클러스터 적응주기 모델에 대한 비판적 검토

전지혜 ( Jihye Jeon ),이철우 ( Chulwoo Lee ) 한국경제지리학회 2017 한국경제지리학회지 Vol.20 No.2

This study seeks to critically examine the significance and limits of the cluster adaptive cycle model for analysis of cluster evolution and to propose research issues for future analysis of cluster evo-lution based on this critical examination. Until the 1980s, research on industrial complexes including clusters was based on a `static perspective` that focuses on the aspect of economic space at a specific point in time, but the research paradigm has recently shifted to a `dynamic perspective` focusing on `evolution` of `complex adaptive systems`. As a result, the adaptive cycle model has attracted attention as an analysis tool of dynamically evolving clusters. However, the cluster adaptive cycle model has emerged by being ap-propriately modified and expanded according to the properties of the cluster and its evolution. The cluster adaptive cycle model is a comprehensive analysis framework that identifies the characteristics of cluster evolution in terms of resource accumulation, interdependence, and resilience and classifies cluster evolu-tion paths into six different categories. Nevertheless, there is still a need for further discussion and supple-mentation in terms of theoretical and empirical research to expand and deepen the model. Therefore, research issues for future analysis of cluster evolution are to specify and elaborate the cluster evolution model, to emphasize the concept of resilience, and to verify the applicability and usefulness of the model through empirical research.

A Novel Data Clustering Algorithm based on Modified Adaptive Particle Swarm Optimization

Ganglong Duan,Wenxiu Hu,Zhiguang Zhang 보안공학연구지원센터 2016 International Journal of Signal Processing, Image Vol.9 No.3

Fuzzy clustering is a popular unsupervised learning method used in cluster analysis which allows a point in large data sets belongs to two or more clusters. Prior work suggests that Particle Swarm Optimization based approach could be a powerful tool for solving clustering problems. In this paper, we propose a data clustering algorithm based on modified adaptive particle swarm optimization. We choose to use artificial bee colony algorithm combined with PSO technique to modify the traditional clustering methods due to its fast convergence and the presence of adaptive mechanisms based on the evolutionary factor. On the one hand, Particle Swarm Optimization is proven to be an effective and robust technique for fuzzy clustering. On the other hand, the artificial bee colony algorithm has the capability to generate diversity within the swarm when the guide bees are in the exploration mode. Through numerical analysis and experimental simulation, we verify that our algorithm performs much better compared with other state-of-the-art algorithms. Future research schedule is also discussed in the final part.

DEU-GRIT을 활용한 군집화된 대학 신입생의 대학 생활 적응 (학업, 사회, 정서 및 진로 적응) 특성 분석 연구: D 대학을 중심으로

이정희,강창완,박애영,고미나,강정임,신지영,최서현,이진현,함수민,김보성 한국감성과학회 2020 감성과학 Vol.23 No.3

This study is aimed at analyzing the characteristics of adaptation to college life among freshmen of D University by utilizing the DEU-GRIT test. To achieve this, we examined differences in level of college life adaptation according to gender and major, and explored the characteristics of college life adaptation among college freshmen who were clustered based on their level of academic, social, emotional, and career adaptation, which is an important factor in adaptation to college life. Moreover, based on the DEU-GRIT test standards, we also explored the characteristics of their college life adaptation by sorting out low level groups of freshmen with difficulties adapting. The results demonstrated that male students had a higher level of emotional adaptation than female students, and that science and technology majors had higher levels of academic and emotional adaptation than students in the humanities and society majors. Furthermore, a cluster analysis based on 4 factor levels of college life adaptation showed that cluster 1 had a high level of adaptability to all factors, but cluster 3 had a low level of adaptability to all factors. Moreover, while cluster 2 showed a high level of emotional adaptation, cluster 4 showed a low level of emotional adaptation. On the other hand, a lower-standard group of DEU-GRIT tests showed that the level of academic adaptation and social adaptation was higher among female students than among male students. However, the overall level of college life, academic, and emotional adaptation were higher among science and technology students than among those majoring in humanities and social studies. In addition, a cluster analysis based on the level of 4 factors of college life adaptation of lower-standard groups showed that cluster 1 of the lower-standard groups had a high level of emotional adaptation, while cluster 2 of the lower-standard groups had a low level of emotional and career adaptation and a high level of academic and social adaptation. Also, cluster 3 of the lower-standard groups was found to have low levels of all factors, and cluster 4 of the lower-standard groups showed low levels of social and emotional adaptation and high levels of academic and career adaptation. This indicates that, in terms of social and emotional adaptation among freshmen, participation in psychological counseling programs tailored according to gender is necessary and, regarding the major, relative emotional intervention is required due to the high and low expectations of job prospects. In conclusion, the emotional and social adaptation to college life among the lower-level groups is likely to serve as a protective factor for college life adaptation, suggesting that specific intervention should be contemplated.

Design of a new adaptive neuro-fuzzy inference system based on a solution for clustering in a data potential field

Nguyen, S.D.,Choi, S.B. North-Holland 2015 Fuzzy Sets and Systems Vol. No.

<P>In this study, we propose a new method for building adaptive neuro-fuzzy inference systems (ANFIS) via datasets. In order to improve the performance of conventional ANFIS to handle noisy data, we focus on ameliorating the cluster-data space established from a given dataset. To achieve this, we propose a weighted clustering process in the joint input-output data space. Thus, during the clustering process, the cluster with the smallest potential distance, which is a combination of the Euclidean distance and the size of the clusters, has priority when obtaining the surveyed sample. Based on this principle, we formulate a new algorithm for synthesizing an ANFIS via the proposed data potential field, called ANFIS-PF, which has the following features: it establishes a data potential field that covers the entire initial data space, a cluster-data space is built based on the generated data potential field, and the ANFIS is synthesized using this cluster-data space. Finally, we performed experiments using datasets with and without noise to demonstrate the effectiveness of the proposed method in several applications, including dynamic-response noisy datasets obtained from a magnetorheological damper. (C) 2015 Elsevier B.V. All rights reserved.</P>

Uncertain fuzzy self-organization based clustering: interval type-2 fuzzy approach to adaptive resonance theory

Majeed, Shakaiba,Gupta, Aditya,Raj, Desh,Rhee, Frank Chung-Hoon Elsevier science 2018 Information sciences Vol.424 No.-

<P><B>Abstract</B></P> <P>Conventional unsupervised learning algorithms require knowledge of the desired number of clusters beforehand. Even if such knowledge is not required in advance, empirical selection of the parameter values may limit the adaptive capability of the algorithm, thereby restricting the clustering performance. An inherent uncertainty in the number and size of clusters requires integration of fuzzy sets into a clustering algorithm. In this paper, we propose a type-1 (T1) fuzzy ART method that adaptively selects the vigilance parameter value, which is critical in determining the network dynamics. This results in improved clustering performance due to the added flexibility in dynamic selection of the number of clusters with the use of fuzzy sets. To further manage the uncertainty associated with memberships, we extend the proposed T1 fuzzy ART with adaptive vigilance to an interval type-2 (IT2) fuzzy ART method. Type reduction and defuzzification are then performed using the KM algorithm to obtain a defuzzified vigilance parameter value. We evaluate our proposed methods on several data sets to validate their effectiveness.</P>

Variation in the Korean Integration of English Word-Final /s/

Iverson, Gregory K.,Lee, Ahrong 서울대학교 언어교육원 (구 서울대학교 어학연구소) 2006 語學硏究 Vol.42 No.2

The adaptation of English word-final /s/ in words that are borrowed into Korean presents an interesting perceptual match-up inasmuch as Korean contrasts two types of voiceless fricatives, “lax” /s/ and “tense” /s’/, either of which, depending on context, may serve as the rendition of English /s/. Which of these is selected appears to correlate with the familiar durational difference between lax and tense consonants in Korean. In particular, S Kim (1999) showed that English words are borrowed consistently with tense /s’/ when the fricative is not in a cluster in the source language (where it is phonetically longer), whereas the result is lax /s/ when the fricative (phonetically shorter) does form part of a consonant cluster in the source language. Hypothesizing that Koreans are sensitive to this durational property, S Kim concluded that positional sub-phonemic length differences in the English fricative are apprehended directly by Korean listeners (and so presumably are contrastive in the language; cf. Iverson & A Lee 2006). Davis & M-H Cho (2006), on the other hand, maintain that the phonetic length correlation is not robust, noting specifically that S Kim was troubled by the adaptation of final clusters consisting of sonorant consonant plus /s/ (as in dance, false), which are rendered with the tense fricative despite being part of a source language cluster. Yet in a comprehensive acoustic study of English /s/ over a full range of environments, Klatt (1974) showed that while /s/ is shorter by 40% in clusters with stops (an [s] that Koreans adapt as lax), it is shorter by only 15% in clusters with sonorant consonants (an [s] which Koreans adapt as tense). Thus, it would appear that Koreans adapt instances of the English fricative following a sonorant as tense /s’/ because these are above the threshold of brevity that marks the non-tense obstruents in Korean, and in any case are appreciably longer than English [s] in an obstruent cluster (as first reported by A Lee (2006a); cf. A Lee (2006b) on adaptation of the stops in English s-clusters). The paper reports on a perception experiment further testing and supporting the hypothesis that phonetic length of the source fricative correlates directly with its adaptation as either lax or tense in Korean.

Speaker Adaptation Using i-Vector Based Clustering

( Minsoo Kim ),( Gil-jin Jang ),( Ji-hwan Kim ),( Minho Lee ) 한국인터넷정보학회 2020 KSII Transactions on Internet and Information Syst Vol.14 No.7

We propose a novel speaker adaptation method using acoustic model clustering. The similarity of different speakers is defined by the cosine distance between their i-vectors (intermediate vectors), and various efficient clustering algorithms are applied to obtain a number of speaker subsets with different characteristics. The speaker-independent model is then retrained with the training data of the individual speaker subsets grouped by the clustering results, and an unknown speech is recognized by the retrained model of the closest cluster. The proposed method is applied to a large-scale speech recognition system implemented by a hybrid hidden Markov model and deep neural network framework. An experiment was conducted to evaluate the word error rates using Resource Management database. When the proposed speaker adaptation method using i-vector based clustering was applied, the performance, as compared to that of the conventional speaker-independent speech recognition model, was improved relatively by as much as 12.2% for the conventional fully neural network, and by as much as 10.5% for the bidirectional long short-term memory.

HIERARCHICAL CLUSTER ANALYSIS by arboART NEURAL NETWORKS and its APPLICATION to KANSEI EVALUATION DATA ANALYSIS

( Shigekazu Ishihara ),( Keiko Ishihara ),( Mitsuo Nagamachi ) 한국감성과학회 2002 춘계학술대회 Vol.2002 No.-

ART (Adaptive Resonance Theory [1]) neural network and its variations perform non-hierarchical clustering by unsupervised learning. We propose a scheme arboART for hierarchical clustering by using several ART1.5-SSS networks. It classifies multidimensional vectors as a cluster tree, and finds features of clusters. The Basic idea of arboART is to use the prototype formed in an ART network as an input to other ART network that has looser distance criteria (Ishihara, et al., [2,3]). By sending prototype vectors made by ART to one after another, many small categories are combined into larger and more generalized categories. We can draw a dendrogram using classification records of sample and categories. We have confirmed its ability using standard test data commonly used in pattern recognition community. The clustering result is better than traditional computing methods, on separation of outliers, smaller error (diameter) of clusters and causes no chaining. This methodology is applied to Kansei evaluation experiment data analysis.

지자체 기후변화 적응 대책 특성 및 개선 방향

안윤정,강영은,박창석,김호걸 한국환경영향평가학회 2016 환경영향평가 Vol.25 No.4

There is a growing interest in establishing a regional climate change adaptation policy as the climate change impact in the region and local scale increases. This study focused on the analysis of 32 regions on its characteristics of local climate change adaptation plans. First, statistic program R was used for conducting cluster analysis based on the frequency and budgets of adaptation plan. Further, we analyzed damage frequency from newspapers regarding climate change impacts in eight categories which were caused by extreme weather events on 2,565 cases for 24 years. Lastly, the characteristics of climate change adaptation plan was compared with damage frequency patterns for evaluating the adequacy of climate change adaptation plan on each cluster. Four different clusters were created by cluster analysis. Most clusters clearly have their own characteristics on certain sectors. There was a high frequency of damage in ‘disaster’ and ‘health’ sectors. Climate change adaptation plan and budget also invested a lot on those sectors. However, when comparing the relative rate among regional governments, there was a difference between types of damage and climate change adaptation plan. We assumed that the difference could come from that each region established their adaptation plans based on not only the frequency of damage, but vulnerability assessment, and expert opinions as well. The result of study could contribute to policy making of climate change adaptation plan. 기후변화에 대한 영향 및 위험은 지역적, 국지적 차원에서 더욱 확장되므로 기후변화에 따른 지역적 영향 및 특성을 반영한 기후변화 적응대책 마련의 필요성이 커지고 있다. 이에 본 연구에서는 기초지자체 적응대책의 분야별 사업 수 및 예산의 특성을 분석하고 피해사례와 비교·검토하여 기후변화 적응대책의 개선방향을 제시하고자 했다. 기초지자체 적응대책 특성 분석을 위해 군집분석을 통하여 지자체 적응대책의 유형별 특성을 파악 했다. 적응대책의 계획이 실제 기후변화로 인한 피해 내용을 잘 반영하고 있는지 검토하기 위하여 과거 24년 동안 지자체별 관측 영향 결과(신문기사 2,565건)와 비교·분석을 수행했다. 군집분석 수행 결과 군집은 4가지 유형으로 구분되었다. 전국 피해 유형으로 재난재해, 건강 부분에서 공통적으로 피해 빈도가 높은 것으로 분석되었으며, 적응계획 또한 재난재해, 건강, 농업, 물 관리순서로 높은 비율을 보였다. 하지만 피해 사례 반영의 비중과 단기 및 장기 미래에 대한 고려 수준에 따라서 각 군집별로 피해빈도와 적응계획의 특성에서 차이를 보였다. 본 연구의 결과는 향후 기초지자체 특성및 지역별 실질적 피해에 기반 한 적응대책 마련의 기초자료로 활용될 수 있을 것이라고 판단된다.

Novel approaches to pin cluster synchronization on complex dynamical networks in Lur’e forms

Tang, Ze,Park, Ju H.,Feng, Jianwen Elsevier 2018 Communications in nonlinear science & numerical si Vol.57 No.-

<P><B>Abstract</B></P> <P>This paper investigates the cluster synchronization of complex dynamical networks consisted of identical or nonidentical Lur’e systems. Due to the special topology structure of the complex networks and the existence of stochastic perturbations, a kind of randomly occurring pinning controller is designed which not only synchronizes all Lur’e systems in the same cluster but also decreases the negative influence among different clusters. Firstly, based on an extended integral inequality, the convex combination theorem and S-procedure, the conditions for cluster synchronization of identical Lur’e networks are derived in a convex domain. Secondly, randomly occurring adaptive pinning controllers with two independent Bernoulli stochastic variables are designed and then sufficient conditions are obtained for the cluster synchronization on complex networks consisted of nonidentical Lur’e systems. In addition, suitable control gains for successful cluster synchronization of nonidentical Lur’e networks are acquired by designing some adaptive updating laws. Finally, we present two numerical examples to demonstrate the validity of the control scheme and the theoretical analysis.</P> <P><B>Highlights</B></P> <P> <UL> <LI> Cluster synchronization of complex dynamical networks consisted of identical or nonidentical Lur’e systems is considered. </LI> <LI> Novel design approaches for randomly occurring pinning controller to cluster synchronization are proposed. </LI> <LI> Verification of presented results via two real example is given. </LI> </UL> </P>