On strategies for building effective ensembles of relative clustering validity criteria

Jaskowiak, Pablo Andretta; Moulavi, Davoud; Furtado, Antonio Carlos; Campello, Ricardo J. G. B.; Zimek, Arthur; Sander, Jöerg

doi:10.1007/s10115-015-0851-6

Cited by 27 publications

(17 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this work we have examined the usage of the method with scientific documents from a particular domain only, whereas the overall applicability to clustering other kinds of documents should be researched. It is also worth examining the applications of the ensembles [9] of multiple CVIs pointed out as suitable for assessing the document clusters by our method.…”

Section: Discussionmentioning

confidence: 99%

“…In [9] multiple strategies of building ensembles of clustering validity indices are evaluated. Authors have conducted an experiment on various synthetic and real-life datasets and examined the correlation of the ensembles of relative indices with regards to external validity index.…”

Section: A Ensembles Of Clustering Validity Indicesmentioning

confidence: 99%

“…Another problem regarding ensembles of CVI is the choice of aggregation strategy. In [9] authors have compared various strategies of combining the results of the indices. The authors found, that score-based strategies, like mean or median of the normalized values of indices in the ensemble, appeared to give inferior results to methods based on rank aggregation.…”

Section: A Ensembles Of Clustering Validity Indicesmentioning

confidence: 99%

See 2 more Smart Citations

Clustering Validity Indices Evaluation with Regard to Semantic Homogeneity

Dziopa¹

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Clustering validity indices are methods for examining and assessing the quality of data clustering results. Various studies provide a thorough evaluation of their performance using both synthetic and real-world datasets. In this work, we describe various approaches to the topic of evaluation of a clustering scheme. Moreover, a new solution to a problem of selecting an appropriate clustering validity index is presented. The approach is applied to a problem of selecting a suitable clustering validity index for a real-world task of clustering biomedical articles using the MeSH ontology.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: A Ensembles Of Clustering Validity Indicesmentioning

confidence: 99%

Section: A Ensembles Of Clustering Validity Indicesmentioning

confidence: 99%

See 1 more Smart Citation

Clustering Validity Indices Evaluation with Regard to Semantic Homogeneity

Dziopa¹

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

show abstract

“…The dynamic compensation is based on monitoring of the slope angle of the signal curve and subtracting a straight line with a slope coefficient tgα. Removal of artifacts and the further search for the maximum value is based on the method of cluster analysis for pairwise weighted average [2]. As the criterion for connecting into clusters Euclidean distance is applied.…”

Section: Data Analysis Unitmentioning

confidence: 99%

Development of a Computer Based Problem-Oriented Automated Remote Leakage Detection Control System

Kosach

Kovshov

2017

ITM Web Conf.

View full text Add to dashboard Cite

Abstract. We explore the architecture and the concept of building an automation remote leakage detection system, in particular -a system for collection and processing of information. As practice shows, the development of those systems is a lengthy and time-consuming process involving management of heterogeneous hardware and analysis of large amounts of data. To implement the task, it is proposed to use the SCADA industry standard, universal communication protocols, and for data processing -open source database management systems and cluster analysis algorithms. A universal approach makes it possible to develop a scalable problem-oriented system. This approach takes into account the information received and processed during the leakage detection process.

show abstract

“…Como a própria definição de cluster admite diferentes interpretações e portanto certa subjetividade, diferentes critérios de validação existem que refletem diferentes aspectos de uma solução ou grupo candidato (VENDRAMIN; CAMPELLO; HRUSCHKA, 2010). Mais ainda do que no contexto supervisionado (classificação) mencionado anteriormente, é bem conhecido na literatura de agrupamento de dados que não se deve confiar exclusivamente em um critério de validação único, mas ao invés utilizar mais de um critério num contexto de análise exploratória (BEZDEK;PAL, 1998;JASKOWIAK et al, 2016). Neste contexto, certos estudos da literatura vão além de simplesmente propor um novo critério a ser considerado junto a outros existentes, mas têm sido capazes de identificar cenários nos quais determinados critérios são mais ou menos apropriados (VENDRAMIN; CAMPELLO; HRUSCHKA, 2010), o que pode guiar o usuário ou analista na escolha de critérios a serem utilizados em uma aplicação em particular.…”

Section: Motivaçãounclassified

Seleção de grupos a partir de hierarquias: uma modelagem baseada em grafos

Anjos¹

View full text Add to dashboard Cite

Sozinho, nenhum objetivo pode ser alcançado. Por trás de cada realização sempre existe uma infinidade de pessoas envolvidas sem as quais tal feito jamais seria possível. Dentre tantas pessoas que, de uma maneira ou de outra, me ajudaram ao longo de todos esses anos, algumas delas não poderiam deixar de ser mencionadas. Em primeiro lugar, agradeço ao Prof. Ricardo por: seu exemplo de caráter, comprometimento e profissionalismo; seus valiosos conselhos profissionais e pessoais; e, acima de tudo, sua compreensão nos momentos mais difíceis ao longo dessa caminhada.

show abstract

On strategies for building effective ensembles of relative clustering validity criteria

Cited by 27 publications

References 55 publications

Clustering Validity Indices Evaluation with Regard to Semantic Homogeneity

Clustering Validity Indices Evaluation with Regard to Semantic Homogeneity

Development of a Computer Based Problem-Oriented Automated Remote Leakage Detection Control System

Seleção de grupos a partir de hierarquias: uma modelagem baseada em grafos

Contact Info

Product

Resources

About