• 음성과학
• /
• 제5권1호
• /
• pp.7-21
• /
• 1999

This paper proposes an improved training procedure in speech recognition based on the continuous density of the Hidden Markov Model (CDHMM). Of the three parameters (initial state distribution probability, state transition probability, output probability density function (p.d.f.) of state) governing the CDHMM model, we focus on the third parameter and propose an efficient algorithm that determines the p.d.f. of each state. It is known that the resulting CDHMM model converges to a local maximum point of parameter estimation via the iterative Expectation Maximization procedure. Specifically, we propose two independent algorithms that can be embedded in the segmental K -means training procedure by replacing relevant key steps; the adaptation of the number of mixture Gaussian p.d.f. and the initialization using the CDHMM parameters previously estimated. The proposed adaptation algorithm searches for the optimal number of mixture Gaussian humps to ensure that the p.d.f. is consistently re-estimated, enabling the model to converge toward the global maximum point. By applying an appropriate threshold value, which measures the amount of collective changes of weighted variances, the optimized number of mixture Gaussian branch is determined. The initialization algorithm essentially exploits the CDHMM parameters previously estimated and uses them as the basis for the current initial segmentation subroutine. It captures the trend of previous training history whereas the uniform segmentation decimates it. The recognition performance of the proposed adaptation procedures along with the suggested initialization is verified to be always better than that of existing training procedure using fixed number of mixture Gaussian p.d.f.

• 응용통계연구
• /
• 제25권5호
• /
• pp.757-773
• /
• 2012

Traditional value at risk(S-VaR) has a difficulity in predicting the future risk of financial asset prices since S-VaR is a backward looking measure based on the historical data of the underlying asset prices. In order to resolve the deficiency of S-VaR, an economic value at risk(E-VaR) using the risk neutral probability distributions is suggested since E-VaR is a forward looking measure based on the option price data. In this study E-VaR is estimated by assuming the generalized gamma distribution(GGD) as risk neutral density function which is implied in the option. The estimated E-VaR with GGD was compared with E-VaR estimates under the Black-Scholes model, two-lognormal mixture distribution, generalized extreme value distribution and S-VaR estimates under the normal distribution and GARCH(1, 1) model, respectively. The option market data of the KOSPI 200 index are used in order to compare the performances of the above VaR estimates. The results of the empirical analysis show that GGD seems to have a tendency to estimate VaR conservatively; however, GGD is superior to other models in the overall sense.

• Communications for Statistical Applications and Methods
• /
• 제25권2호
• /
• pp.173-184
• /
• 2018

In medical or public health research, it is common to encounter clustered or longitudinal count data that exhibit excess zeros. For example, health care utilization data often have a multi-modal distribution with excess zeroes as well as a multilevel structure where patients are nested within physicians and hospitals. To analyze this type of data, zero-inflated count models with mixed effects have been developed where a count response variable is assumed to be distributed as a mixture of a Poisson or negative binomial and a distribution with a point mass of zeros that include random effects. However, no study has considered a situation where data are also censored due to the finite nature of the observation period or follow-up. In this paper, we present a weighted version of zero-inflated Poisson model with random effects accounting for variable individual follow-up times. We suggested two different types of weight function. The performance of the proposed model is evaluated and compared to a standard zero-inflated mixed model through simulation studies. This approach is then applied to Medicaid data analysis.

• 음성과학
• /
• 제10권4호
• /
• pp.137-147
• /
• 2003

Speech synthesis can classify by synthesis way, that is waveform coding, source coding and mixture coding. Specially, waveform coding is suitable for high quality synthesis. However, it is not desirable by synthesis techniques of syllable or phoneme unit because it do not separate and handles excitation and formant part. Therefore, there is a need for pitch alteration method applied in synthesis by the rule in waveform coding. This study propose about pitch alteration method that use spectrum scaling after do to flatten spectra by subband linear approximation to minimize spectrum distortion. This paper show evaluation whether show excellency of some measure compared with LPC, Cepstrum, lifter function and method that propose. estimation method seeks distribution of each flattened signal and measured degree of flattened spectra Signal flattened is normalized, So that highest point amounts to zero, and distribution of signal ,whose average is zero, is calculated. this show result that measure the spectrum distortion rate to estimate performance of method that propose. The average spectrum distortion rate was kept below the average 2.12%, so the method that propose is superiors than existent method.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2005년도 춘계학술대회 논문집
• /
• pp.252-257
• /
• 2005

Injection molding is a very important industrial process for the manufacturing of plastics objects. During an injection molding process of composites, the fiber-matrix separation and fiber orientation are caused by the flow of molten polymer/fiber mixture. As a result, the product tends to be nonhomogeneous and anisotropic. Hence, it is very important to clarify the relations between separation' orientation and injection molding conditions. So far, there is no research on the measurement of fiber orientation using image processing. In this study, the effects of fiber content ratio and molding condition on the fiber orientation-angle distributions are studied experimentally. Using the image processing method, the fiber orientation distribution of weld-line in injection-molded products is assessed. And the effects of fiber content and injection mold-gate conditions on the fiber orientation are also discussed.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1998년도 하계학술대회 논문집 E
• /
• pp.1752-1754
• /
• 1998

Recently the research about electron transport characteristic and energy distribute function in mixture gases within Helium, has been used and developed widely as industrial quality improvement of extinguish characteristic, electrical dielectric strength ability of application of each species high voltage apparatus, gas plasma etching progress of work to use manufacture of semiconductor, thin film molding by CVD, insulation film to use ultra LSI, etc. This paper analyze electron transport characteristic in the range E/N $1{\sim}60$[Td], pressure $0.1{\sim}6.0$[Torr] by MCS. It is necessary to seek electron drift velocity, diffusion coefficient, lonization coefficients, characteristic energy, mean energy and electron energy distribution function as electron transport characteristic.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 추계학술대회 논문집 전기물성,응용부문
• /
• pp.98-100
• /
• 2004

In this study, PST thin films were etched with inductively coupled $Cl_2/(Cl_2+Ar)$ plasmas. The etch characteristics of PST thin films as a function of $Cl_2/(Cl_2+Ar)$ gasmixtures were analyzed by using quadrupole mass spectrometer (QMS). Systematic studies were carried out as a function of the etching parameters, including the RF power and the working pressure. The maximum PST film etch rate is 56.2 nm/min, because a small addition of $Cl_2$ to the $Cl_2$/Ar mixture increased the chemical effect. It was proposed that sputter etching is the dominant etching mechanism while the contribution of chemical reaction is relatively low due to low volatility of etching products.

• 한국지반환경공학회 논문집
• /
• 제15권4호
• /
• pp.13-27
• /
• 2014

Catchments soil erosion, one of the most serious problems in the mountainous environment of the world, consists of a complex phenomenon involving the detachment of individual soil particles from the soil mass and their transport, storage and overland flow of rainfall, and infiltration. Sediment size distribution during erosion processes appear to depend on many factors such as rainfall characteristics, vegetation cover, hydraulic flow, soil properties and slope. This study involved laboratory flume experiments carried out under simulated rainfall in a 3.0 m long ${\times}$ 0.8 m wide ${\times}$ 0.7 m deep flume, set at $17^{\circ}$ slope. Five experimental cases, consisting of twelve experiments using three different sediments with two different rainfall conditions, are reported. The experiments consisted of detailed observations of particle size distribution of the out-flow sediment. Sediment water mixture out-flow hydrograph and sediment mass out-flow rate over time, moisture profiles at different points within the soil domain, and seepage outflow were also reported. Moisture profiles, seepage outflow, and movement of overland flow were clearly found to be controlled by water retention function and hydraulic function of the soil. The difference of grain size distribution of original soil bed and the out-flow sediment was found to be insignificant in the cases of uniform sediment used experiments. However, in the cases of non-uniform sediment used experiments the outflow sediment was found to be coarser than the original soil domain. The results indicated that the sediment transport mechanism is the combination of particle segregation, suspension/saltation and rolling along the travel distance.

• Steel and Composite Structures
• /
• 제22권3호
• /
• pp.691-712
• /
• 2016

In this editorial, buckling analytical investigation of the nanocomposite plate with square cut out reinforced by carbon nanotubes (CNTs) surrounded by Pasternak foundation is considered. The plate is presumed has square cut out in center and resting on Pasternak foundation. CNTs are used as amplifier in plate for diverse distribution, such as uniform distribution (UD) and three patterns of functionally graded (FG) distribution types of CNTs (FG-X, FG-A and FG-O). Moreover, the effective mechanical properties of nanocomposite plate are calculated from the rule of mixture. Domain decomposition method and orthogonal polynomials are applied in order to define the shape function of nanocomposite plate with square cut out. Finally, Rayleigh-Ritz energy method is used to obtain critical buckling load of system. A detailed parametric study is conducted to explicit the effects of the dimensions of plate, length of square cut out, different distribution of CNTs, elastic medium and volume fraction of CNTs. It is found from results that increase the dimensions of plate and length of square cut out have negative impact on buckling behavior of system but considering CNTs in plate has positive influence.

• 응용통계연구
• /
• 제30권1호
• /
• pp.57-68
• /
• 2017

대량의 데이터에 있어 전반적인 특성 및 구조를 파악하는데 유용하기 때문에 다양한 분야에서 군집분석을 사용하고 있다. Dempster 등 (1977)에서 정의된 expectation-maximization(EM) 알고리즘은 가장 보편적으로 사용되는 군집분석 방법이다. 선형모형의 유한혼합물(finite mixture of linear model) 기법 또한 군집분석 방법 중 많이 사용되는 방법이며 베이지안 군집방법은 Bernardo와 Giron (1988)이 군집에 대한 가중치 확률만 모를 경우 처음 적용하였다. 우리는 이 연구에서 일반적인 선형모형의 유한혼합물이 아닌 군집특정(cluster-specific) 변량효과를 모형에 포함하여 베이지안 분석방법인 깁스표집법(Gibbs sampling)을 사용한다. 제안한 모형의 특성 및 표집법에 대하여 설명하였고 모의실험 및 실제 데이터 분석을 통하여 모형의 유용성을 파악하였다. Hurn 등 (2003)의 CO2 데이터에 모형을 적용하여 변량효과가 없는 모형, 개체특정(subject-specific) 변량효과 모형과 비교하였다.