• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.12
• /
• pp.2937-2943
• /
• 2013

This paper introduces a real-time flocking simulation framework through radial basis function(RBF). The proposed framework first divides the entire environment into a grid structure and then assign a vector per each cell. These vectors are automatically calculated by using RBF function, which is parameterized from user-input control lines. Once the construction of vector field is done, then, flocks determine their path by following the vector field flow. The collision with static obstacles are modeled as a repulsive vector field, which is ultimately over-layed on the existing vector field and the inter-individual collision is also handled through fast lattice-bin method.

• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.1966-1967
• /
• 2011

본 연구에서는 방사형 기저 함수를 이용한 다항식 신경회로망(Polynomial Neural Network) 분류기를 제안한다. 제안된 모델은 PNN을 기본 구조로 하여 1층의 다항식 노드 대신에 다중 출력 형태의 방사형 기저 함수를 사용하여 각 노드가 방사형 기저 함수 신경회로망(RBFNN)을 형성한다. RBFNN의 은닉층에는 fuzzy 클러스터링을 사용하여 입력 데이터의 특성을 고려한 적합도를 사용하였다. 제안된 분류기는 입력변수의 수와 다항식 차수가 모델의 성능을 결정함으로 최적화가 필요하며 본 논문에서는 Multiobjective Particle Swarm Optimization(MoPSO)을 사용하여 모델의 성능뿐만 아니라 모델의 복잡성 및 해석력을 고려하였다. 패턴 분류기로써의 제안된 모델을 평가하기 위해 Iris 데이터를 이용하였다.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2012.05a
• /
• pp.332-335
• /
• 2012

본 논문에서는 CAT(Cellular Automata Transform) 성질을 이용하여 의료 영상의 단계적 변환을 제안한다. 적용 방법은 먼저, CAT 초기 값과 다양한 규칙에 따라 단계적으로 Gateway Values와 rule matrix에 대한 전이행렬 T를 이용하여 CAT 기저함수를 생성한다. 그런 다음, 생성된 CAT 기저함수를 의료 영상에 곱하여 영상을 다양한 방법으로 변환한다. 마지막으로, 키 공간 분석을 통하여 제안한 방법이 높은 영상 변환 및 보안의 성질을 가졌음을 검증한다.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2009.04a
• /
• pp.488-491
• /
• 2009

본 논문에서는 응력함수와 Kantorovich method를 이용하여 기저판(substrate)에 인장과 굽힘이 작용할 때 복합재 패치의 3차원 응력을 해석하였다. 면내 방향과 면외 방향의 두 응력함수에 가상 공액일의 법칙(Complementary virtual work principle)을 적용하였으며 복합재 패치의 자유 경계조건과 바닥의 기저판으로부터 전달되는 전단 수직 응력 조건을 부여하였다. 이를 통해서 패치 구조물의 지배방정식을 연립 미분 방정식 형태의 고유치 문제로 변환하여 응력함수를 구하였다. 위 방법의 타당성과 효용성을 검증하기 위한 수치 예제로 cross-ply, angle-ply, quasi-isotropic의 패치 적층 배열을 고려하였으며, 층간 응력함수 값이 자유 경계에서 최고치를 나타내고 패치 중심부로 갈수록 급격히 감소하는 모습을 확인하였다. 제안된 기법은 기저판에 인장하중이 작용하는 경우뿐만 아니라 굽힘 하중이 작용하는 경우에도 정확한 예측이 가능하여, 패치 구조물의 층간 응력을 포함한 3차원 응력을 해석하는데 있어서 효율적인 해석 도구로서 사용할 수 있을 것이라 사료된다.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.10 no.4
• /
• pp.550-563
• /
• 1999

A wavelet-Galerkin scheme based on the time-dependent Maxwell's equations is presented. Daubechies wavelet with two vanishing wavelet moments is expanded for basis function in spatial domain and Yee's leap-frog approach is applied. The shifted interpolation property of Daubechies wavelet family leads to the simplified formulations for inhomogeneous media without the additional matrices for the integral or material operator. The stability condition is formulated. The dispersion characteristics are analyzed and compared with those of finite difference time domain and multiresolution time domain methods. The analyses show the excellent trade-off between the regularity and the support width of the basis function. Although the basis function has only two vanishing wavelet moments, it is enough to provide negligible dispersive error in the numerical analysis and its compact support enables only several involved terms per nodes. The storage effectiveness, execution time reduction and accuracy of this scheme are demonstrated by calculating the resonant frequencies of the homogeneous and inhomogeneous cavities.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.3
• /
• pp.182-187
• /
• 2003

In this thesis, we propose the algorithm to improve the performance of speaker verification that is mapping feature parameters by using RBF neural network. There is a big difference between wire vector region and wireless one which comes from the same speaker. For wire/wireless speakers model production, speaker verification system should distinguish the wire/wireless channel that based on speech recognition system. And the feature vector of untrained channel models is mapped to the feature vector(LPC Cepstrum) of trained channel model by using RBF neural network. As a simulation result, the proposed algorithm makes 0.6%∼10.5% performance improvement compared to conventional method such as cepstral mean subtraction.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.1C
• /
• pp.105-112
• /
• 2009

In this paper, we propose a digital watermarking method using Multi-resolution Characteristic of 2D CAT (2D cellular automata transform). Firstly, we select the gateway values to generate a basis function and the basis function transforms images into cellular automata space. Then, we embed the random bit sequence as watermark in specific parts of cellular automata transform coefficients. The proposed method not only verifies higher fidelity than the existing method but also stronger stability on JPEG lossy compression, filtering, sharpening and noise through tests for robustness. Moreover, the proposed scheme allows only one 2D CAT basis function per gateway value. Since there are $2^{96}$ possible gateway values.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.14 no.3
• /
• pp.381-390
• /
• 2001

In this paper, a new improved crack analysis technique by Element-Free Galerkin(EFG) method is proposed, in which the singularity and the discontinuity of the crack successfully described by adding enrichment terms containing a singular basis function to the standard EFG approximation and a discontinuity function implemented in constructing the shape function across the crack surface. The standard EFG method requires considerable addition of nodes or modification of the model. In addition, the proposed method significantly decreases the size of system of equation compared to the previous enriched EFG method by using localized enrichment region near the crack tip. Numerical example show the improvement and th effectiveness of the previous method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.22 no.6
• /
• pp.735-740
• /
• 2012

In this paper, we introduce design methodologies of polynomial radial basis function neural network classifier with the aid of Principal Component Analysis(PCA) and Linear Discriminant Analysis(LDA). By minimizing the information loss of given data, Feature data is obtained through preprocessing of PCA and LDA and then this data is used as input data of RBFNNs. The hidden layer of RBFNNs is built up by Fuzzy C-Mean(FCM) clustering algorithm instead of receptive fields and linear polynomial function is used as connection weights between hidden and output layer. In order to design optimized classifier, the structural and parametric values such as the number of eigenvectors of PCA and LDA, and fuzzification coefficient of FCM algorithm are optimized by Artificial Bee Colony(ABC) optimization algorithm. The proposed classifier is applied to some machine learning datasets and its result is compared with some other classifiers.

• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.1343-1344
• /
• 2015

본 연구에서는 증분형 FCM(Incremental Fuzzy C-Means: Incremental FCM) 클러스터링 알고리즘을 기반으로 방사형 기저함수 신경회로망(Radial Basis Function Neural Networks: RBFNN) 패턴 분류기를 설계한다. 방사형 기저함수 신경회로망은 조건부에서 가우시안 함수 또는 FCM을 사용하여 적합도를 구하였지만, 제안된 분류기에서는 빅 데이터간의 적합도를 구하기 위해 증분형 FCM을 사용한다. 또한, 빅 데이터를 학습하기 위해 결론부에서 재귀최소자승법(Recursive Least Square Estimation: RLSE)을 사용하여 다항식 계수를 추정한다. 마지막으로 추론부에서는 증분형 FCM에서 구한 적합도와 재귀최소자승법으로 구한 다항식을 이용하여 최종 출력을 구한다.