A self-organizing maps algorithm for gene expression data clustering based on feature's distribution

In order to solve the problem that traditional SOM algorithm simply regards all the features as equal importance, a novel similarity computation method is proposed in this paper. This method uses feature's intra-cluster distribution and inter-cluster distribution to evaluate different features with different weights, and integrate features' weights in similarity computation. Experiment results demonstrate that this novel similarity computation method can effectively improve precision on gene expression data clustering.