Gene expression data classification using automatic differential evolution based algorithm

Microarray technology facilitates to monitor the expression levels of multiple genes simultaneously over a number of time points. For analyzing such microarray gene expression data, fuzzy clustering plays an important role. However, this fact has inspired us to propose a new automatic cluster detection method using modified differential evolution based fuzzy symmetric clustering technique. Here, allocation of gene data points to different clusters has been made based on a newly developed symmetry based cluster validity index. In order to get a best fuzzy partition of a gene dataset, FSym-index needs to be maximized. The search capability of the modified differential evolution based optimization technique in conjunction with the fuzzy symmetry based cluster validity index is applied to ascertain the good fuzzy partition as well as to obtain actual number of cluster present in a data set. The superiority of the proposed fuzzy-VMODEPS is compared with those obtained by some recent automatic clustering techniques, VGAPS, GCUK and HNGA over publicly available three real life gene expression data-sets.