Sparse Non-negative Tensor Factorization Using Columnwise Coordinate Descent
11:30:00 - 12:00:00
308 , Mathematics Research Center Building (ori. New Math. Bldg.)
The SVD and PCA methods are often used in visual data representation, analysis and visualization. In this talk, we will emphasize on a higher-order generalization of the SVD and PCA methods to extract data-dependent non-negative basis functions. They can be used for data compression, visualization, and detection of hidden information (factors). In this talk, I will emphasize on the technical derivation of sparse non-negative tensor factorization using columnwise coordinate descent. We conduct two experiments to illustrate how this procedure achieves the goal to get a sparse non-negative tensor factorization.