ICA based supervised gene classification of Microarray data in yeast functional genome

In order to study the function of unknown genes in functional genome, traditional supervised classification algorithms were applied to gene classification with Microarray expression profiles. But the results show that the classification precision is poor and the accuracies achieved for different classes varies dramatically. Because the gene expression profiles are mixed with different biologically meaningful information and there is much noise in the genomic-scale dataset. Independent component analysis (ICA) is a method for multi-channel signal processing to separate mixed signals. Through linear transformation, ICA minimizes the statistical dependence of the components of the represented variables. So in this paper ICA based supervised gene classification in yeast functional genome is presented, which is a hybrid method of ICA with supervised classification approaches. This method recognizes the hidden patterns under the gene expression profiles and reduces the noise that is abundant in the gene expression profiles efficiently. Experimental results show that this method improves the performance of precision and recall