Neural network-based analysis of DNA microarray data

The analysis of DNA microarray expression data has become an important subject in bioinformatics. Scientists have adopted different approaches to select the informative genes those can distinguish different types of cancers. In this paper, we show the use of a dimension reduction technique such as singular value decomposition (SVD) to capture the genes with similar patterns. We propose a novel method of selection of feature genes based on information loss using SVD. To assign the samples to known classes, we design a multi-layer perceptron-based classifier with reduced dimensional input vectors. We provide performance comparison between different selection methods in terms of classification rate