• 한국지능시스템학회논문지
• /
• 제14권5호
• /
• pp.517-524
• /
• 2004

1960년대 퍼지 이론이 소개된 이후 데이터 마이닝을 포함한 기계 학습 분야의 군집화 작업에서 퍼지 이론이 폭넓게 사용되었다. 퍼지 C-평균 알고리즘은 가장 많이 사용되는 퍼지 군집화 알고리즘이다. 이 알고리즘은 하나의 데이터 개체가 서로 다른 소속 정도를 가지고 각 군집에 할당될 수 있도록 한다. 퍼지 C-평균 알고리즘도 K-평균 알고리즘과 같은 일반적인 군집화 알고리즘과 마찬가지로 초기 군집수와 군집 중심의 위치에 의해 최종 군집 결과의 성능 차이가 나타난다. 군집화를 위한 이러한 초기 설정은 주관적이며 이 때문에 적절치 못한 결과를 얻게 될 수도 있다. 본 논문에서는 이 문제를 해결할 수 있는 방법으로 주어진 학습 데이터의 속성을 기반으로 한 초기 군집수와 군집 중심을 결정하는 개선된 밀도 기반의 퍼지 C-평균 알고리즘을 제안하였다. 제안 방법은 격자를 사용하여 초기 군집 중심의 위치와 군집수를 결정하였다. 기존에 많이 이용되었던 객관적인 기계 학습 데이터를 이용하여 제안 알고리즘의 성능비교를 수행하였다.

• 한국경제지리학회지
• /
• 제17권4호
• /
• pp.688-701
• /
• 2014

본 연구는 부산신발산업의 집적과 쇠락요인을 산업클러스터의 경쟁력이라는 이론적 관점에서 접근한다. 이를 위해 M. Porter의 클러스터 모델을 재구성한 '산업클러스터의 구성요소와 연계 모형'을 제시하고 그것에 기초하여 부산신발산업의 집적화와 쇠락에 영향을 미친 요인들을 검토하였다. 그리고 신발산업의 부산지역 집적 및 쇠락과정과 현황을 고찰하고 부산지역 신발업체들을 대상으로 관련 설문조사를 실시하였다. 분석 결과, 부산신발산업의 집적 및 쇠락과정은 상당히 복잡한 요인들에 의해 영향을 받았지만, 노동집약적 경공업의 특성상 노동력 등 투입요소부문이 가장 결정적인 영향변수였던 것으로 나타났다. 그리고 1970년대 중반 이후 OEM 생산을 본격화하면서 글로벌 생산네트워크에 편입된 바, 그것에의 진입과 배제과정에서 산업클러스터의 성쇠가 보다 급격히 발생한 것으로 판단된다.

• 대한수학회논문집
• /
• 제20권4호
• /
• pp.813-826
• /
• 2005

Regarding all particles at a fixed site as a cluster, the size of the largest cluster under the zero range invariant measures is well studied by Jeon et al.[5] for the case of density one. Here, the density of the finite zero-range process is given by the ratio between the number m of particles and the number n of sites. In this paper, we study the lower density case, i.e., the case m = o(n). Especially, when m ~ $n^{\beta}$,0 < ${\beta}$ < 1, we show that there is an interesting cutoff point around $\beta$ = 1/2.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제17권4호
• /
• pp.267-278
• /
• 2013

We consider zero range processes with density dependent jump rates g given by $g=g(n,k)=g_1(n)g_2(k/n)$ with $g_1(x)=x^{-\alpha}$ and $$g_2(x)=\{^{x^{-\alpha}\;if\;a&lt;x}_{Mx^{-\alpha}\;if\;x{\leq}a}$$. (0.1) In this case, with 1/2 < a < 1 and ${\alpha}$ > 0, we show that non-complete condensation occurs with maximum cluster size an. More precisely, for any ${\epsilon}$ > 0, there exists $M^*$ > 0 such that, for any 0 < M ${\leq}M^*$, the maximum cluster size is between (a - ${\epsilon}$)n and (a + ${\epsilon}$)n for large n. This provides a simple example of non-complete condensation under perturbation of rates which are deep in the range of perfect condensation (e.g. ${\alpha}$ >> 1) and supports the instability of the condensation transition.

• Transactions on Electrical and Electronic Materials
• /
• 제4권1호
• /
• pp.1-6
• /
• 2003

We have studied the size confinement effect on the properties of melting-like transition of small icosahedral copper clusters using a classical molecular dynamics simulation based on a well fitted empirical potential. We investigated the caloric curves of icosahedron nanoclusters and the significant depression in the melting temperatures of the copper nanoclusters was compared with that of the bulk copper. A structural transitions from decahedral to icosahedral shapes were shown. As the cluster size increased, the melting temperature increased, and the latent heat increased but seem to be saturated. However, the specific heat was unrelated to the cluster size.

• Journal of Information Processing Systems
• /
• 제14권4호
• /
• pp.926-940
• /
• 2018

In clustering-based approaches, cluster heads closer to the sink are usually burdened with much more relay traffic and thus, tend to die early. To address this problem, distance-aware clustering approaches, such as energy-efficient unequal clustering (EEUC), that adjust the cluster size according to the distance between the sink and each cluster head have been proposed. However, the network lifetime of such approaches is highly dependent on the distribution of the sensor nodes, because, in randomly distributed sensor networks, the approaches do not guarantee that the cluster energy consumption will be proportional to the cluster size. To address this problem, we propose a novel approach called CACD (Clustering Algorithm Considering node Distribution), which is not only distance-aware but also node density-aware approach. In CACD, clusters are allowed to have limited member nodes, which are determined by the distance between the sink and the cluster head. Simulation results show that CACD is 20%-50% more energy-efficient than previous work under various operational conditions considering the network lifetime.

• 대한지역사회영양학회지
• /
• 제5권1호
• /
• pp.83-91
• /
• 2000

The aims of this study are to evaluate the quality of hospital food services and the evaluate the quality in selected hospitals trough the use of the questionnaires. A survey of 30 hospital food and nutrition service department was undertaken and detailed information was collected from each, including, surveys of 1, 016 patient. Statistical data analysis was completed using the SAS/win 6.11 package for descriptive analysis, t-test X$^2$-test ANOVA principal component analysis , and cluster analysis and cluster analysis. In the case of patient satisfaction with hospital food and food services, overall satisfaction scores of male and female were 3.54 and 3.45 showing higher levels than the average score(3.00) The aspect of the food and food service which received the lowest ratings by patients was 'meal rounding while dining'. After conduction of factor analysis of variables affecting the patients meal satisfaction 3 groups including the 'menu satisfaction factor', 'service satisfaction factor ' and 'nutrition management satisfaction factor ' were selected. 3 clusters were categorized by the 'service cluster' 'nutrition management cluster', 'men cluster', and 'menu nutrition service cluster' after conducting a cluster analysis with influencing variables affecting patients meal satisfaction. The overview results of patient satisfaction by cluster were : in the case of the service group, such factors as taste, portion size, dealing with complaints while dining meal rounding while dining should be managed with caution In case of the nutrition management group, such factors as taste, portion size, temperature of the food intake, and dependence on hospital food should be managed with care, In the case of the menu groups, such factors as punctuality of meal times, contaminated substances in meals and serving mistakes, cleanliness of dishes, kindness of the server meal rounding while dining should by particularly managed with importance.

• 한국임상수의학회지
• /
• 제28권4호
• /
• pp.422-426
• /
• 2011

For a national survey in which wide geographic region or an entire country is targeted, multi-stage sampling approach is widely used to overcome the problem of simple random sampling, to consider both herd- and animallevel factors associated with disease occurrence, and to adjust clustering effect of disease in the population in the calculation of sample size. The aim of this study was to establish sample size for estimating bovine tuberculosis (TB) in Korea using stratified two-stage sampling design. The sample size was determined by taking into account the possible clustering of TB-infected animals on individual herds to increase the reliability of survey results. In this study, the country was stratified into nine provinces (administrative unit) and herd, the primary sampling unit, was considered as a cluster. For all analyses, design effect of 2, between-cluster prevalence of 50% to yield maximum sample size, and mean herd size of 65 were assumed due to lack of information available. Using a two-stage sampling scheme, the number of cattle sampled per herd was 65 cattle, regardless of confidence level, prevalence, and mean herd size examined. Number of clusters to be sampled at a 95% level of confidence was estimated to be 296, 74, 33, 19, 12, and 9 for desired precision of 0.01, 0.02, 0.03, 0.04, 0.05, and 0.06, respectively. Therefore, the total sample size with a 95% confidence level was 172,872, 43,218, 19,224, 10,818, 6,930, and 4,806 for desired precision ranging from 0.01 to 0.06. The sample size was increased with desired precision and design effect. In a situation where the number of cattle sampled per herd is fixed ranging from 5 to 40 with a 5-head interval, total sample size with a 95% confidence level was estimated to be 6,480, 10,080, 13,770, 17,280, 20.925, 24,570, 28,350, and 31,680, respectively. The percent increase in total sample size resulting from the use of intra-cluster correlation coefficient of 0.3 was 22.2, 32.1, 36.3, 39.6, 41.9, 42.9, 42,2, and 44.3%, respectively in comparison to the use of coefficient of 0.2.

• 산업기술연구
• /
• 제21권B호
• /
• pp.149-153
• /
• 2001

A study on the particle growth in $TEOS/O_2$ plasma was performed, and particle size and its distribution was measured by the electrical aerosol analyzer (EAA), light scattering particle size analyzer and the particle size was also determined by SEM. The effects of process variables such as total gas flow rate, reactor pressure, supplied power and initial reactant concentration on the particle growth were investigated. From the EAA results, the particle size distribution is divided into three groups of the cluster size and the small and large size particles. The particle size distribution measured by the light scattering particle size analyzer becomes bimodal, because the cluster size particles smaller than 20 nm in diameter cannot be detected by the light scattering particle size analyzer. The size of particles measured by the light scattering particle size analyzer is in good agreements with those by the SEM. Also we could understand that the particle formation is very sensitive to the changes of reactor pressure and reactant concentration. As the total gas flow rate increases, the particle size decreases because of the shorter residence time. As the reactor pressure, or the reactant concentration increases, the particle concentration increases and the particles grow more quickly by the faster coagulation between particles.

• 천문학회지
• /
• 제37권5호
• /
• pp.323-328
• /
• 2004

In this paper we review the observational results on Relic radio sources in clusters of galaxies. We discuss their observational properties, structures and radio spectra. We will show that Relics can be divided according to their size, morphology, and location in the galaxy cluster. These differences could be related to physical properties of Relic sources. The comparison with cluster conditions suggests that Relics could be related to shock waves originated by cluster mergers.