• The Korean Journal of Applied Statistics
• /
• v.30 no.1
• /
• pp.83-93
• /
• 2017

A typical approach to visualizing k (${\geq}2$)-group multidimensional data is to use Fisher's canonical discriminant analysis (CDA). CDA finds the best low-dimensional subspace that accommodates k group centroids in the Mahalanobis space. This paper proposes an alternative visualization procedure functioning in the Euclidean space, which finds the primary dimension with maximum discrimination of k group centroids and the secondary dimension with maximum dispersion of all observational units. This hybrid procedure is especially useful when the number of groups k is two.

• The Korean Journal of Applied Statistics
• /
• v.30 no.4
• /
• pp.567-578
• /
• 2017

Grouped multivariate data can be tested for differences between two or more groups using multivariate analysis of variance (MANOVA). However, this method cannot be used if several assumptions of MANOVA are violated. In this case, multidimensional scaling (MDS) and analysis of distance (AOD) can be applied to grouped dissimilarities based on the various distances. A permutation test is a non-parametric method that can also be used to test differences between groups. MDS is used to calculate the coordinates of observations from dissimilarities and AOD is useful for finding group structure using the coordinates. In particular, AOD is mathematically associated with MANOVA if using the Euclidean distance when computing dissimilarities. In this paper, we study the between and within group structure by applying MDS and AOD to the grouped dissimilarities. In addition, we propose a new test statistic using the group structure for the permutation test. Finally, we investigate the relationship between AOD and MANOVA from dissimilarities based on the Euclidean distance.

• Journal of Korea Multimedia Society
• /
• v.15 no.2
• /
• pp.215-225
• /
• 2012

In multimedia database, a multidimensional space-based indexing has been used to increase search efficiency. However, this method is inefficient in terms of ubiquity because it uses Euclidean distance as a scale of distance calculation. On the contrary, a metric space-based indexing method, in which metric axiom is prerequisite is widely available because a metric scale other than Euclidean distance could be used. This paper is attempted to propose a way of improving VP-tree, one of the metric space indexing methods. The VP-tree calculates the distance with an object which is ultimately linked to the a leaf node depending on the node fit for the search range from a root node and examines if it is appropriate with the search range. Because search speed decreases as the number of distance calculations at the leaf node increases, however, this paper has proposed a method which uses the latest interface on query object as the base point of trigonometric inequality for improvement after focusing on the trigonometric inequality-based range compression method in a leaf node. This improvement method would be able to narrow the search range and reduce the number of distance calculations. According to a system performance test using 10,000 video data, the new method reduced search time for similar videos by 5-12%, compared to a conventional method.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.4
• /
• pp.427-431
• /
• 2010

Multidimensional scaling (MDS) is a widely used method for dimensionality reduction, of which purpose is to represent high-dimensional data in a low-dimensional space while preserving distances among objects as much as possible. MDS has mainly been applied to data visualization and feature selection. Among various MDS methods, the classical MDS is not readily applicable to data which has large numbers of objects, on normal desktop computers due to its computational complexity. More precisely, it needs to solve eigenpair problems on dissimilarity matrices based on Euclidean distance. Thus, running time and required memory of the classical MDS highly increase as n (the number of objects) grows up, restricting its use in large-scale domains. In this paper, we propose an efficient approximation algorithm for the classical MDS based on divide-and-conquer and CUDA. Through a set of experiments, we show that our approach is highly efficient and effective for analysis and visualization of data consisting of several thousands of objects.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.8
• /
• pp.68-77
• /
• 2000

Line special frequency (LSF) parameters have been widely used in low bit-rate speech coding due to their efficiency for representing the short-time speech spectrum. In this paper, a new distance measure based on the masking properties of human ear is proposed for quantizing LSF parameters whereas most conventional quantization methods are based on the weighted Euclidean distance measure. The proposed method derives the perceptual distance measure from the definition of noise-to-mask ratio (NMR) which has high correspondence with the actual distortion received in the human ear and uses it for quantizing LSF parameters. In addition, we propose an adaptive bit allocation scheme, which allocates minimal bits to LSF parameters maintaining the perceptual transparency of given speech frame for reducing the average bit-rates. For the performance evaluation, we has shown the ratio of perceptually transparent frames and the corresponding average bit-rates for the conventional and proposed methods. By jointly combining the proposed distance measure and adaptive bit allocation scheme, the proposed system requires only 770 bps for obtaining 95.5% perceptually transparent frames, while the conventional systems produce 89.9% at even 1800 bps.