Chapter 10: Tensor Decomposition Based Unsupervised Feature Extraction Applied to Bioinformatics
Although so called “large p small n” problem is typical in bioinformatics, there are no effective feature selection methods applicable to them. In this chapter, we propose tensor decomposition based unsupervised feature extraction. The proposed method is applied to post-traumatic stress disorder medicated heart diseases and 26 non-small cell lung cancer cell lines, off-target effect of miRNA transfection, in silico drug discovery from gene expression, and social insects with multiple castes. In spite of the variety of targeted problems, the proposed method turn out to work pretty well.