Robust data clustering

By Ana L. N. Fred, Anil K. Jain. 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings., 2003

401 citations recorded by the source corpus.

We address the problem of robust clustering by combining data partitions (forming a clustering ensemble) produced by multiple clusterings. We formulate robust clustering under an information-theoretical framework; mutual information is the underlying concept used in the definition of quantitative measures of agreement or consistency between data partitions. Robustness is assessed by variance of the cluster membership, based on bootstrapping. We propose and analyze a voting mechanism on pairwise associations of patterns for combining data partitions. We show that the proposed technique attempts to optimize the mutual information based criteria, although the optimality is not ensured in all situations. This evidence accumulation method is demonstrated by combining the well-known K-means algorithm to produce clustering ensembles. Experimental results show the ability of the technique to identify clusters with arbitrary shapes and sizes.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9e7a86fd9e15bf37d937a79ccb7efb78bb070f74

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 Computer Vision and Pattern Recognition

Citations in the corpus

Evaluation of Stability of k-Means Cluster Ensembles with Respect to Random Initialization (2006, methodology): “Among these are random initializations of the clustering algorithm, subsampling, or resampling the data [5], [8], [9], [13], [16], [26], [27], applying different types of clustering algorithms [1], [16], [18], [37], using subsets of features [16], [32], “weakening” the clustering algorithm [16], [34”
A Comprehensive Survey on Community Detection With Deep Learning (2021, background): “Combo [61] optimizes normalized mutual information (NMI) [62], [63] and conductance (CON) [64].”
Clustering ensembles: models of consensus and weak partitions (2005, methodology): “including the co-association based hierarchical methods [15, 16, 17 ], hypergraph algorit hms [47, 29,”
Using diversity in cluster ensembles (2004, background): “A way to avoid relabeling is to treat the two partitions as (nominal) random variables, say X and Y , and calculate the mutual information between them [7] [15] [16].”
Community detection in networks: A user guide (2016, methodology): “This has been achieved by introducing the normalised mutual information (NMI), obtained by dividing the mutual information by the arithmetic average19 of the entropies of X and Y [84]”
You are who you know: inferring user profiles in online social networks (2010, methodology): “To measure how similar these two community structures are, we use the normalized mutual information metric [9].”