Prof. Yoshihiro Taguchi
Chuo University, japan
Title of Plenary Speech
Kernel tensor decomposition based unsupervised feature extraction applied to bioinformatics
Abstract of Plenary Speech
In genomic science, so called “large p small n” problem is very usual, since the number of genes (i.e., features, =p) is as many as a few tens thousands whereas the number of samples (=n) is at most a few hundreds. Feature selection in “large p small n” is always difficult. In order to address this problem, I proposed tensor decomposition (TD) based unsupervised feature extraction (FE) and applied mainly to bioinformatics. TD based unsupervised FE was further extended to employ kernel trick to take the non-linearity into consideration. The target of the method is wide-ranged from biomarker identification, identification of disease causing genes and drug repositioning. I introduce this in may talk and discuss how to apply this method to various applications.
