Abstract:

To project high-dimensional data into low-dimensional space can be effectively utilized to visualize and explore properties of data. An approach of NLPCA (NonLinear Principal Component Analysis)and SOM(Self-Organizing Map) neural network is presented for clustering and visualization of gene expression data. The experiment results show that the performance of clustering gene expression data based on the SOM network is efficient.

Key words: self-organizing map neural network;nonlinear principal components analysis, cluster analysis, visualization