Classifier Based on Non-negative Matrix Factorization for Tumor Data Classification

With the development of DNA microarrys technology, it is very important to classify the different tumor types correctly in cancer diagnosis and drug discovery. In this paper, we discuss how to use the nonnegative matrix factorization (NMF) to extract features and illustrate how to adopt classification model to improve the classification accuracy. For the DNA microarrys, the gene expression data is firstly preprocessed for normalization. NMF is then applied to extract features. Finally, we use the Back Propagation Neural Network (BPNN) as the classifier to classify the different samples. In our experiments, we adopt the leukemia and colon datasets to test the validity. The experimental results show that the proposed method yields a good recognition rate.