Resampling Method for Unsupervised Estimation of Cluster Validity

By Erel Levine, Eytan Domany. Neural Computation, 2000

294 citations recorded by the source corpus.

We introduce a method for validation of results obtained by clustering analysis of data. The method is based on resampling the available data. A figure of merit that measures the stability of clustering solutions against resampling is introduced. Clusters that are stable against resampling give rise to local maxima of this figure of merit. This is presented first for a one-dimensional data set, for which an analytic approximation for the figure of merit is derived and compared with numerical measurements. Next, the applicability of the method is demonstrated for higher-dimensional data, including gene microarray expression data.

Subfields: Artificial Intelligence, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/1519597a92d8591190613703a36f360daa70e96b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

Citations in the corpus

An Unsupervised Aspect-Sentiment Model for Online Reviews (2010, methodology): “Available at http:// gibbslda.sourceforge.net/. approach (Levine and Domany, 2001; Lange et al., 2004; Niu et al., 2007) is to use a cluster validation procedure.”
Survey of clustering algorithms (2005, background): “We offer more detailed discussion on how to identify appropriate number of clusters, which is particularly important in cluster validity, in the last part of the section.”
Clustering Stability: An Overview (2010, methodology): “• Compare the clustering of the original data set with the clusterings obtained on subsamples (Levine and Domany, 2001).”
Computational cluster validation in post-genomic data analysis (2005, background): “In order to assess the degree of stability of a partitioning, several papers (Ben-Hur et al., 2002; Levine and Domany, 2001) repeatedly draw overlapping subsamples of the same dataset (the individual subsamples are drawn without replacement).”
Icasso: software for investigating the reliability of ICA estimates by clustering and visualization (2003, background): “4 can be additionally motivated by the fact that for two normalized zero mean vectors ui and uj their correlation coefficient rij = u T i uj and their Euclidean distance ‖ ui − uj ‖∝ √ 1 − rij [14].”
Consensus Clustering: A Resampling-Based Method for Class Discovery and Visualization of Gene Expression Microarray Data (2003, methodology): “…perturbations of the original data set, so as to assess the stability of the clustering results with respect to sampling variability (Ben-Hur, Elisseeff, & Guyon, 2002; Bhattacharjee et al., 2001; Dudoit & Fridlyand, 2002; Jain & Moreau, 1988; Levine & Domany, 2001; Tibshirani et al., 2001a).”
Evaluation of Stability of k-Means Cluster Ensembles with Respect to Random Initialization (2006, methodology): “Different cross-validation protocols can be used, the two most widely explored being 2-fold cross-validation [15], [31] and bootstrap resampling or subsampling [7], [9], [22], [23], [24], [27], [28].”
Validating the independent components of neuroimaging time series via clustering and visualization (2004, methodology): “These included four of the Dunn-like indices in (Bezdek and Pal, 1998) and the R-index defined in (Levine and Domany, 2001).”
Stability-Based Validation of Clustering Solutions (2004, methodology): “In the three ring data set (depicted in Figures 3A and 3C), which was also used by Levine and Domany (2001), three ring-shaped clusters can be naturally distinguished.”