Tensor decomposition-based unsupervised feature extraction applied to matrix products for multi-view data processing.

摘要: In the current era of big data, amount data available is continuously increasing. Both number and types samples, or features, are on rise. The mixing distinct features often makes interpretation more difficult. However, separate analysis individual requires subsequent integration. A tensor a useful framework to deal with in an integrated manner without them. On other hand, not easy obtain since it measurements huge numbers combinations features; if there m kinds each which has N dimensions, needed as many Nm, too large measure. this paper, I propose new method where generated from combinatorial measurements, was decomposed back matrices, by unsupervised feature extraction performed. order demonstrate usefulness proposed strategy, applied synthetic well three omics datasets. It outperformed matrix-based methodologies.