• The Korean Journal of Applied Statistics
• /
• v.22 no.6
• /
• pp.1203-1214
• /
• 2009

In this paper we introduce the time series clustering methods in the time and frequency domains and discuss the merits or demerits of each method. We analyze 15 daily stock prices of KOSPI 200, and the nonparametric method using the wavelet shows the best clustering results. For the clustering of nonstationary time series using the spectral density, the EMD method remove the trend more effectively than the differencing.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.437-438
• /
• 2018

문서 군집화 성능을 높이기 위한 한 방법으로 차원 축소를 적용한 문서 벡터로 군집화를 실시하는 방법이 있다. 본 발표에서는 특이값 분해(SVD), 커널 주성분 분석(Kernel PCA), Doc2Vec 등의 차원 축소 기법을, K-평균 군집화(K-means clustering), 계층적 병합 군집화(hierarchical agglomerative clustering), 스펙트럼 군집화(spectral clustering)에 적용하고, 그 성능을 비교해 본다.

• Journal of Information Processing Systems
• /
• v.14 no.3
• /
• pp.727-739
• /
• 2018

This study evaluates the viewpoints of user focus incidents using microblog sentiment analysis, which has been actively researched in academia. Most existing works have adopted traditional supervised machine learning methods to analyze emotions in microblogs; however, these approaches may not be suitable in Chinese due to linguistic differences. This paper proposes a new microblog sentiment analysis method that mines associated microblog emotions based on a popular microblog through user-building combined with spectral clustering to analyze microblog content. Experimental results for a public microblog benchmark corpus show that the proposed method can improve identification accuracy and save manually labeled time compared to existing methods.

• The KIPS Transactions:PartB
• /
• v.10B no.6
• /
• pp.665-672
• /
• 2003

This paper describes a speech synthesis technique by concatenating unit phoneme. At that time, a major problem is that discontinuity is happened from connection part between unit phonemes, especially from connection part between unit phonemes recorded by different persons. To solve the problem, this paper uses clustered diphone, and proposes a spectral smoothing technique, not only using formant trajectory and distribution characteristic of spectrum but also reflecting human's acoustic characteristic. That is, the proposed technique performs unit phoneme clustering using distribution characteristic of spectrum at connection part between unit phonemes and decides a quantity and a scope for the smoothing by considering human's acoustic characteristic at the connection part of unit phonemes, and then performs the spectral smoothing using weights calculated along a time axes at the border of two diphones. The proposed technique removes the discontinuity and minimizes the distortion which can be occurred by spectrum smoothing. For the purpose of the performance evaluation, we test on five hundred diphones which are extracted from twenty sentences recorded by five persons, and show the experimental results.

• Journal of Korea Multimedia Society
• /
• v.16 no.1
• /
• pp.19-28
• /
• 2013

In this research, we presented an effective clustering method based on ICA for the analysis of huge Raman hyperspectral dental data. The hyperspectral dataset captured by HR800 micro Raman spectrometer at UMKC-CRISP(University of Missouri-Kansas City Center for Research on Interfacial Structure and Properties), has 569 local points. Each point has 1,005 hyperspectal dentin data. We compared the clustering effectiveness and the clustering time for the case of using all dataset directly and the cases of using the scores after PCA and ICA. As the result of experiment, the cases of using the scores after PCA and ICA showed, not only more detailed internal dentin information in the aspect of medical analysis, but also about 7~19 times much shorter processing times for clustering. ICA based approach also presented better performance than that of PCA, in terms of the detailed internal information of dentin and the clustering time. Therefore, we could confirm the effectiveness of ICA for the analysis of Raman hyperspectral dental data.

• Proceedings of the KSRS Conference
• /
• 2007.03a
• /
• pp.130-133
• /
• 2007

본 연구는 원격 탐사의 영상 처리에서 영상 분할의 상위 수준으로 웅집 계층 clustering의 dendrogram을 통한 무감독 영상 분류를 제안한다. 제안된 알고리즘은 분광 영역에서 정의된 RAG(Regional Agency Graph)와 min-heap 자료 구조를 이용하여 MCSNP(Mutual Closest Spectral Neighbor Pair)의 집 합을 검색하면서 합병을 수행하는 계층 clustering 방법이다. 계산 시간과 저장 기억의 사용에 대한 효율을 증가시키기 위해 분광적 인접성올 정의 하는 분광 공간(spectral space)내의 다중창을 사용하였고 RNV(Region Neighbor Vector)을 이용하여 합병에 의하여 변하는 RAG 갱신하였고 적정한 단계 수가 주어 진다면 제안된 알고리즘은 집단 합병의 계층적 관계를 쉽게 해석 할 수 있는 dendrogram을 생성한다. 본 연구는 생성된 dendrogram을 이용한 nested-hierarchical 분석을 통하여 피복 형태의 계층적 관계를 해석한다. 이러한 해석은 피복 형태의 정확한 분류를 위한 의사 결정에 중요한 정보를 공급한다.

• Korean Journal of Remote Sensing
• /
• v.6 no.2
• /
• pp.115-133
• /
• 1990

Study was made on the spectral characteristics of rock samples including bentonites collected from the northern Ulsan area. The geology of the area consists mainly of sediments of the Kyongsang Series and Bulguksa granite, the Tertiary volcanics, andesites and tuffs. Relative reflectances of meshed samples(2.5~10mm) to BaSO$_4$ are measured at 6 Landsat TM spectral windows (excluding the thermal band) with HHRR, and their reflection charactristics were analysed. In addition, three different data selection schemes including the Eulidean distance, multiple regression, and PCA weight methods were applied to the 30 TM ratio channels, derived from the above 6 bands. The selected data sets were subject to two unsupervised classification techniques(FA and ISODATA) in order to compare the effectiveness for classification of particularly bentonite from others. As a result, in ISODATA analysis the multiple regression model shows the best, followed by the Euliean distances one. The PCA weight model seems to show some confusion. In FA, though difficult for quantitative analysis, the best still seems to be the regression model. Among ratio bands, rations of band 7 or 5 against other bands represent the best contribution in classification of bentonites from others.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.8 no.2
• /
• pp.44-54
• /
• 2005

In this study the tidal area in Seo han bay, North Korea was detected and extracted by using various satellite images (ASTER, KOMPSAT EOC, Landsat TM/ETM+) and GIS spatial analysis. Especially, the micro-landform was classified through the spectral characteristic of each satellite image and the change of tidal flat size was detected on passing year. For this, the spectral characteristics of eight tidal flat area in Korea, which are called as Seo han bay, Gwang ryang bay, Hae iu bay, Gang hwa bay, A san bay, Garorim bay, Jul po bay and Soon chun bay, were analyzed by using multi band of multi spectral satellite images such as Landsat TM/ETM+. Moreover, the micro-landform tidal flat in Seo han bay, North Korea was extracted by using ISODATA clustering based on the result of spectral characteristic. In addition, in order to detect the change of tidal flat size on passing years, the ancient topography map (1918-1920) was constructed as GIS DB. Also, the tidal flat distribution map based on the temporal satellite images were constructed to detect the tidal flat size for recent years. Through this, the efficient band to classify the micro-landform and detect its boundary was clarified and one possibility of KOMPSAT EOC application could be also introduced by extracting the spatial information of tidal flat efficiently.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.134-137
• /
• 2005

Unsupervised spectral angle classification (USAC) is the algorithm that can extract ground object information with the minimum 'Spectral Angle' operation on behalf of 'Spectral Euclidian Distance' in the clustering process. In this study, our algorithm uses the unit vector instead of the spectral distance to compute the mean of cluster in the unsupervised classification. The proposed algorithm (MUSAC) is applied to the Hyperion and ETM+ data and the results are compared with K-Meails and former USAC algorithm (FUSAC). USAC is capable of clearly classifying water and dark forest area and produces more accurate results than K-Means. Atmospheric correction for more accurate results was adapted on the Hyperion data (Hyperion-FLAASH) but the results did not have any effect on the accuracy. Thus we anticipate that the 'Spectral Angle' can be one of the most accurate classifiers of not only multispectral images but also hyperspectral images. Furthermore the cluster unit vector can be an efficient technique for determination of each cluster mean in the USAC.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.8
• /
• pp.3295-3311
• /
• 2020

With the wide application of hyperspectral images, it becomes more and more important to compress hyperspectral images. Conventional recursive least squares (CRLS) algorithm has great potentiality in lossless compression for hyperspectral images. The prediction accuracy of CRLS is closely related to the correlations between the reference bands and the current band, and the similarity between pixels in prediction context. According to this characteristic, we present an improved CRLS with adaptive band selection and adaptive predictor selection (CRLS-ABS-APS). Firstly, a spectral vector correlation coefficient-based k-means clustering algorithm is employed to generate clustering map. Afterwards, an adaptive band selection strategy based on inter-spectral correlation coefficient is adopted to select the reference bands for each band. Then, an adaptive predictor selection strategy based on clustering map is adopted to select the optimal CRLS predictor for each pixel. In addition, a double snake scan mode is used to further improve the similarity of prediction context, and a recursive average estimation method is used to accelerate the local average calculation. Finally, the prediction residuals are entropy encoded by arithmetic encoder. Experiments on the Airborne Visible Infrared Imaging Spectrometer (AVIRIS) 2006 data set show that the CRLS-ABS-APS achieves average bit rates of 3.28 bpp, 5.55 bpp and 2.39 bpp on the three subsets, respectively. The results indicate that the CRLS-ABS-APS effectively improves the compression effect with lower computation complexity, and outperforms to the current state-of-the-art methods.