Multiway generalized canonical correlation analysis

• DOI: 10.1093/biostatistics/kxaa010
• Publication Date: 2020-02-17T12:12:16Z
• Authors: Arnaud Gloaguen; Cathy Philippe; Vincent Frouin; Giulia Gennari; Ghislaine Dehaene...; Laurent Le Brusquet; Arthur Tenenhaus
• Keywords: Canonical correlation; Canonical analysis; Component analysis; Canonical correspondence analysis; Correspondence analysis

Regularized generalized canonical correlation analysis (RGCCA) is a general multiblock data framework that encompasses several important multivariate methods such as principal component analysis, partial least squares regression, and versions of analysis. In this article, we extend RGCCA to the case where at one block has tensor structure. This method called multiway (MGCCA). Convergence properties MGCCA algorithm are studied, computation higher-level components discussed. The usefulness shown on simulation cognitive study in human infants using electroencephalography (EEG).