Removal of Mirror Bicliques from a Symmetric Matrix
Michael Raj Dominic Savio, A. Sankar, R V. Nataraj · 2014
We introduce a novel algorithm to remove all mirror bicliques, i.e. duplicate patterns, while generating all maximal bicliques from a symmetric matrix. Our algorithm MD-Miner+ basically designed by exploiting the symmetric nature of the dataset, that extends DMiner algorithm, where the patterns are generated only once and all the mirror patterns are pruned, whereas DMiner generates all mirror patterns. Our algorithm is compared with DMiner algorithm and observed that ours take only half the running time on sparse datasets and outperform on dense datasets during experiments conducted on benchmark datasets.