GAPS: A clustering method using a new point symmetry-based distance measure

Modeling Tumor Growth for Kidney Cancer Based on Nuclei Clusters of Pathology Slides

“…geometric mean of clusters) inside a cluster and by maximizing the separation between different clusters [8], [17], [16]. Several algorithms are reported in the literature for computation of the best number of clusters [6], [9], [18], [19]. Jain et al [9] presents various approaches for clustering including evolutionary approaches, artificial neural networks, and search based approaches to determine the number of clusters and corresponding groups.…”

“…Jain et al [9] presents various approaches for clustering including evolutionary approaches, artificial neural networks, and search based approaches to determine the number of clusters and corresponding groups. Liu et al [6], Murthy et al [18], and Bandyopadhyay et al [19] implement genetic algorithm to identify the best structure for clustering. Ray et al [17] propose a validity measure to draw best structure of clusters based on sum of the distances between cluster centers and data points inside all clusters (i.e.…”

Performance Assessment of Some Clustering Algorithms Based on a Fuzzy Granulation-Degranulation Criterion

“…The algorithm which provides the smallest quan error is regarded as the best suitable for that particular data set. Five clustering algorithms, viz., a newly developed point symmetry based genetic clustering technique (GAPS) [3], GAK-means algorithm [9], Average-linkage clustering algorithm [6] (source code was obtained from http://bioinformatics.oxfordjournals.org/cgi/content/abstract), Self Organizing Map (SOM) [8] (source obtained from http://www.cs.tau.ac.il/∼rshamir/expander), Expectation Maximization (EM) algorithm [5] (matlab source code was obtained from http://www.mathworks.com/matlabcentral/fileexchange/) are used as the underlying partitioning techniques. The parameters of the genetic clustering algorithms (GAPS and GAK-means) are as follows: population size is equal to 100 and maximum number of generations is equal to 30.…”

“…Unlike the K-means it does not depend on any distance measure, and accommodates categorical and continuous data in a superior manner. Recently, some clustering methods have been proposed that exploit the symmetry property within the clusters [3]. These methods are found to be superior to several other techniques when the clusters do offer a symmetric structure.…”

“…Based on the proposed performance index the appropriate clustering algorithm for a particular data set is determined. The five clustering algorithms are the well-known genetic algorithm based K-means clustering technique (GAK-means) which is developed in [9], in order to overcome the limitation of K-means algorithm to get stuck at sub optimal solution, a newly developed genetic point symmetry based clustering technique (GAPS) [3], the well-known Expectation Maximization clustering algorithm [5], the Average Linkage clustering algorithm [7] and Self Organizing Map (SOM) [8]. Let the clustering algorithms be denoted by A i where i = 1, .…”