• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.32-35
• /
• 1999

기존의 반연속 HMM의 파라미터들 중에서 평균 벡터와 분산 행렬은 Maximum Likelihood Estimation 방법을 사용하여 학습한다. 본 논문에서는 평균 벡터를 위하여 Fuzzy c-means(FCM) 알고리즘을 사용하였고 분산 행렬을 위하여 FCM 알고리즘의 평균 벡터를 적용, 변형한 새로운 함수를 사용하여 화자적응에 적용하였다. 이러한 평균 벡터와 분산 행렬의 추정 방법은 새로운 화자에 대한 적응 능력을 갖는다. 제안한 방법을 적용한 한국어 격리 단어에 대한 컴퓨터 모의 실험결과 새로운 화자에 대해 적응함을 확인하였다.

• Annual Conference on Human and Language Technology
• /
• 1999.10e
• /
• pp.142-149
• /
• 1999

외국문명의 영향으로 많은 외래어가 한국어 문서 내에서 사용되고 있으며, 이러한 단어는 주로 전문용어, 고유명사, 신조어 등으로 사전에 등록되지 않는 것이 많다. 본 논문에서는 이중언어 코퍼스로부터 자동으로 외래어 사전을 추출해 내는 확률적 정렬 방법과 실험결과를 소개한다. 확률적 정렬 방법은 통계적 음차 표기 모델에서 사용된 방법을 변형하여 적용한 것이며, 문서단위로 정렬된 두 종류의 영-한 이중언어 코퍼스에 대해 실험하여 재현률과 정확률을 측정하였다 성능은 전처리단계인 한국어 미등록어 추정에 영향을 많이 받았는데, 미등록어 추정을 대략하였을 경우, 재현률은 평균 58%였고, 정확률은 평균74%이었으며, 수동으로 미등록어 명사를 분리했을 경우, 재현률 평균86%, 정확률 평균91%로 외래어와 대응되는 원어를 추출해 냈다.

• The KIPS Transactions:PartB
• /
• v.8B no.4
• /
• pp.389-394
• /
• 2001

본 논문에서는 독립 성분 해석(Independent Component Analysis, ICA) 기법과 인근 평균 및 정규화를 이용한 영상 분류 방법을 제안하였다. ICA에 잡음을 주어 영상을 분류하였을 때, 잡음에 대한 강인성을 증가시키기 위하여, 제안된 인근 평균 및 정규화를 전처리로 적용하였다. 제안된 방법은 전처리 없이 ICA에 주성분 해석(Principal Component Analysis, PCA)을 이용한 것에 비해 잡음에 대한 강인성을 증가시키는 것을 모의 실험을 통하여 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.05a
• /
• pp.1049-1052
• /
• 2013

저자들은 1980 년대 후반부터 역사학적 논쟁들을 통계적 수학적 방법을 이용하여 분석해보고자 하는 시도들을 하여 왔다. 최근에는 역사학의 논쟁점 중이 하나인 화랑세기의 진위 여부를 분석하기 위한 일련의 연구를 시도하였다. 이러한 일련의 시도의 일환으로 이 연구는 기획되었다. 이 글에서는 일본의 고대 기년 추정 방법을 참고하면서 세대 평균치와 재위 평균치의 추정이라는 방법을 통하여 "삼국사기" 초기 기년 문제에 대한 해명을 시도해 보고자 한다. 종래에는 세대 평균치라는 개념을 사용하지 않아서 편차가 심한 왕별 재위기간 평균치만 사용하여 재위기간 추정의 오차가 컸다. 또한 여러 가지 추론들이 주관적인 판단에 의지한 부분이 많아서 폭넓은 지지를 얻지 못한 경우가 많았다. 따라서 세대 평균치와 재위 평균치는 반드시 일치하지 않는다는 점을 분명히 하는 한편, 양자를 결합한 추정 방안을 마련해 볼 것이다.

• The Korean Journal of Applied Statistics
• /
• v.13 no.2
• /
• pp.447-455
• /
• 2000

We propose a weighted kernel pattern recognition method using the K -means clustering algorithm to reduce computation and storage required for the full kernel classifier. This technique finds a set of reference vectors and weights which are used to approximate the kernel classifier. Since the hierarchical clustering method implemented in the 'Weighted Parzen Window (WP\V) classifier is not able to rearrange the proper clusters, we adopt the K -means algorithm to find reference vectors and weights from the more properly rearranged clusters \Ve find that the proposed method outperforms the \VP\V method for the repre~entativeness of the reference vectors and the data reduction.

• The Korean Journal of Applied Statistics
• /
• v.17 no.1
• /
• pp.135-144
• /
• 2004

We propose a reproducibility (validity) assessment procedure of K-means cluster analysis by randomly partitioning the data set into three parts, of which two subsets are used for developing clustering rules and one subset for testing consistency of clustering rules. Also, as an alternative to Rand index and corrected Rand index, we propose an entropy-based consistency measure between two clustering rules, and apply it to determination of the number of clusters in K-means clustering.

• The Korean Journal of Applied Statistics
• /
• v.31 no.6
• /
• pp.707-719
• /
• 2018

Various time series representation methods have been proposed for efficient time series clustering and classification. Lin et al. (DMKD, 15, 107-144, 2007) proposed a symbolic aggregate approximation (SAX) method based on symbolic representations after approximating the original time series using piecewise local mean. The performance of SAX therefore depends heavily on how well the piecewise local averages approximate original time series features. SAX equally divides the entire series into an arbitrary number of segments; however, it is not sufficient to capture key features from complex, large-scale time series data. Therefore, this paper considers data-adaptive local constant approximation of the time series using the unbalanced Haar wavelet transformation. The proposed method is shown to outperforms SAX in many real-world data applications.

• The Korean Journal of Applied Statistics
• /
• v.30 no.2
• /
• pp.301-309
• /
• 2017

It is often possible to test for differences in population means when two or more samples are extracted from each N population. However, it is not possible to test for the mean difference if one sample is extracted from each population since a sample mean does not exist. But, by dividing a group of samples extracted one by one into two groups and generating a sample mean, we can identify a heterogeneity that may exist within the group by comparing the differences of the groups' mean. Therefore, we propose a minimum combination t-test method that can test the mean difference by the number of combinations that can be divided into two groups. In this paper, we proposed a method to test differences between means to check heterogeneity in a group of extracted samples. We verified the performance of the method by simulation study and obtained the results through real data analysis.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.6
• /
• pp.445-451
• /
• 2016

This paper introduces a postprocessing method, an iteration method for melody, and an average neural network method for learning a large number of songs in order to improve musically insufficient parts in automatic composition using existing artificial neural network. The melody of songs composed by artificial neural networks is produced according to the melodies of trained songs, so it can not be a specific tonality and it is difficult to have a repetitive composition. In order to solve these problems, we propose a postprocessing method that converts the melody composed by artificial neural networks into a melody having a specific tonality according to music theory and an iteration method for melody by iteratively composing measure divisions of artificial neural networks. In addition, the existing training method of many songs has some disadvantages. To solve this problem, we adopt an average neural network that is made by averaging the weights of artificial neural networks trained each song. From some experiments, it was confirmed that the proposed method solves the existing problems.

• Journal of Korea Water Resources Association
• /
• v.34 no.4
• /
• pp.317-326
• /
• 2001

This study evaluates the errors involved in the area average rainfall amounts estimated by the arithmetic mean method, the Thiessen's weighting method, and the optimal weighting method from the estimation theory. This study was applied to the upstream part of Nam-Han river basin (upper part of Youngwal) and the following results could be obtained. First, in case the raingauges are located evenly over the basin, no obvious difference can be found in the area average rainfall amounts from the arithmetic mean method or from the Thiessen's weighting method. However, as these two methods cannot consider the spatial variability of rainfall, the estimation error could be higher when the spatial variability of rainfall is high. In our application the estimation error from the arithmetic mean method or the Thiessen's weighting method was also found to be higher than that from the method from the information theory, which considers the spatial variability of rainfall. Thus, we could conclude that for the rainy season of Korea or for the mountain area when and where the spatial variability of rainfall is high, a proper method of considering the spatial variability of rainfall should be used regardless of the basin size. The isohyetal method generally used for the large basins or the optimal weighting method from the estimation theory used in this study could be good alternatives for this case.