• 음성과학
• /
• 제11권2호
• /
• pp.237-245
• /
• 2004

In general, Gaussian mixture model(GMM) is used to estimate the speaker model for speaker identification. The parameter estimates of the GMM are obtained by using the expectation-maximization (EM) algorithm for the maximum likelihood(ML) estimation. However, if the number of mixtures isn't defined well in the GMM, those parameters are obtained inappropriately. The problem to find the number of components is significant to estimate the optimal parameter in mixture model. In this paper, to estimate the optimal number of mixtures, we propose the method that starts from the sufficient mixtures, after, the number is reduced by investigating the mutual information between mixtures for GMM. In result, we can estimate the optimal number of mixtures. The effectiveness of the proposed method is shown by the experiment using artificial data. Also, we performed the speaker identification applying the proposed method comparing with other approaches.

• 약학회지
• /
• 제48권1호
• /
• pp.6-12
• /
• 2004

Near-infrared spectra of methanol-water mixtures and acetonitrile-water mixtures were acquired to find interactions between solvents widely used for reverse-phase liquid chromatography. Mixtures were prepared to give a series of increasing mole fractions of methanol or acetonitrile in water. Data matrices of acquired spectra were analyzed to determine the proper number of principal components of each mixture system using Malinowski's factor indicator function. Initial guess of score matrix and loading matrix were calculated by nonlinear iterative partial least squares (NIPALS) algorithm for faster computation. Iterative target transform factor analysis (ITTFA) was applied to convert the initial estimation of score matrix to true concentration profile and loading matrix to pure spectra of pure components of the mixtures. In case of methanol-water the number of principal components was found to be 4 and those initial guess of factors were converted to the pure spectra of water methanol and two kinds of complexes. In case of acetonitrile-water the number of pure components of the mixtures was found to be 3 and the pure spectrum of acetonitrile-water complex was found. The nonlinear characteristics of concentration profiles of complexes in the solvent mixtures may give a good criteria in understanding their elution characteristics in reverse-phase liquid chromatogrsphy.

• 한국도로학회논문집
• /
• 제18권4호
• /
• pp.55-61
• /
• 2016

PURPOSES : The objective of this study was to determine the relationship between the dielectric characteristics of asphalt mixtures and the air voids present in them using ground penetrating radar (GPR) testing. METHODS : To measure the dielectric properties of the asphalt mixtures, the reflection coefficient method and the approach based on the actual thickness of the asphalt layer were used. An air-couple-type GPR antenna with a center frequency of 1 GHz was used to measure the time for reflection from the asphalt/base layer interface. A piece of aluminum foil was placed at the interface to be able to determine the reflection time of the GPR signal with accuracy. An asphalt pavement testbed was constructed, and asphalt mixtures with different compaction numbers were tested. After the GPR tests, the asphalt samples were cored and their thicknesses and number of air voids were measured in the laboratory. RESULTS : It was found the dielectric constant of asphalt mixtures tends to decrease with an increase in the number of air voids. The dielectric constant values estimated from the reflection coefficient method exhibited a slight correlation to the number of air voids. However, the dielectric constant values measured using the approach based on the actual asphalt layer thickness were closely related to the asphalt mixture density. Based on these results, a regression equation to determine the number of air voids in asphalt mixtures using the GPR test method was proposed. CONCLUSIONS : It was concluded that the number of air voids in an asphalt mixture can be calculated based on the dielectric constant of the mixture as determined by GPR testing. It was also found that the number of air voids was exponentially related to the dielectric constant, with the coefficient of determination, $R^2$, being 0.74. These results suggest that the dielectric constant as determined by GPR testing can be used to improve the construction quality and maintenance of asphalt pavements.

• 한국도로학회논문집
• /
• 제10권4호
• /
• pp.19-29
• /
• 2008

최근에 $135^{\circ}C^{\circ}C$ 이하의 온도에서 생산되는 중온형 아스팔트 혼합물의 새로운 생산 기술이 전세계적으로 개발되고 있다. 본 연구에서는 가열 아스팔트보다 낮은 온도에서 아스팔트를 효과적으로 골재에 분산시켜 코팅할 수 있는 폼드 아스팔트 기술을 이용하여 중온형 아스팔트 혼합물을 생산하였으며, 최근 개발된 새로운 Simple Performance Testing 장비를 이용하여 다양한 온도와 하중조건 하에서 중온형 폼드 아스팔트 혼합물의 공용성 특성을 평가하였다. 중온형 폼드 아스팔트 혼합물은 FG 64-22의 아스팔트를 거품상태로 만들어 중온으로 가열된 골재에 뿌려서 제조하였으며, 중온형 아스팔트 혼합물은 중온의 골재에 FG 64-22의 아스팔트를 액상 상태로 첨가하여 제조하였다. 중온형 폼드 아스팔트 혼합물은 중온형 아스팔트 혼합물보다 높은 동탄성계수와 Flow Number를 나타내었다. 따라서, $100^{\circ}C$로 가열된 골재를 사용하여 생산된 중온형 폼드 아스팔트 혼합물은 중온형 아스팔트 혼합물에 비하여 피로균열 및 소성변형 저항에 우수한 것으로 평가되었다.

• Journal of the Korean Data and Information Science Society
• /
• 제19권3호
• /
• pp.751-759
• /
• 2008

Mixtures of factor analyzers(MFA) is useful to model the distribution of high-dimensional data on much lower dimensional space where the number of observations is very large relative to their dimension. Mixtures of common factor analyzers(MCFA) can reduce further the number of parameters in the specification of the component covariance matrices as the number of classes is not small. Moreover, the factor scores of MCFA can be displayed in low-dimensional space to distinguish the groups. We propose the factor scores of MCFA as new low-dimensional features for classification of high-dimensional data. Compared with the conventional dimension reduction methods such as principal component analysis(PCA) and canonical covariates(CV), the proposed factor score was shown to have higher correct classification rates for three real data sets when it was used in parametric and nonparametric classifiers.

• Bulletin of the Korean Chemical Society
• /
• 제27권5호
• /
• pp.718-724
• /
• 2006

The heterogeneous association behaviour of various concentration binary mixtures of mono alkyl ethers of ethylene glycol with ethyl alcohol were investigated by dielectric measurement in benzene solutions over the entire concentration range at 25 ${^{\circ}C}$. The values of static dielectric constant $\epsilon_0$ of the mixtures were measured at 1 MHz using a four terminal dielectric liquid test fixture and precision LCR meter. The high frequency limiting dielectric constant $\epsilon_\infty$ values were determined by measurement of refractive index $n_D$ ($\epsilon_\infty\;=\;n_D\;^2$). The measured values of $\epsilon_0$ and $\epsilon_\infty$ were used to evaluate the values of excess dielectric constant $\epsilon^E$, effective Kirkwood correlation factor $g^{eff}$ and corrective correlation factor $g_f$ of the binary polar mixtures to obtain qualitative and quantitative information about the H-bond complex formation. The non-linear behaviour of the observed $\epsilon_0$ values of the polar molecules and their mixtures in benzene solvent confirms the variation in the associated structures with change in polar mixture constituents concentration and also by dilution in non-polar solvents. Appearance of the maximum in $\epsilon^E$ values at different concentration of the polar mixtures suggest the formation of stable adduct complex, which depends on the molecular size of the mono alkyl ethers of ethylene glycol. Further, the observed $\epsilon^E$ < 0 also confirms the heterogeneous H-bond complex formation reduces the effective number of dipoles in these polar binary mixtures. In benzene solutions these polar molecules shows the maximum reduce in effective number of dipoles at 50 percent dilutions. But ethyl alcohol rich binary polar mixtures in benzene solvent show the maximum reduce in effective number of dipoles in benzene rich solutions.

• Geomechanics and Engineering
• /
• 제12권4호
• /
• pp.639-655
• /
• 2017

The binary mixture consists of two types of granular media with different physical attributes and sizes, which can be characterized by the percentage of large granules by weight (P) and the particle size ratio (${\alpha}$). Researchers determine that two thresholds ($P_S$ and $P_L$) exist for the peak shear strength of binary mixtures, i.e., at $P{\leq}P_S$, the peak shear strength is controlled by the small granules; at $P{\leq}P_L$, the peak shear strength is controlled by the large granules; at $P_S{\leq}P{\leq}P_L$, the peak shear strength is governed by both the large and small granules. However, the thresholds of binary mixtures with different ${\alpha}$ values, and the explanation related to the inner details of binary mixtures to account for why these thresholds exist, require further confirmation. This paper considers the mechanical behavior of binary mixtures with DEM analysis. The thresholds of binary mixtures are found to be strongly related to their coordination numbers $Z_L$ for all values of ${\alpha}$, where $Z_L$ denotes the partial coordination number only between the large particles. The arrangement structure of the large particles is examined when P approaches the thresholds, and a similar arrangement structure of large particles is formed in both 2D and 3D particle systems.

• 전자공학회논문지S
• /
• 제35S권3호
• /
• pp.166-175
• /
• 1998

When we make a speaker adapted model using MAPE (maximum a posteriori estimation), the adapted model has one mixture in each state. This is because we cannot estimate a number of a priori distribution from a speaker-independent model in each state. If the model is represented by one mixture in each state, it is not well adadpted to specific speaker because it is difficult to represent various speech informationof the speaker with one mixture. In this paper, we suggest the method using several mixtures to well represent various speech information of the speaker in each state. But, because speaker-specific training dat is not sufficient, this method can't be used in every state. So, we make the number of mixtures in each state variable in proportion to the number of frames and to the determinant ofthe variance matrix in the state. Using the proposed method, we reduced the error rate than methods using one branch in each state.

• Communications for Statistical Applications and Methods
• /
• 제18권3호
• /
• pp.357-365
• /
• 2011

This paper investigates a penalized likelihood method for estimating the parameter of normal mixtures in multivariate settings with full covariance matrices. The proposed model estimates the number of components through the addition of a penalty term to the usual likelihood function and the construction of a penalized likelihood function. We prove the consistency of the estimator and present the simulation results on the multi-dimensional nor-mal mixtures up to the 8-dimension.

• International Journal of Air-Conditioning and Refrigeration
• /
• 제16권4호
• /
• pp.117-123
• /
• 2008

The present study is directed at flow and heat transfer of $CO_2$ and oil mixtures in a circular tube. PAG and POE oils are considered in this study. Flow characteristics of $CO_2$ and oil mixtures have been investigated by flow visualization. Pressure drop has been measured in the range of operating mass flow rate from 0.1 to 0.4 kg/min in a circular tube. Heat transfer characteristics of $CO_2$/oil mixtures have been investigated using a counterflow heat exchanger. In case of pure liquid $CO_2$ as well as $CO_2$ and POE mixtures, flow are seen to be uniform so that $CO_2$ and POE oil are still miscible even at flowing state. However, it is found that $CO_2$ and PAG are not miscible. Pressure drop of $CO_2$/PAG mixtures are much higher than that of $CO_2$/POE mixtures as well as pure $CO_2$ at a fixed mass flow rate. As the concentration of POE oil is increased from 0 to 5 wt%, pressure drop is increased. However, heat transfer rate and heat transfer coefficient of $CO_2$/POE mixtures are much higher than that of $CO_2$/PAG mixtures. The f-factor correlation and Nusselt number correlation for $CO_2$/POE oil mixtures are suggested in this paper.