• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.05a
• /
• pp.99-102
• /
• 2003

본 논문에서는 퍼지 클러스터링 알고리즘에 의해 구해진 퍼지 클러스터들에 대한 평가척도를 제안한다. 제안된 척도는 퍼 지 클러스터들간의 중첩성(overlapping)과 분리성 (separation)을 이용한다. 중첩성은 클러스터간 인접도를 이용하여 계산하며, 분리성은 데이터에 대한 상관성 정도를 나타낸다. 따라서 중첩성이 낮고 분리성이 높을수록 좋은 클러스터 결과라고 할 수 있다. 표준 데이터 집합을 대상으로 기존의 척도들과 비교실험 함으로써 제안된 척도의 신뢰성을 알아보았다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.4
• /
• pp.455-460
• /
• 2003

A new cluster validity index is proposed that determines the optimal partition and optimal number of clusters for fuzzy partitions obtained from the fuzzy c-means algorithm. The proposed validity index exploits an overlap measure and a separation measure between clusters. The overlap measure is obtained by computing an inter-cluster overlap. The separation measure is obtained by computing a distance between fuzzy clusters. A good fuzzy partition is expected to have a low degree of overlap and a larger separation distance. Testing of the proposed index and nine previously formulated indexes on well-known data sets showed the superior effectiveness and reliability of the proposed index in comparison to other indexes.

• Proceedings of the Korean Society for Information Management Conference
• /
• 2001.08a
• /
• pp.167-172
• /
• 2001

클러스터링 성능 평가를 위한 척도로 여러 공식이 개발되어 사용되어왔다. 이들 평가척도는 가급적 범위가 제한되며, 가환성이 있고, 비편향적이며 단일 척도일 필요가 있다. 기존 평가 척도에 대해서 검토한 후 비편향적인 단일 척도 WACS를 개발하였다. 클러스터 수를 달리하는 클러스터링 결과에 대해 여러 평가척도를 적용해서 성능을 평가하는 실험을 통해서 WACS 척도가 평가척도로서의 요건을 만족시킨다는 것을 확인하였다.

• The Journal of the Korea Contents Association
• /
• v.9 no.9
• /
• pp.80-89
• /
• 2009

A fast and efficient search algorithm to determine an optimal number of clusters in clustering algorithms is presented. The method is based on cluster validity index which is a measure for clustering optimality. As the clustering procedure progresses and reaches an optimal cluster configuration, the cluster validity index is expected to be minimized or maximized. In this Paper, a fast non-exhaustive search method for finding the optimal number of clusters is designed and shown to work well in clustering. The proposed algorithm is implemented with the k-mean++ algorithm as underlying clustering techniques using CB and PBM as a cluster validity index. Experimental results show that the proposed method provides the computation time efficiency without loss of accuracy on several artificial and real-life data sets.

• Proceedings of the KAIS Fall Conference
• /
• 2011.05a
• /
• pp.389-392
• /
• 2011

본 논문에서는 영상의 색 항등성을 달성하기 위해 본질 영상의 핵심인 불변 방향을 K-means 클러스터링을 이용해 검출하는 개선된 알고리즘을 제안한다. 우선, RGB 영상을 K-means 클러스터링 기법에 의해 다수의 클러스터로 분할한다. 이 때, 클러스터 간의 거리 측정은 유클리드 거리이다. 그리고 분할된 클러스터 중 가장 많은 색을 가진 클러스터만을 x-색도 공간으로 도시하여 해당되는 후보 불변 방향을 계산한다. 검출된 후보 불변 방향은 방향별로 프로젝션된 히스토그램에서 3개 이상의 프로젝션된 데이터를 가진 bin들의 개수가 가장 적은 방향이다. 그 후, 분할된 다른 여러 클러스터에 해당되는 후 보 불변 방향을 계산하여 가장 많은 빈도로 나타나는 방향을 영상의 최종 불변 방향으로 결정한다. 실험에서 Ebner에 의해 제안된 데이터집합을 실험 영상으로 사용하였고, 색항등성 측도를 평가 척도로 사용하였다. 실험 결과, 제안한 기법은 형광성 표면을 가진 형광 데이터집합에 보다 적합하였으며, 엔트로피 기법보다 색항등성이 1.5배 이상 높았다.

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 2004.06a
• /
• pp.67-72
• /
• 2004

않은 양의 유즈넷 뉴스 중에서 사용자가 찾고자 하는 정확한 정보를 빠른 시간 안에 검색하고, 원하는 정보만 필터링 하는 것은 중요하다. 그러나 뉴스 문서는 이메일과 달라서 미리 자신에게 맞는 뉴스그룹을 등록해 주어야만 정보를 얻을 수 있다. 본 연구에서는 다양한 뉴스그룹들 중에서 사용자의 취향과 유사한 뉴스그룹들을 코호넨 신경망을 이용하여 추천해주는 방법을 제시한다. 신경망을 학습시키기 위한 뉴스 문서의 키워드들을 선택하기 위해 예제 문서들로부터 후보 용어들을 추출하고 퍼지 추론을 적용하여 대표 용어들을 선택한다. 하지만 신경망의 학습 패턴을 관찰해 보면, 많은 부분이 비어있는 희소성 문제를 발견할 수 있다. 이에 본 연구에서는 통계적인 결정계수를 도입하여 불필요한 차원을 제거한 후 신경망을 학습시키는 새로운 방법을 제안한다. 제안된 방법은 모든 차원을 활용할 때 보다 클러스터내 거리와 클러스터간 거리의 척도를 이용한 클러스터 중첩도 면에서 우수한 분류 성능을 보여줌을 확인하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.9B
• /
• pp.1731-1741
• /
• 1999

Color images are often corrupted by the noise due to noisy sensors or channel transmission errors. Some filters such as vector media and vector $\alpha$-trimmed mean filter have bee used for color noise removal. In this paper, We propose the fuzzy cluster filters based on the possibilistic c-means clustering, because the possibilistic c-means clustering can get robust memberships in noisy environments. Also, we propose weighted vector $\alpha$-trimmed mean filter to improve the conventional vector $\alpha$-trimmed mean filter. In this filter, the central data are more weighted than the outlying data. In this paper, we implemented the color noise generator to evaluate the performance of the proposed filters in the color noise environments. The NCD measure and visual measure by human observer are used for evaluation the performance of the proposed filters. In the experiment, proposed fuzzy cluster filters in the sense of NCD measure gave the best performance over conventional filters in the mixed noise. Simulation results showed that proposed weighted vector $\alpha$-trimmed mean filters better than the conventional vector $\alpha$-trimmed mean filter in any kinds of noise.

• Journal of Environmental Impact Assessment
• /
• v.26 no.2
• /
• pp.160-170
• /
• 2017

The purpose of this study is to perform clustering of the habitat types and to identify the characteristics of species in the habitat types using mammal data (70,562) of the 3rd National Ecosystem Survey conducted from 2006 to 2012. The 15 habitat types recorded in the field-paper of the 3rd National ecosystem survey were reclassified, which was followed by the statistical analysis of mammal habitat types. In the habitat types cluster analysis, non-hierarchical cluster analysis (k-means cluster analysis), hierarchical cluster analysis, and non-metric multidimensional scaling method were applied to 14 habitat types recorded more than 30 times. A total of 7 Orders, 16 Families, and 39 Species of mammals were identified in the 3rd National Ecosystem Survey collected nationwide. When 11 clusters were classified by habitat types, the simple structure index was the highest (ssi = 0.07). As a result of the similarities and hierarchies between habitat types suggested by the hierarchical clustering analysis, the residential areas were the most different habitat types for mammals; the next following type was a cluster together with rivers and coasts. The results of the non-metric multidimensional scaling analysis demonstrated that both Mus musculus and Rattus norvegicus restrictively appeared in a residential area, which is the most discriminating habitat type. Lutra lutra restrictively appeared in coastal and river areas. In summary, according to our results, the mammalian habitat can be divided into the following four types: (1) the forest type (using forest as the main habitat and migration route); (2) the river type (using water as the main habitat); (3) the residence habitat (living near residential area); and (4) the lowland type (consuming grain or seeds as the main feeding resource).

• Journal of Digital Contents Society
• /
• v.12 no.4
• /
• pp.541-550
• /
• 2011

MotionTypography is crucial to effective emotional communication in digital society. In this paper, I study movements to represent emotion using motiontypography and approach two goals: First to define an emotional measure by means of emotion assesses by the public, Second to research image characteristics corresponding to movements. In this dissertation, we collect emotional words by literature and experimental surveys and extract representative emotional words using KJ method and clustering analysis. The results of research, the emotional axes selected for motiontypography represent 'calm to active' and 'soft to stiff', the viewers feel a specific emotional state from some movements of motiontypography. If, we investigate the relationship of motiontypographic visual elements with emotional words are achieved together, I think it will serve as a motiontypographic guideline that enables helping the public to easily produce motiontypography.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.12
• /
• pp.3718-3726
• /
• 1999

Color image is more effective than gray one in human visual perception. Therefore, color image processing becomes important area. Color images are often corrupted by noises due to the input sensor, channel transmission errors and so on. Some filtering techniques such as vector median, mean filter, and vector $\alpha-trimmed$ mean filter have been used for color noise removal. Among them, vector $\alpha-trimmed$ mean filter gave the best performance in the mixed color noise. But, there are edge shift and blurring effect because vector $\alpha-trimmed$ mean filter is uniformly processed across the image. So, we proposed a conditional fuzzy cluster filter to improve this problems. Simulation results showed that the proposed scheme improves the NCD measure and visual quality over the conventional vector $\alpha-trimmed$ mean filter in the mixed color noise.