• The Korean Journal of Applied Statistics
• /
• v.26 no.1
• /
• pp.141-149
• /
• 2013

We present methods for studying the log-density ratio that enables the selection of the predictors and the form to be included in the logistic regression model. Under bivariate normal distributional assumptions, we investigate the form of the log-density ratio as a function of two predictors. If two covariance matrices are equal, then the crossproduct and quadratic terms are not needed. If the variables are uncorrelated, we do not need the crossproduct terms, but we still need the linear and quadratic terms. We also explore other conditions in which the crossproduct and quadratic terms are not needed in the logistic regression model.

• The Korean Journal of Applied Statistics
• /
• v.22 no.5
• /
• pp.1047-1057
• /
• 2009

The nonparametric bivariate two-sample test of Bennett (1967) is extended to the multivariate k sample test. This test has been easily modified for a monotone trend among k samples. Often in applications it is important to consider a set of multivariate response variables simultaneously, rather than individually, and also important to consider testing k samples altogether. Different approaches of estimating the null covariance matrices of the test statistics resulted in the same limiting form. The multivariate k sample test is applied to the non-normal data of a randomized trial conducted for a period of four weeks in mental hospitals. The purpose of the trial is to compare the efficacy of three different interventions for a relief of the frequently occurring problems of constipation, caused as a side effect of antipsychotic drugs during hospitalization. The bowel movement status of patient for a week is summarized into a single severity score, and severity scores of four weeks comprise a four-dimensional multivariate variable. It is desirable with this trial data to consider a multivariate testing among k samples.

• Korean Journal of Remote Sensing
• /
• v.1 no.1
• /
• pp.5-27
• /
• 1985

A hill-sliding technique was devised to extract Gaussian clusters from the multivariate probability density estimates of sample data for the first step of iterative unsupervised classification. The underlying assumption in this approach was that each cluster possessed a unimodal normal distribution. The key idea was that a clustering function proposed could distinguish elements of a cluster under formation from the rest in the feature space. Initial clusters were extracted one by one according to the hill-sliding tactics. A dimensionless cluster compactness parameter was proposed as a universal measure of cluster goodness and used satisfactorily in test runs with Landsat multispectral scanner (MSS) data. The normalized divergence, defined by the cluster divergence divided by the entropy of the entire sample data, was utilized as a general separability measure between clusters. An overall clustering objective function was set forth in terms of cluster covariance matrices, from which the cluster compactness measure could be deduced. Minimal improvement of initial data partitioning was evaluated by this objective function in eliminating scattered sparse data points. The hill-sliding clustering technique developed herein has the potential applicability to decomposition of any multivariate mixture distribution into a number of unimodal distributions when an appropriate diatribution function to the data set is employed.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.43 no.5
• /
• pp.405-412
• /
• 2015

This paper represents a filter design to estimate the airspeed of a spin-stabilized, trajectory-correctible artillery ammunition. Due to the limited power and space in operational point of view, the airspeed sensor is not installed, and thus the airspeed need to be estimated using limited sensor measurements. The only IMU measurements(three-axis specific forces and angular rates) are used in this application. The extended Kalman filter algorithm is applied since a linear filter can not cover the its wide operational range in airspeed and altitude. In the implementation of the EKF, the state and measurement equations are transformed into the no-roll frame for simple form of Jacobian matrix. The simulation study is conducted to evaluate the performance of the filter under various environment conditions of sensor noise and wind turbulence. In addition, the effect of the choice in filter design parameters, i.e. process error covariance matrices is analyzed on the performance of the estimation of airspeed and angular rates.

• The Journal of the Korea Contents Association
• /
• v.9 no.10
• /
• pp.249-257
• /
• 2009

The purpose of this study is to find a classification method with high accuracy in regard with sasang constitutional diagnosis. The BMI, blood pressure, pulse wave, and Sasang constitution diagnosed by a specialist was collected from 2848 subjects who were apparently healthy. Through a selective procedure, the data of 1635 subjects was used in the analysis. The results with the classification methods such as the discriminant analysis, regression, decision tree and neural network were compared with the diagnosis of a Sasang constitutional specialist. In result, the discriminant analysis method was hard to qualify the assumption of the equality of covariance matrices within constitutional groups. Moreover, without BMI, the decision tree and neural network methods were very sensitive to the change of the analysis data. Therefore, the Logistic regression and the decision tree is recommended on condition that the decisive factors of constitution are well concerned.

• Journal of Advanced Navigation Technology
• /
• v.19 no.5
• /
• pp.389-394
• /
• 2015

This paper considers the transfer alignment in the inertial navigation system which has lever-arm and the time delay in the velocity measurement. We suggest a method to improve the performance of the velocity matching. First, we analyze the estimation performance of the velocity matching through the tuning of the two covariance matrices of process noise and measurement noise. Next we provide some maneuvering conditions of the vehicles to improve the estimation performance using the observability analysis. The analysis results are verified using the computer simulations, which show that cruise movements do not provide the azimuth estimation of the vehicles, while east or north accelerating movement can provide.

• The Journal of the Acoustical Society of Korea
• /
• v.30 no.6
• /
• pp.315-323
• /
• 2011

This paper proposes two methods of estimating prior distribution to improve the performance of rapid speaker adaptation based on maximum a posteriori linear regression (MAPLR). In general, prior distribution of the transformation matrix used in MAPLR adaptation is estimated from all of the training speakers who are employed to construct the speaker-independent model, and it is applied identically to all new speakers. In this paper, we propose a method in which prior distribution is estimated from a group of reference speakers, selected using adaptation data, so that the acoustic characteristics of the selected reference speakers may be similar to that of the new speaker. Additionally, in MAPLR adaptation with block-diagonal transformation matrix, we propose a method in which the mean matrix and covariance matrix of prior distribution are estimated from two groups of transformation matrices obtained from the same training speakers, respectively. To evaluate the performance of the proposed methods, we examine word accuracy according to the number of adaptation words in the isolated word recognition task. Experimental results show that, for very limited adaptation data, statistically significant performance improvement is obtained in comparison with the conventional MAPLR adaptation.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.1
• /
• pp.24-31
• /
• 1994

In this paper, we report on the development of a speaker independent continuous speech recognition system using continuous hidden Markov models. The continuous hidden Markov model consists of mean and covariance matrices and directly models speech signal parameters, therefore does not have quantization error. Filter bank coefficients with their 1st and 2nd-order derivatives are used as feature vectors to represent the dynamic features of speech signal. We use the segmental K-means algorithm as a training algorithm and triphone as a recognition unit to alleviate performance degradation due to coarticulation problems critical in continuous speech recognition. Also, we use the one-pass search algorithm that Is advantageous in speeding-up the recognition time. Experimental results show that the system attains the recognition accuracy of $83\%$ without grammar and $94\%$ with finite state networks in speaker-indepdent speech recognition.

• Nuclear Engineering and Technology
• /
• v.55 no.12
• /
• pp.4447-4464
• /
• 2023

This paper focuses on the development of a new computational model of the CNESTEN's TRIGA Mark II research reactor using the 3D continuous energy Monte-Carlo code TRIPOLI-4 (T4). This new model was developed to assess neutronic simulations and determine quantities of interest such as kinetic parameters of the reactor, control rods worth, power peaking factors and neutron flux distributions. This model is also a key tool used to accurately design new experiments in the TRIGA reactor, to analyze these experiments and to carry out sensitivity and uncertainty studies. The geometry and materials data, as part of the MCNP reference model, were used to build the T4 model. In this regard, the differences between the two models are mainly due to mathematical approaches of both codes. Indeed, the study presented in this article is divided into two parts: the first part deals with the development and the validation of the T4 model. The results obtained with the T4 model were compared to the existing MCNP reference model and to the experimental results from the Final Safety Analysis Report (FSAR). Different core configurations were investigated via simulations to test the computational model reliability in predicting the physical parameters of the reactor. As a fairly good agreement among the results was deduced, it seems reasonable to assume that the T4 model can accurately reproduce the MCNP calculated values. The second part of this study is devoted to the sensitivity and uncertainty (S/U) studies that were carried out to quantify the nuclear data uncertainty in the multiplication factor k_eff. For that purpose, the T4 model was used to calculate the sensitivity profiles of the k_eff to the nuclear data. The integrated-sensitivities were compared to the results obtained from the previous works that were carried out with MCNP and SCALE-6.2 simulation tools and differences of less than 5% were obtained for most of these quantities except for the C-graphite sensitivities. Moreover, the nuclear data uncertainties in the k_eff were derived using the COMAC-V2.1 covariance matrices library and the calculated sensitivities. The results have shown that the total nuclear data uncertainty in the k_eff is around 585 pcm using the COMAC-V2.1. This study also demonstrates that the contribution of zirconium isotopes to the nuclear data uncertainty in the k_eff is not negligible and should be taken into account when performing S/U analysis.

• Journal of Animal Science and Technology
• /
• v.51 no.5
• /
• pp.345-352
• /
• 2009

This study utilized 332,625 records of linear type scores consisting for 15 primary traits, 22,175 final score and 84,612 pedigree information of 22,175 Holstein cows from 1993 to 2007 in Korea to estimate genetic parameters for 16 type traits. Genetic and error (co)variances between two traits selected from 16 traits were estimated using bi-trait pairwise analyses with DFREML package. The estimated heritabilities for stature (ST), strength (STR), body depth (BD), dairy form (DF), rump angle (RA), thurl width (TW), rear legs side view (RLSV), foot angle (FA), fore udder attachment (FUA), rear udder height (RUH), rear udder width (RUW), udder cleft (UC), udder depth (UD), front teat placement (FTP), front teat length (FTL) and final score (FS) were 0.31, 0.21, 0.25, 0.10, 0.29, 0.19, 0.09, 0.06, 0.12, 0.13, 0.12, 0.08, 0.26, 0.20, 0.28 and 0.15, respectively. ST had the highest positive genetic correlation with BD (0.90), while RLSV had the highest negative genetic correlation with FA (-0.56). RA had negative genetic correlation with most udder traits (-0.17~-0.02). Especially, RUW had the higher positive genetic correlation with STR (0.60), BD (0.62), and TW (0.49), however, UD had the higher negative genetic correlation with STR (-0.40) and BD (-0.40). FTL had negative genetic correlation with FUA, RUH, RUW, UC and UD. FS had positive genetic correlation with UC, UD and FTP (0.12, 0.18 and 0.20). However, additional research is needed on the use of these parameters in the genetic evaluation because estimated genetic and error variance-covariance matrices were not positive definite.