• 응용통계연구
• /
• 제33권2호
• /
• pp.115-122
• /
• 2020

K-평균 클러스터링은 매우 널리 사용되고 있으나 유사도가 구면체 또는 타원체로 정의되어 각 클러스터가 볼록 집합 형태인 자료에는 좋은 결과를 주지만 그렇지 않은 경우에는 매우 형편 없는 결과를 나타낸다. 스펙트럴 클러스터링은 K-평균 클러스터링의 단점을 잘 보완해 줄 뿐아니라 여러 형태의 자료나 고차원 자료 등에 대해서도 좋은 결과를 나타내서 최근 인공 신경망 모형에 많이 이용되고 있다. 하지만, 개선되어야 할 단점도 여전히 많다. 본 논문에서는 스펙트럴 클러스터링에 대해 알기 쉽게 소개하고, 클러스터 갯수의 추정, 척도모수의 추정, 고차원 자료의 차원 축소 등 스펙트럴 클러스터링에 대한 최근의 연구 동향을 소개한다.

• 천문학논총
• /
• 제9권1호
• /
• pp.69-84
• /
• 1994

The spectral types of stars can be classified by using Balmer discontinuity($D_B$) and wavelength(${\lambda}_B$) expressed in terms of effective temperatures appeared in Balmer discontinuity. In this research, in order to classify stars, we used the well established observational data of high dispersion spectrophotometry for the spectral types and luminosity classes of stars in the Breger(1976) catalogue. Balmer discontinuity by effective temperatures of stars was accurately measured, and the ${\lambda}_B$ was replaced to luminosity classes of MK system, because of the close relationship between the As and luminosity classes. We measured the energy gradients(${\phi}_R$) of stars which were expressed as a function of spectral types in the interval of ${\lambda}{\lambda}4,000{\sim}4600{\AA}$, and then obtained a new physical parameter(${\phi}$) from the $D_B$ and ${\phi}_B$. The new parameter, ${\phi}$ can be used instead of HD classifications of stars and can be used widely for spectrophotometrical classifications of stars.

• 물과 미래
• /
• 제10권1호
• /
• pp.101-117
• /
• 1977

The statatistical characteristics and spectra of sea waves at Mookho were analysed by several statistical methods. As the results, the following conclusions are obtained: 1. Values of surface elevation of sea wave are better fitted to Gram Charlier distribution than Gaussian distribution. This proves that sea waves have not only characters of irregularity but also non-linearity. 2. Distribution of maxima of surface elevation practically follows the distribution of Cartwright and Longuet-Higgins, also spectral width parameter is found to be increased with the increase of root mean square of surface elevation. 3. Sea wave may have spectrum of broad frequency band, however distributions of wave heights and periods follow the Rayleigh distribution which is derived from the assumption of narrow frequency band. 4. Ratios among mean wave heights from observed data show good agreements with theoretical values from Rayleigh distribution. 5. Spectral density and spectral width parameter increase with increase of wind velocity. And wave period at optimum band gas higher value than significant wave period by about 10 percent.

• Communications for Statistical Applications and Methods
• /
• 제2권2호
• /
• pp.231-242
• /
• 1995

In this paper we consider kernel type estimator of the spectral density at a point in the analysis of stationary time series data. The kernel entails choice of smoothing parameter called bandwidth. A data-based bandwidth choice is proposed, and it is obtained by solving an equation similar to Sheather(1986) which relates to the probability density estimation. A Monte Carlo study is done. It reveals that the spectral density estimates using the data-based bandwidths show comparatively good performance.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 2002년도 하계학술발표대회 논문집 제21권 1호
• /
• pp.25-28
• /
• 2002

Voice activity detection is important Problem in the speech recognition and communication. This paper introduces feature parameter which is reconstructed by the spectral entropy of information theory for the robust voice activity detection in the noise environment, analyzes and compares it with the energy method of voice activity detection and performance. In experiment, we confirmed that the spectral entropy is more feature parameter than the energy method for the robust voice activity detection in the various noise environment.

• 대한수학회지
• /
• 제54권4호
• /
• pp.1175-1187
• /
• 2017

The spectral problem $$-y^{{\prime}{\prime}}+q(x)y={\lambda}y,\;0 < x < 1, \atop y(0)cos{\beta}=y^{\prime}(0)sin{\beta},\;0{\leq}{\beta}<{\pi};\;{\frac{y^{\prime}(1)}{y(1)}}=h({\lambda})$$ is considered, where ${\lambda}$ is a spectral parameter, q(x) is real-valued continuous function on [0, 1] and $$h({\lambda})=a{\lambda}+b-\sum\limits_{k=1}^{N}{\frac{b_k}{{\lambda}-c_k}},$$ with the real coefficients and $a{\geq}0$, $b_k$ > 0, $c_1$ < $c_2$ < ${\cdots}$ < $c_N$, $N{\geq}0$. The sharpened asymptotic formulae for eigenvalues and eigenfunctions of above-mentioned spectral problem are obtained and the uniform convergence of the spectral expansions of the continuous functions in terms of eigenfunctions are presented.

• ETRI Journal
• /
• 제38권3호
• /
• pp.540-550
• /
• 2016

Spectral clustering is a powerful tool for exploratory data analysis. Many existing spectral clustering algorithms typically measure the similarity by using a Gaussian kernel function or an undirected k-nearest neighbor (kNN) graph, which cannot reveal the real clusters when the data are not well separated. In this paper, to improve the spectral clustering, we consider a robust similarity measure based on the shared nearest neighbors in a directed kNN graph. We propose two novel algorithms for spectral clustering: one based on the number of shared nearest neighbors, and one based on their closeness. The proposed algorithms are able to explore the underlying similarity relationships between data points, and are robust to datasets that are not well separated. Moreover, the proposed algorithms have only one parameter, k. We evaluated the proposed algorithms using synthetic and real-world datasets. The experimental results demonstrate that the proposed algorithms not only achieve a good level of performance, they also outperform the traditional spectral clustering algorithms.

• ETRI Journal
• /
• 제44권5호
• /
• pp.769-779
• /
• 2022

Spectral clustering has become a typical and efficient clustering method used in a variety of applications. The critical step of spectral clustering is the similarity measurement, which largely determines the performance of the spectral clustering method. In this paper, we propose a novel spectral clustering algorithm based on the local similarity measure of shared neighbors. This similarity measurement exploits the local density information between data points based on the weight of the shared neighbors in a directed k-nearest neighbor graph with only one parameter k, that is, the number of nearest neighbors. Numerical experiments on synthetic and real-world datasets demonstrate that our proposed algorithm outperforms other existing spectral clustering algorithms in terms of the clustering performance measured via the normalized mutual information, clustering accuracy, and F-measure. As an example, the proposed method can provide an improvement of 15.82% in the clustering performance for the Soybean dataset.

• Journal of information and communication convergence engineering
• /
• 제1권4호
• /
• pp.205-208
• /
• 2003

Voice activity detection is a important problem in the speech recognition and speech communication. This paper introduces new feature parameter which are reconstructed by spectral entropy of information theory for robust voice activity detection in the noise environment, then analyzes and compares it with energy method of voice activity detection and performance. In experiments, we confirmed that spectral entropy and its reconstructed parameter are superior than the energy method for robust voice activity detection in the various noise environment.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2006년도 하계종합학술대회
• /
• pp.243-244
• /
• 2006

In this study, we propose a speech/music discrimination method using spectral peak track analysis. The proposed method uses the spectral peak track's duration at the same frequency channel for feature parameter. And use the duration threshold to discriminate the speech/music. Experiment result, correct discrimination ratio varies according to threshold, but achieved a performance comparable to another method and has a computational efficient for discrimination.