• Communications for Statistical Applications and Methods
• /
• v.6 no.3
• /
• pp.929-938
• /
• 1999

We propose a criterion for testing homogeneity of diagonal covariance matrices of K multivariate normal populations. It is based on a factorization of usual likelihood ratio intended to propose and develop a criterion that makes use of properties of structures of the diagonal convariance matrices. The criterion then leads to a simple test as well as to an accurate asymptotic distribution of the test statistic via general result by Box (1949).

• Journal of the Korean Institute of Intelligent Systems
• /
• v.19 no.5
• /
• pp.648-654
• /
• 2009

In this paper, a method called bilateral diagonal 2DLDA is proposed for face recognition. Two methods called Dia2DPCA and Dia2DLDA were suggested to reserve the correlations between the variations in the rows and columns of diagonal images. However, these methods work in the row direction of these images. A row-directional projection matrix can be obtained by calculating the between-class and within-class covariance matrices making an allowance for the column variation of alternative diagonal face images. In addition, column-directional projection matrix can be obtained by calculating the between-class and within-class covariance matrices making an allowance for the row variation in diagonal images. A bilateral projection scheme was applied using left and right multiplying projection matrices. As a result, the dimension of the feature matrix and computation time can be reduced. Experiments carried out on an ORL face database show that the proposed method with three different distance measures, namely, Frobenius, Yang and AMD, is more accurate than some methods, such as 2DPCA, B2DPCA, 2DLDA, etc.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.4
• /
• pp.361-368
• /
• 2015

This paper reports an unscented Kalman filter approach for localization of a mobile robot in an indoor environment. The method proposes a new model of measurement uncertainty which adjusts the error covariance according to the measured distance. The method also uses non-zero off diagonal values in error covariance matrices of motion uncertainty and measurement uncertainty. The method is tested through experiments in an indoor environment of 100*40 m working space using a differential drive robot which uses Laser range finder as an exteroceptive sensor. The results compare the localization performance of the proposed method with the conventional method which doesn't use adaptive measurement uncertainty model. Also, the experiment verifies the improvement due to non-zero off diagonal elements in covariance matrices. This paper contributes to implementing and evaluating a practical UKF approach for mobile robot localization.

• Speech Sciences
• /
• v.10 no.4
• /
• pp.149-157
• /
• 2003

In this paper, we proposed the principal component analysis (PCA) fuzzy mixture model for speaker identification. A PCA fuzzy mixture model is derived from the combination of the PCA and the fuzzy version of mixture model with diagonal covariance matrices. In this method, the feature vectors are first transformed by each speaker's PCA transformation matrix to reduce the correlation among the elements. Then, the fuzzy mixture model for speaker is obtained from these transformed feature vectors with reduced dimensions. The orthogonal Gaussian Mixture Model (GMM) can be derived as a special case of PCA fuzzy mixture model. In our experiments, with having the number of mixtures equal, the proposed method requires less training time and less storage as well as shows better speaker identification rate compared to the conventional GMM. Also, the proposed one shows equal or better identification performance than the orthogonal GMM does.

• 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.