. Furthermore, it is difficult to achieve an accurate diagnosis by DGA without experienced experts. In order to resolve these drawbacks, this paper proposes a novel diagnosis method using fuzzy clustering and a radial basis neural network(RBFNN). In the neural network, fuzzy clustering is effective for selecting the efficient training data and reducing learning process time. After fuzzy clustering, the RBF neural network is developed to analyze and diagnose the state of the transformer. The proposed method measures the possibility and degree of aging as well as the faults occurred in the transformer. To demonstrate the validity of the proposed method, various experiments are performed and their results are presented.

• The Journal of Information Technology
• /
• v.8 no.1
• /
• pp.119-124
• /
• 2005

Learning procedure in the neural network is updating of weights between neurons. Unadequate initial learning coefficient causes excessive iterations of learning process or incorrect learning results and degrades learning efficiency. In this paper, adaptive learning algorithm is proposed to increase the efficient in the learning algorithms of Kohonens Self-Organization Neural networks. The algorithm updates the weights adaptively when learning procedure runs. To prove the efficiency the algorithm is experimented to clustering of the random weight. The result shows improved learning rate about 42~55% ; less iteration counts with correct answer.

• Investigative Magnetic Resonance Imaging
• /
• v.6 no.2
• /
• pp.137-146
• /
• 2002

Purpose : This paper introduces an improved classification approach which adopts a deterministic relaxation method and an agglomerative clustering technique for the classification of MRI using neural network. The proposed approach can solve the problems of convergency to local optima and computational burden caused by a large number of input patterns when a neural network is used for image classification. Materials and methods : Application of Hopfield neural network has been solving various optimization problems. However, major problem of mapping an image classification problem into a neural network is that network is opt to converge to local optima and its convergency toward the global solution with a standard stochastic relaxation spends much time. Therefore, to avoid local solutions and to achieve fast convergency toward a global optimization, we adopt MFA to a Hopfield network during the classification. MFA replaces the stochastic nature of simulated annealing method with a set of deterministic update rules that act on the average value of the variable. By minimizing averages, it is possible to converge to an equilibrium state considerably faster than standard simulated annealing method. Moreover, the proposed agglomerative clustering algorithm which determines the underlying clusters of the image provides initial input values of Hopfield neural network. Results : The proposed approach which uses agglomerative clustering and deterministic relaxation approach resolves the problem of local optimization and achieves fast convergency toward a global optimization when a neural network is used for MRI classification. Conclusion : In this paper, we introduce a new paradigm to classify MRI using clustering analysis and deterministic relaxation for neural network to improve the classification results.

• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06a
• /
• pp.491-493
• /
• 2011

Traffic sign detection is the domain of automatic driver assistant systems. There are literatures for traffic sign detection using color information, however, color-based method contains ill-posed condition and to extract the region of interest is difficult. In our work, we propose a method for traffic sign detection using k-means clustering method, back-propagation neural network, and projection histogram features that yields the robustness for ill-posed condition. Using the color information of traffic signs enables k-means algorithm to cluster the region of interest for the detection efficiently. In each step of clustering, a cluster is verified by the neural network so that the cluster exactly represents the location of a traffic sign. Proposed method is practical, and yields robustness for the unexpected region of interest or for multiple detections.