• The KIPS Transactions:PartB
• /
• v.10B no.6
• /
• pp.673-680
• /
• 2003

In this paper, we study the enhancement of VQ (Vector Quantization) design for text independent speaker recognition. In a concrete way, we present a non-iterative method which makes a vector quantization codebook and this method performs non-iterative learning so that the computational complexity is epochally reduced The proposed Classified Space VQ (CSVQ) design method for text Independent speaker recognition is generalized from Semi-noniterative VQ design method for text dependent speaker recognition. CSVQ contrasts with the existing desiEn method which uses the iterative learninE algorithm for every traininE speaker. The characteristics of a CSVQ design is as follows. First, the proposed method performs the non-iterative learning by using a Classified Space Codebook. Second, a quantization region of each speaker is equivalent for the quantization region of a Classified Space Codebook. And the quantization point of each speaker is the optimal point for the statistical distribution of each speaker in a quantization region of a Classified Space Codebook. Third, Classified Space Codebook (CSC) is constructed through Sample Vector Formation Method (CSVQ1, 2) and Hyper-Lattice Formation Method (CSVQ 3). In the numerical experiment, we use the 12th met-cepstrum feature vectors of 10 speakers and compare it with the existing method, changing the codebook size from 16 to 128 for each Classified Space Codebook. The recognition rate of the proposed method is 100% for CSVQ1, 2. It is equal to the recognition rate of the existing method. Therefore the proposed CSVQ design method is, reducing computational complexity and maintaining the recognition rate, new alternative proposal and CSVQ with CSC can be applied to a general purpose recognition.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.419-422
• /
• 2024

본 논문에서는 다항식 보간법의 일종인 이동최소자승법(Moving least squares, MLS)을 네트워크로 학습하여, Divergence-constrained MLS 벡터장을 효율적으로 표현하는 방법을 제안한다. 벡터장을 구성하기 위해 MLS는 스칼라가 아닌 벡터 보간을 해야 하므로 행렬과 벡터의 크기가 더 커지며, 이는 계산량이 커짐을 나타낸다. 고차 보간(High-order interpolation)이 가능한 특징은 장점이 되지만, 계산량이 매우 크기 때문에 시뮬레이션에는 활용이 어렵다. Divergence-constrained MLS를 유체 시뮬레이션에 적용한 경우가 있지만, 실제로 슈퍼컴퓨터(Supercomputer)를 해야 장면 제작이 가능하므로 효용성이 떨어진다. 본 논문에서는 이러한 문제를 해결하기 위해 네트워크 학습을 통한 Divergence-constrained MLS 벡터장을 표현할 수 있는 결과를 보여준다.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.7 no.3
• /
• pp.39-45
• /
• 2003

Modified Lanczos vector superposition method is proposed for efficient dynamic analysis of structures, The proposed method is based on the modified Lanczos algorithm that generates stiffness-orthonormal Lanczos vectors. The proposed Lanczos vector superposition method has the same accuracy and efficiency as the conventional Lonczos vector superposition method in the analysis of structures under single input loads. On the other hand, the proposed method is more efficient than the conventional method in the analysis of structures under multi-input loads. The effectiveness of the proposed method is verified by analyzing two numerical examples.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.5
• /
• pp.77-85
• /
• 2004

This paper proposes a multi-level vector error diffusion method using 64 primary colors to improve color impulse artifact in bright region. Vector error diffusion method causes color impulse artifact in bright region because we only use the Euclidean distance measure in quantization process. In order to reduce this artifact, the proposed method divides input color into chromatic color and achromatic color according to chroma value. In the case of chromatic color, input color is classified into bright region, middle bright region, and dark region according to lightness value. N candidate primary color is organized using lightness difference between input vector and 60 chromatic primary color vector in the case of bright region. Then, primary color with minimum vector norm between input vector and N candidate primary color in addition to 4 achromatic primary colors is selected as output color. As a result of experiments, the proposed method showed visually pleasing halftone output.

• The KIPS Transactions:PartB
• /
• v.8B no.4
• /
• pp.343-350
• /
• 2001

백터 양자화기 설계는 다차원의 목적함수를 최소화하는 학습 알고리즘을 필요로 한다. 일반화된 Lloyd 방법(GLA)은 벡터 양자화기 설계를 위해 오늘날 가장 널리 사용되는 알고리즘이다. GLA 는 일괄처리(batch) 방식으로 코드북을 생성하며 목적함수를 단조 감소시키는 강하법(descent algorithm)의 일종이다. 한편 Kohonen 학습법(KLA)은 학습벡터가 입력되는 동안 코드북이 갱신되는 온라인 벡터 양자화기 설계 알고리즘 이다. KLA는 원래 신경망 학습을 위해 Kohonen에 의해 제안되었다. KLA 역시 GLA와 마찬가지로 강하법의 일종이라 할 수 있다. 따라서 이들 두 알고리즘은, 비록 사용하기 편리하고 안정적으로 동작을 하지만, 극소(local minimum) 점으로 수렴하는 문제를 안고 있다. 우리는 이 문제와 관련하여 simulated annealing(SA) 방법의 응용을 논하고자 한다. SA는 현재까지 극소에 빠지지 않고 최소(global minimum)로 수렴하면서, 해의 수렴이 (통계적으로) 보장되는 유일한 방법이라 할 수 있다. 우리는 먼저 GLA에 SA를 응용한 그 동안의 연구를 개괄한다. 다음으로 온라인 방식의 벡터 양자화가 설계에 SA 방법을 응용함으로써 SA 방법에 기초한 새로운 온라인 학습 알고리즘을 제안한다. 우리는 이 알고리즘을 OLVQ-SA 알고리즘이라 부르기로 한다. 가우스-마코프 소스와 음성데이터에 대한 벡터양자화 실험 결과 제안된 방법이 KLA 보다 일관되게 우수한 코드북을 생성함을 보인다.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.3
• /
• pp.111-120
• /
• 2004

This paper proposes a vector error diffusion method for smear artifact reduction in the boundary region. This artifact mainly results from a large accumulation of quantization errors. In particular, color bands with a smear artifact, the width of a few pixels appear along the edges. Accordingly, to reduce this artifact, the proposed halftoning process excludes the large accumulated Quantization error by comparing the vector norms and vector angles between the error-corrected vector and eight primary color patches. When the vector norm of the error corrected vector is larger than those of eight primary color patches, the quantization error vector is excluded from the quantization error distribution process. In addition, the quantization error is also excluded when the angle between eight primary color patches and error corrected vector is large. As a result, the proposed method enables a visually pleasing halftone pattern to be generated by all three color separations into account in a device- independent color space and reduces smear artifact in the boundary regions.

• Proceedings of the KIEE Conference
• /
• 1997.07f
• /
• pp.2224-2226
• /
• 1997

공간벡터 펄스폭변조법과 관련하여 과변조시의 전류제어성능을 향상시킬 수 있는 벡터 과변조의 새로운 연산 알고리즘으로 유효전압벡터를 반지름으로 하는 원을 이용하여 수정된 기준전압벡터가 위치할 섹터를 구하고, 전압제한 6각형을 나타내는 전압벡터식을 이용하여 동일한 수식으로 벡터과변조하는 간결한 새로운 알고리즘을 제시하고, 이를 전류제어 시뮬레이션에 적용하였다.

• The KIPS Transactions:PartB
• /
• v.10B no.1
• /
• pp.67-72
• /
• 2003

In this paper, we study the enhancement of VQ (Vector Quantization) design for text dependent speaker recognition. In a concrete way, we present the non-Iterative method which makes a vector quantization codebook and this method Is nut Iterative learning so that the computational complexity is epochally reduced. The proposed semi-noniterative VQ design method contrasts with the existing design method which uses the iterative learning algorithm for every training speaker. The characteristics of a semi-noniterative VQ design is as follows. First, the proposed method performs the iterative learning only for the reference speaker, but the existing method performs the iterative learning for every speaker. Second, the quantization region of the non-reference speaker is equivalent for a quantization region of the reference speaker. And the quantization point of the non-reference speaker is the optimal point for the statistical distribution of the non-reference speaker In the numerical experiment, we use the 12th met-cepstrum feature vectors of 20 speakers and compare it with the existing method, changing the codebook size from 2 to 32. The recognition rate of the proposed method is 100% for suitable codebook size and adequate training data. It is equal to the recognition rate of the existing method. Therefore the proposed semi-noniterative VQ design method is, reducing computational complexity and maintaining the recognition rate, new alternative proposal.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2004.10a
• /
• pp.109-112
• /
• 2004

Of late, the thrust vectoring control, using fluidic co-flow and counter-flow concepts, has been received much attention since it not only improves the maneuverability of propulsive engine but also reduces an additional material load due to the trailing control wings, which in turn reduce the aerodynamic drag. However, the control effects are not understood well since the flow field involves very complicated non: physics such as shock wave/boundary layer interaction, separation and significant unsteadiness. Existing data are not enough to achieve the effectiveness and usefulness of the thrust vectoring control, and systematic work is required for the purpose of practical applications In the present study, computational study has been performed to investigate the effects of the thrust vector control using the fluidic co-and counter-flow concepts. The results obtained show that, for a given pressure ratio, the thrust deflection angle has a maximum value at a certain suction flow rate, which is at less than $5\%$ of the mass flow rate of the primary jet. With a longer collar, the same vector angle is achievable with smaller mass flow rate.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.3
• /
• pp.535-544
• /
• 2023

In this paper, we study the effect of gradient vector calculation method(analytical method, central finite difference method) on adaptive beamforming to control weight distribution during iterated calculation when LMS algorithm (repeating method) is used to realize desired beam pattern. To this end, a quasi-ideal beam having an arbitrarily set beam width, a rotating beam, and a multi-beam were reviewed as examples. Numerical experiments applied the step parameters of the appropriate values to the adaptive beamforming system through trial and error equally to the two calculations, and compared the convergence characteristics of objective functions that evaluate adaptability and error using two methods for calculating gradient vectors.