On combining multiple clusterings: an overview and a new perspective

Li, Tao; Ogihara, Mitsunori; Ma, Sheng

doi:10.1007/s10489-009-0160-4

Cited by 29 publications

(20 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The combination of base clusterings can be constructed by three kinds of method: the consensus functions [Strehl and Ghosh 2002], the categorical clusterings [Gionis et al 2007], and the direct optimizations [Christou 2011]. The consensus functions focus on the total agreement of all the base clusterings from different perspectives [Li et al 2010]. The clustering ensemble can also be converted to the problem of clustering categorical data (categorical clustering [Guha et al 2000;Andritsos et al 2004], for short) by viewing each attribute as a way of producing a base clustering of the data.…”

Section: Process Of Clustering Ensemblementioning

confidence: 99%

“…We can construct consensus functions by the following approaches: direct best matching [Li et al 2010], graph-based mappings [Strehl and Ghosh 2002;Fern and Brodley 2004], statistical mixture models [Topchy et al 2005], pairwise comparisons [Gionis et al 2007;Li et al 2010] and a number of other models. They are all built on the co-associations or pairwise agreements between clusterings (e.g., partition difference PD [Li et al 2010] that focuses on the similarity between partitions and QMI [Topchy et al 2005] that works on the consensus function based on quadratic mutual information), between data objects (e.g., CSPA 1 [Strehl and Ghosh 2002]) that induces a similarity measure from partitions and re-clusters objects, or between clusters (e.g., MCLA [Strehl and Ghosh 2002] that collapses groups of clusters into meta-clusters and competes for each object to determine the combined clustering). While the clustering ensemble based on consensus functions largely captures the common structure of the base clusterings, and achieves a combined clustering with better quality than individual clusterings, it also faces several issues that have not been explored well in the consensus design.…”

Section: Process Of Clustering Ensemblementioning

confidence: 99%

See 1 more Smart Citation

Coupled Clustering Ensemble by Exploring Data Interdependence

Wang

Chi

She

et al. 2018

ACM Trans. Knowl. Discov. Data

View full text Add to dashboard Cite

Clustering ensembles combine multiple partitions of data into a single clustering solution. It is an effective technique for improving the quality of clustering results. Current clustering ensemble algorithms are usually built on the pairwise agreements between clusterings that focus on the similarity via consensus functions, between data objects that induce similarity measures from partitions and re-cluster objects, and between clusters that collapse groups of clusters into meta-clusters. In most of those models, there is a strong assumption on IIDness (i.e. independent and identical distribution), which states that base clusterings perform independently of one another and all objects are also independent. In the real-world, however, objects are generally likely related to each other through features that are either explicit or even implicit. There is also latent but definite relationship among intermediate base clusterings because they are derived from the same set of data. All these demand a further investigation of clustering ensembles that explores the interdependence characteristics of data. To solve this problem, a new coupled clustering ensemble (i.e. CCE) framework that works on the interdependence nature of objects and intermediate base clusterings is proposed in this paper. The main idea is to model the coupling relationship between objects by aggregating the similarity of base clusterings, and the interactive relationship among objects by addressing their neighborhood domains. Once these interdependence relationships are discovered, they will act as critical supplements to clustering ensembles. We verified our proposed framework by using three types of consensus function: clustering-based, object-based, and cluster-based. Substantial experiments on multiple synthetic and real-life benchmark data sets indicate that CCE can effectively capture the implicit interdependence relationships among base clusterings and among objects with higher clustering accuracy, stability, and robustness compared to 14 state-of-the-art techniques, supported by statistical analysis. In addition, we show that the final clustering quality is dependent on the data characteristics (e.g. quality and consistency) of base clusterings in terms of sensitivity analysis. Finally, the applications in document clustering, as well as on the data sets with much larger size and dimensionality, further demonstrate the effectiveness, efficiency, and scalability of our proposed models.

show abstract

Section: Process Of Clustering Ensemblementioning

confidence: 99%

Section: Process Of Clustering Ensemblementioning

confidence: 99%

Coupled Clustering Ensemble by Exploring Data Interdependence

Wang

Chi

She

et al. 2018

ACM Trans. Knowl. Discov. Data

View full text Add to dashboard Cite

show abstract

“…We thus propose a disparate clustering algorithm to find the design alternatives (as shown in Algorithm I). Multiple clustering can be implemented using different attributes in K-means or by changing the cluster merging rules in hierarchical clustering [20]. We can use a contingency table to capture the relationships between members in BOM clusters and EI clusters.…”

Section: Finding Sustainable Design Alternatives Based On Multiplementioning

confidence: 99%

Data-driven solutions for building environmental impact assessment

Zhou

Zhu

et al. 2015

Proceedings of the 2015 IEEE 9th International Conference on Semantic Computing (IEEE ICSC 2015)

Self Cite

View full text Add to dashboard Cite

Life cycle assessment (LCA) as a decision support tool for evaluating the environmental load of products has been widely used in many fields. However, applying LCA in the building industry is expensive and time consuming. This is due to the complexity of building structure along with a large amount of high-dimensional heterogeneous building data.So far building environmental impact assessment (BElA) is an important yet under-addressed issue. This paper gives a brief survey of BElA and investigates potential advantages of using data mining techniques to discover the relationships between building materials and environment impacts. We formulate three important BElA issues as a series of data mining problems, and propose corresponding solution schemes. Specifically, first, a feature selection approach is proposed based on the practical demand and construction characteristics to perform assessment analysis. Second, a unified framework for solving constraint based clustering ensemble selection is proposed to extend the environmental impact assessment range from the building level to the regional level. Finally, a multiple disparate clustering method is presented to help sustainable new buildings design. We expect our proposal would shed light on data-driven approaches for environment impact assessment.

show abstract

“…The more two objects or clusters resemble each other, the larger is the similarity [5]. The other similarity between attributes [6] can also be converted into the difference of similarities between pairwise attribute values [7]. Therefore, the similarity between attribute values plays a fundamental role in similarity analysis.…”

Section: Introductionmentioning

confidence: 99%

Coupled Attribute Similarity Learning on Categorical Data

Wang¹,

Xiang²,

Zhou³

et al. 2015

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

Attribute independence has been taken as a major assumption in the limited research that has been conducted on similarity analysis for categorical data, especially unsupervised learning. However, in real-world data sources, attributes are more or less associated with each other in terms of certain coupling relationships. Accordingly, recent works on attribute dependency aggregation have introduced the co-occurrence of attribute values to explore attribute coupling, but they only present a local picture in analyzing categorical data similarity. This is inadequate for deep analysis, and the computational complexity grows exponentially when the data scale increases. This paper proposes an efficient data-driven similarity learning approach that generates a coupled attribute similarity measure for nominal objects with attribute couplings to capture a global picture of attribute similarity. It involves the frequency-based intra-coupled similarity within an attribute and the inter-coupled similarity upon value co-occurrences between attributes, as well as their integration on the object level. In particular, four measures are designed for the inter-coupled similarity to calculate the similarity between two categorical values by considering their relationships with other attributes in terms of power set, universal set, joint set, and intersection set. The theoretical analysis reveals the equivalent accuracy and superior efficiency of the measure based on the intersection set, particularly for large-scale data sets. Intensive experiments of data structure and clustering algorithms incorporating the coupled dissimilarity metric achieve a significant performance improvement on state-of-the-art measures and algorithms on 13 UCI data sets, which is confirmed by the statistical analysis. The experiment results show that the proposed coupled attribute similarity is generic, and can effectively and efficiently capture the intrinsic and global interactions within and between attributes for especially large-scale categorical data sets. In addition, two new coupled categorical clustering algorithms, i.e., CROCK and CLIMBO are proposed, and they both outperform the original ones in terms of clustering quality on UCI data sets and bibliographic data.

show abstract

On combining multiple clusterings: an overview and a new perspective

Cited by 29 publications

References 54 publications

Coupled Clustering Ensemble by Exploring Data Interdependence

Coupled Clustering Ensemble by Exploring Data Interdependence

Data-driven solutions for building environmental impact assessment

Coupled Attribute Similarity Learning on Categorical Data

Contact Info

Product

Resources

About